cc/td/doc/product/software/ios121/121relnt
hometocprevnextglossaryfeedbacksearchhelp
PDF

Table of Contents

Caveats for Cisco IOS Release 12.1

Caveats for Cisco IOS Release 12.1

September 28, 2000

Text Part Number 78-10725-05 Rev. B3

Note   You can find the most current Cisco IOS documentation on Cisco Connection Online (CCO). These electronic documents may contain updates and modifications made after the hard copy documents were printed.

This document lists severity 1 and 2 caveats, as well as select severity 3 caveats, for Cisco IOS Release 12.1, up to and including Release 12.1(4). Caveats describe unexpected behavior or defects in Cisco IOS software releases. Severity 1 caveats are the most serious caveats; severity 2 caveats are less serious.

To help us improve this document, please send us your comments. If you are reading Cisco product documentation on the World Wide Web, you can submit comments electronically at http://www.cisco.com/feedback/ or contact caveats-doc@cisco.com. For more information, see the "Documentation CD-ROM" section.

How to Use This Document

This document describes open and resolved severity 1 and 2 caveats and select severity 3 caveats:

Within the sections the caveats are sorted by technology in alphabetical order. For example, AppleTalk caveats are listed separately from, and before, IP caveats. The caveats are also sorted alphanumerically by caveat number.

If You Need More Information

The most up-to-date documentation can be found on the Web through CCO and on the latest Documentation CD-ROM. These electronic documents might contain updates and modifications made after the paper documents were printed. For information on CCO, see the "Cisco Connection Online" section. For more information on the CD-ROM, see the "Documentation CD-ROM" section.

For more information on caveats and features in Cisco IOS Release 12.1, see the following sources:

The following table lists the most recent release notes when this caveats document was published: 
Release Notes

Cisco IOS Release

Revision Date

 Release Notes for Cisco IOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco 2600 Series for  CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco 3600 Series for  CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco 4000 Series for  CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco 4500 Series for  CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco AS5200  Universal Access Server for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for the Cisco AS5300  Universal Access Server for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for the Cisco AS5800  Universal Access Server for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for the Cisco MC3810 for  CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco uBR900 Series  for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco uBR7200 Series  for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000


 Release Notes for Cisco Catalyst 5000  Family RSM/VIP2 for CiscoIOS Release12.1


 Release12.1(4)


 September 4, 2000

Open Caveats—12.1(4)

This section describes possibly unexpected behavior by Release 12.1(4). This section describes severity 1 and 2 caveats and select severity 3 caveats.

Basic System Services

  When hardware compression is enabled, packets are normally fast-switched. If you disable fast switching and then enable fast switching again, fast switching remains disabled.
  Workaround: Reconfigure hardware compression by entering the no compression command followed by the compression stac command.
  Applying a named access list (for example, "internet" ) through RADIUS works for PPP but fails for Serial Line IP (SLIP) users with shell service. The following error message occurs:
  AAA/AUTHOR/EXEC: received invalid access-class value 0. (Should be 1 - 199) AAA/AUTHOR/EXEC: acl internet does not exist.
  There is no workaround.
  A Cisco RSP8 Route Switch Processor (RSP) might run out of buffer headers used for communication between the Versatile Interface Processors (VIPs) and the RSP. This situation usually occurs on a Cisco 7500 series router that is configured with several T1 port adapters (PAs). Communication between the VIP and the RSP is adversely affected, and messages such as the VIP notifying the RSP that a line has gone down might not be completed. You can diagnose this condition by using the show controllers cbus privileged EXEC command to determine if the number of buffers in BufhdrQ is near zero. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0 T and that is configured with an HSSI4/0/0 High-Speed Serial Interface might exhibit the following error message:
  %RSP-3-RESTART: interface Hssi4/0/0
  There is no workaround.
  On an RSP2 Route Switch Processor, a Versatile Interface Processor (VIP) might reload after you upgrade the RSP2 to Cisco IOS Release 12.1(1). The RSP2 exhibits the following error messages:
  %VIP2-1-MSG: slot8 CYASIC Error Interrupt register 0x24200000
%VIP2-1-MSG: slot8 DMA Transmit Error
%VIP2-1-MSG: slot8 CYASIC Other Interrupt register 0x200000
%VIP2-1-MSG: slot8 CYBUS Error Cmd/Addr 0xD001B8A
%VIP2-1-MSG: slot8 MPUIntfc/PacketBus E
  There is no workaround.
  Frame Relay (FRF.9) hardware compression does not keep input byte count for compressed data on both interface level and Frame Relay Permanent Virtual Circuit (PVC) level. The display from these commands appears inconsistent with the output byte counts and with the FRF.9 software compression statistics displayed by the show interface, show compress, and show frame-relay pvc EXEC commands. Also, the corresponding MIB representation becomes inconsistent. There is no workaround.
  On asynchronous interfaces that are configured with TCP header compression, NetFlow statistics reports only flow one way even if NetFlow is enabled on every interface. There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1(1) might reload with a bus error. There is no workaround.
  A Frame Relay interface on a Cisco MC3810 multiservice concentrator must be cleared after output queue wedges fail to send Local Management Interface (LMI) packets resulting in a "Line Protocol Down" message. There is no workaround.
  The backup BRI port on an EBS card intermittently fails to initialize. There is no workaround.
  When the compression service adapter (CSA) for a Cisco 7200 series router is configured with PPP encapsulation, it loses compression from the configuration upon router reload.
  Workaround: Reenable the compress stac interface configuration command after reload.
  The write memory command on a system that has High System Availability (HSA)(two Route Switch Processors (RSPs) in a Cisco 7500 platform) and the service sync config or the slave sync config privileged EXEC commands configured on the same system will cause the Master RSP to run at 100 percent CPU for a variable amount of time depending on the size of the configuration. There is no workaround.
  Two Cisco 7513 routers with "Compression PA, 3M SRAM" compression cards and "Mx Serial PA, 4 ports" cards receive a high number of restarts after entering the show compress EXEC command when the compress stac interface configuration command is issued on the interface.
  The performance of the router becomes poor and might reload. No cyclic redundancy check (CRC) errors were detected after entering the show compress EXEC command. There is no workaround.
  A Cisco 7500 series router with a Route Switch Processor 8 (RSP8) will reload with a watchdog timeout if you enter the memory cache-policy io uncached command. There is no workaround.
  A Cisco 7513 router that is running Cisco IOS Release 12.1(2) with Gigabit Ethernet Interface Processor (GEIP) reloads. There is no workaround.
  When a Cisco router is reloaded, ICMP Router Discovery Protocol (IRDP) updates might never be sent on an interface if a high rate of broadcasts are received during the router reload.
  Workaround: Shut down and unshut the interface of the trunk after the router comes back up.
  A Cisco 7206VXR router that is running Cisco IOS Release 12.1.2 reloads approximately once a week at "abort." There is no workaround.
  Authentication, Authorization, and Accounting (AAA) and ISDN session time measurements are not equal. There is no workaround.
  When using Flash load helper to copy an image to Flash memory, boot ROMs before Cisco IOS Release 12.0(1) cannot parse the new copy EXEC command syntax that was introduced to parse URL syntax. Specifically, file names of the form tftp://171.69.245.105//c1600-ny-l.121-3 cannot be parsed. In this example, the system would attempt to load /c1600-ny-l.121-3 from the server at address 171.69.245.105. Because of the incorrect directory and filename pair, the copy will fail.
  Workaround: Run the ROM-based BootHelper image directly instead of invoking it from Flash load helper.
  A Cisco 7200 series router reloads with a bus error at PC 0x604A57E8, address 0xFBB4984. Logs will display the following message:
  %SYS-3-NOELEMENT: data_enqueue:Ran out of buffer elements for enqueue -Process= "<interrupt level>", ipl= 4 -Traceback= 60467520 604752C0 60350B68 60353AF8 6035580C 603558D0 %SYS-3-NOELEMENT: data_enqueue:Ran out of buffer elements for enqueue -Process= "<interrupt level>", ipl= 4 -Traceback= 60467520 604752C0 60350B68 60353AF8 6035580C 603558D0
  There is no workaround.
  A Cisco 7507 router that is running Cisco IOS Release 12.2(3) might reload because of a bus error:
  System returned to ROM by bus error at PC 0x0, address 0x0.
  There is no workaround.
  A Cisco MC3810 multiservice access concentrator might experience a software-forced reload with the following messages at the end of the backup BRI connection:
  PC 0x42A78 MC3810 Software (MC3810-A2ISX-M), Version 12.1(2), RELEASE SOFTWARE (fc1)
Image text-base: 0x00023000, data-base: 0x00AF21D4
  The stack trace from system failure shows the following message:
  FP: 0x117B140, RA: 0x42A78 FP: 0x117B178, RA: 0x270FC4 FP: 0x117B1B0, RA: 0x14C560 FP: 0x117B1E0, RA: 0x14C728 FP: 0x117B1E8, RA: 0x26E1DC FP: 0x117B1F0, RA: 0x24CBBC FP: 0x1032644, RA: 0x26DE78 FP: 0x1032654, RA: 0x26DE78
  There is no workaround.
  A Cisco AS5300 access server might reload through a bus error while running Cisco IOS Release 12.1(3.1). There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1(1a) receives the following bus error message:
  System returned to ROM by bus error at PC 0x605430A8, address 0x0 logs indicate traceback= 6038550C 606A3610 60335984 6033CC70 MDT: %IPFAST-2-PAKSTICK: Corrupted pak header for Virtual-Access431, flags 0x80
  There is no workaround.
  If a Cisco 7200 series router with many PA-MCxT1 cards has ISDN configured, some ISDN interfaces might fail to initialize correctly. An ISDN_UNEXPECTED_EVENT message is logged and the ISDN status stays at TEI-ASSIGNED.
  Workaround: Shut/no shut the interface.
  When the BRI interface is up, voice is not supported on E1 0 and E1 1 controllers. There is no workaround.

DECnet

  In certain topologies with DECnet IV routers and the cluster alias, a node might not be removed from the routing table of other routers when it is removed from the LAN. A loop condition might occur in which other routers see that node through each other. There is no workaround.

EXEC and Configuration Parser

  If priority queueing is enabled and there is an active ISDN call, changing the configuration of the priority-list queue-limit global configuration command will drop the call. There is no workaround.
  Privilege EXEC level 1 commands disappear from the configuration in Cisco IOS Release 12.1. There is no workaround.

IBM Connectivity

  If you connect a Route Switch Module (RSM) to a PIX Firewall in a multiring environment and configure the RSM using the multiring trcrf-vlan interface configuration command and the multiring all interface configuration command, the RSM might lose the connection to the PIX Firewall. In this situation, the RSM cannot ping the PIX Firewall. A sniffer trace reveals that the RSM is sending the Address Resolution Protocol (ARP) request as a single route explorer and that the PIX Firewall is answering with a directed frame. However, the control byte in the Routing Information Field (RIF) of this directed response is set to 46 instead of 06 (where 6 is the length of the RIF), and the RSM discards the packet.
  Workaround: Configure the RSM by entering the multiring all all-routes interface configuration command. This command will instruct the RSM to send the initial ARP request as an all-route explorer, and the PIX Firewall will respond with a directed frame with the control byte in the RIF set to 06. The RSM will accept this frame.
  With data-link switching (DLSw) on a Cisco Catalyst 6k-MSFC2 processor router, memory might be used inefficiently for some traffic patterns. This situation is not a memory leak because the memory will return when the DLSw traffic is stopped. This situation can lead to other problems because other Cisco IOS software features might be memory-starved. There is no workaround.
  The dlsw mac-addr global configuration command is not added to the configuration when it is entered in Cisco IOS Releases 11.3T, 12.0, and 12.1. There is no workaround.
  When you bridge IP on a Cisco 4700 series router with and without a Routing Information Field (RIF), and Systems Network Architecture (SNA) traffic is passed between Token Ring and Ethernet, IP packets intermittently fail to pass through the router when source-bridge spanning is added to the Token Ring interface to allow single route explorers (SREs) to be passed into the router. There is no workaround.

Interfaces and Bridging

  If you configure Gigabit Ethernet with IEEE 802.1Q for virtual LAN (VLAN) application, traffic might not be able to pass through the interface. There is no workaround.
  PPP stack Compression might cause connectivity problems on a Cisco AS5300 or AS5400 universal access server. There is no workaround.
  ATM subinterfaces might appear as INACTIVE when you use the show atm vc privileged EXEC command. This symptom occurs when you add new subinterfaces to the router. Using the shut and the no shut commands on the subinterface do not correct the problem.
  Workaround: Use the shut command followed by the no shut command on the main interface in order to bring the state to ACTIVE.
  On a Cisco router that is configured with a Packet OC-3 Interface Processor (POSIP) module on a 7206VXR interface in automatic protection switching (APS) protect mode, the APS POS interface might drop an unusual high number of packets that will be recorded as output drops. There is no workaround.
  TCP frames that are 4096 bytes long might become corrupted when switching between two FDDI interfaces. Payload is corrupted at offset 4028.
  Workaround: Enable the ip tcp path-mtu-discovery command.
  With IP Cisco Express Forwarding (CEF) running, a router processes frames received on an Inter-Switch Link (ISL) trunk even when the frame is not destined to it. This can cause a Layer 3 storm, as packets destined for an IP address on the same subnet as the sender are processed by the router and routed back out the interface on which they arrived. The problem is most often seen when there is a Spanning Tree change in the Layer 2 switch core, as the MAC table will be flushed and frames are flooded out all ports and arrive at the router.
  Workaround:Use 802.1Q encapuslation instead of ISL.
  Alternate workaround: Disable CEF.
  When two SNASwitch routers are connected over Inter-Switch Link (ISL) to a Cisco Catalyst 5000 or 6000 series switch to access a common mainframe, the SNASwitch routers establish a link between themselves regardless of the configured destination MAC (DMAC) address on the SNASwitch link statement or the Fast Ethernet interfaces if the mainframe is unavailable.
  Workaround: Add a bridge-group statement to the Fast Ethernet subinterface on the router. No bridge protocol is required. With this configuration change, the router no longer replies to incoming frames.
  Alternate workaround: Configure a permanent MAC address in the Catalyst switch Content Addressable Memory (CAM) table to prevent explorers from flooding the ISL ports.
  Cisco 7200 and 7500 series routers (and possibly other series) will "glitch" (pulse-low) the request to send (RTS) signal when "pulse-time" is coded for the Data Terminal Ready (DTR) signal. In some instances, RTS follows the behavior of DTR. Other times, RTS will simply glitch low for a few milliseconds.
  This situation might cause interoperability problems with particular data communications equipment (DCE) equipment, or if particular interface wiring (for example, looping RTS back to DCD) is used.
  Workaround: "Dummy up" RTS with a breakout box.
  A packet drop occurs on all PRI interfaces that are PA-MC-4CT1s in a Cisco 7513 router. The rate of drop is regularly one packet every two packets. There is no workaround.
  A Cisco 7200 series router's High-Level Data Link Control (HDLC) encapsulated interface on Packet-Over-SONET (POS) PA-POS is never up after reloading unless you configure the cdp enable or clock source internal interface configuration commands.
  Workaround: Use the clear interface EXEC command or a sequence of the shut and no shut interface commands.

IP Routing Protocols

  If h323 is configured on a remote interface of a router that is configured as a standby gatekeeper, the local interface Enhanced Interior Gateway Routing Protocol (IGRP) might not advertise a route to the remote network. If the standby gatekeeper router in this situation is a Hot Standby Router Protocol (HSRP) router that takes over for the HSRP active router, the other routers on the local network will lose their route to the remote network.
  Workaround: Configure a static route to the remote network with the HSRP virtual address as the next hop.
  IP Network Address Translation (NAT) fails to use addressed from the defined pool. Instead, it uses an address that was already used for static mapping. There is no workaround.
  A point-to-point sub-interface can be up/up but from a routing protocol perspective, the interface is in a down state. The problem appears with "ip unnumbered sub-interfaces." A shut/no shut of the interface does not fix the problem, but restarting the routing process (using the clear ip route * EXEC command for static routes) does correct it.
  A related issue occurs if a sub-interface is marked down (result of Operation, Administration, and Maintenance (OAM) cell missing, Frame Relay-permanent virtual circuit (FR-PVC) down), the routing process is not informed and it becomes necessary to wait for the routing protocol to detect the adjacency down (missing Hellos, Updates).
  Workaround: Use "ip numbered sub-interfaces."
  In intercept mode, TCP intercept might not work correctly with Network Address Translation (NAT). There is no workaround.
  A Cisco router learns a default route from the Border Gateway Protocol (BGP) and redistributes this route into the Enhanced Interior Gateway Routing Protocol (EIGRP). When the interface through which the default route is learned is cleared manually, EIGRP stops advertising the default, even though default route is in the routing table and the EIGRP topology table.
  Workaround: Enter the clear ip route {0.0.0.0} EXEC command.
  The router has a CPU hog when there are a large number of link-state advertisement (LSA) deletions. Two autonomous systems (AS) were accidentally merged. When they were restored, the CPU hogged while removing 60,000 LSAs. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1(1) or a later release and that has fast switching enabled might not forward BOOTP back to the client. The IP helper address cannot be a broadcast address (it must be the address of the server). There is no workaround.
  Static outside Network Address Translation (NAT) might not work with Cisco Express Forwarding (CEF) if CEF was active before NAT. There is no workaround.
  A Cisco router that is running Enhanced Interior Gateway Routing Protocol (EIGRP) with the stub feature on might have a route that is active and not waiting for replies. This situation only occurs in networks where all of the EIGRP neighbors are declared as stub.
  Workaround: Remove the EIGRP stub feature or clear the IP EIGRP neighbors.
  The following syslog might be observed at the rendevous point (RP) of a multicast routing network running Protocol Independent Multicast (PIM) sparse mode:
  %PIM-5-REG_ENCAP_INVALID: Bad register from <IP-address> for (<IP-address>, <Class-D-IP-address>). Trace = ....
  This situation occurs when a host sends a large packet that is fragmented before it arrives at the rendevous point. There is no workaround.
  In some circumstances, the show ip bgp vpnv4 all EXEC command might give additional wrong prefixes out of a single route. There is no workaround.
  IP accounting output packets might be fast-switched, instead of using Cisco Express Forwarding (CEF) switching. There is no workaround.
  After Cisco IOS software bootup, Open Shortest Path First (OSPF) might create a link-state advertisement (LSA) for an interface that is in the "Interface Up, Line protocol Down" state. This problem does not occur if you administratively shut down the interface at boot time and then later bring it up.
  Workaround: Administratively shut down the interface, save using the copy system:running-config nvram:startup-config command, and reboot the router.
  Alternate workaround: Exclude the interface address from OSPF by removing the network statement or restricting your inverse mask.
  Cisco 7513-A and 7513-B routers use Hot Standby Router Protocol (HSRP). After the Cisco 7513-A router reloaded, the Cisco 7513-B router became active and reloaded later. This problem occurred three times when running Cisco IOS Release 12.0(7)T. Priority queue and traffic-shaping were added to all CT1 interfaces recently. There is no workaround.
  Dial-in users (Group-Async1) cannot reach outside networks if the outside interface has an IP address that is included in the following range:
  130.134.192.1 through 130.134.209.254
  Using different subnet masks appears to have no impact on the problem. Masks have been tested ranging from 16-bit to 30-bit with no change.
  Cisco IOS Releases 12.0(3)T IP/FW/IDS and 12.1(2) IP/FW/IDS have been tested with the same results. The problem has also occurred in Cisco IOS Release 11.0(3) IP/FW/IDS. Changing the Network Address Translation (NAT) configuration to use a numbered Group-Async1 interface and overloading the outside interface address appear to produce no improvement in the problem.
  Workaround: Use a different outside address.
  Open Shortest Path First (OSPF) demand-circuit configured concurrently on a link that is also running Multilink PPP does not work. The shortest path first (SPF) algorithm keeps executing when the physical link changes status and sends updates toward the other peer.
  Hello packets are correctly suppressed over the link. There is no workaround.
  When Simple Network Management Protocol (SNMP) polling the BGP4-MIB on a Cisco 7206VXR router with Cisco IOS Release 12.1(1) or Release 12.1(2), the polling stops at bgpIdentifier:
  snmpwalk ip_address read_comm_string .1.3.6.1.2.1.15
  15.1.0 = OCTET STRING: Hex: 0E 15.2.0 = INTEGER: 16922 15.3.1.1.63.108.125.10 = IpAddress: 0.0.0.0
  [snipped for brevity]
  15.3.1.21.63.108.125.10 = INTEGER: 60 15.3.1.21.63.240.0.1 = INTEGER: 60 15.3.1.21.63.240.0.2 = INTEGER: 60 15.3.1.22.63.108.125.10 = INTEGER: 5 15.3.1.22.63.240.0.1 = INTEGER: 30 15.3.1.22.63.240.0.2 = INTEGER: 30 15.3.1.23.63.108.125.10 = INTEGER: 5 15.3.1.23.63.240.0.1 = INTEGER: 30 15.3.1.23.63.240.0.2 = INTEGER: 30 15.3.1.24.63.108.125.10 = Gauge: 0 15.3.1.24.63.240.0.1 = Gauge: 0 15.3.1.24.63.240.0.2 = Gauge: 4 15.4.0 = IpAddress: 63.108.125.16
  There is no workaround.
  The show ip nat translation verbose EXEC command might cause a Cisco router to reload.
  Workaround: Set the terminal length 0 command before executing the show ip nat translation verbose EXEC command.
  When generic routing encapsulation (GRE) traffic has to be encrypted and Cisco express forwarding (CEF) is applied on the tunnel interface, Enhanced IGRP (EIGRP) hellos sent by the remote side over the GRE tunnel are not seen anymore. There is no workaround.
  A Cisco 7200 series router running Cisco IOS Release 12.1(2) and Network Address Translation (NAT) might have high CPU utilization. In the show processes cpu EXEC command, an important part of the CPU is being used by IP Input although the router is configured to fast-switching or cef. There is no workaround.
  When forwarding Dynamic Host Configuration Protocol (DHCP) requests with the ip helper-address interface configuration command through a virtual template, DHCP offers are not forwarded back by the server side router if both server side and client side router have fast-switching enabled on the virtual template interface.
  Workaround: Disable fast-switching on one of the routers (no matter which one).
  When multiple unequal cost paths exist for type 3 link state advertisements (LSAs), shortest path first calculation might trigger an unnecessary route flap in the routing table for these LSAs. There is no workaround.
  When Network Address Translation (NAT) is configured, the FTP ls command might not work. There is no workaround.

Miscellaneous

  When a large number of virtual channels (VCs) are simultaneously congested on a PA-A3 port adapter, a Cisco router might run out of input/output (I/O) memory because of excessive buffering on per-VC queues, and the router might reload.
  Workaround: Upgrade the memory in the router, or configure per-VC queueing parameters to allow a smaller number of packets in the queue.
  Serial tunnel (STUN) Synchronous Data Link Control (SDLC) running through PA-4T+ port adapter might fail on a Cisco 7206 router with half-duplex (HDX) configured. There is no workaround.
  The RADIUS Call Filter Attribute (attribute 243) does not work properly. Instead of operating as an interesting traffic filter, it is operating as a data filter like attribute 242. There is no workaround.
  The D-channel interface gets reenabled automatically after it is shut.
  Workaround: Shut the controller instead of the D-channel interface.
  In Cisco IOS Releases 12.0 and 12.1, the multichannel T1/E1 port adapters show the layer 1 status in the ACTIVE state in the show isdn status [output] EXEC command even after the D channel is shut down.
  Workaround: Shut down the T1 or E1 controller.
  A Cisco 7500 series router might drop encrypted packets if those packets are routed over a Token Ring port adapter. There is no workaround.
  A Cisco router might reload with PPP reliable-link (LAPB) and compression and fancy queuing enabled.
  Workaround: Use first-in-first-out (FIFO) queueing.
  A Cisco uBR924 cable access router loses H.323 configuration after reloading when it is configured to use an H.323 gatekeeper to place phone calls and the following three lines are added to the configuration on the cable interface:
  h323-gateway voip interface
h323-gateway voip id gklocal ipaddr 6.3.0.2 1719
h323-gateway voip h323-id StingerA@cisco.com
  Ater the modem is rebooted, the commands might not be processed correctly because the commands are applied to the interface before the interface acquires an IP address from Dynamic Host Configuration Protocol (DHCP).
  Workaround: Reenter the missing configuration.
  If the serial interface for a T1 PRI D channel (for example, interface serial <x>:23) is configured with the no keepalive TN3270 configuration command, the call management task consumes 99 percent of the CPU. There is no workaround.
  If there is a large network delay for Voice over IP (VoIP) calls on a Cisco AS5300 universal access server that is running Cisco IOS Release 12.0(6)T or later releases, DSP might go into voice mode (where an audio path opens) approximately 1 or 2 seconds after the call connects.
  Workaround: Enter the no ip tcp delayed-ack global configuration command.
  The privilege interface global configuration command might execute slowly and cause high CPU utilization although the command process will eventually complete itself. This command is normally executed during the configuration task and during the initial configuration process after a router reload. There is no workaround.
  A Cisco router might not be able to place outgoing calls if you use the rotary group line configuration command and the interface dialer global configuration command after the router reloads. The router exhibits the following error message:
  No free dialer - starting fast idle timer
  Workaround: Enter the shutdown interface configuration command followed by the no shutdown interface configuration command.
  When an initiating Cisco router is configured to use IP Security (IPSec) with the crypto cisco pregen-dh-pairs command, the initiator will cease to operate if the connection is placed in a down state then returned to an up state by a remote peer. If this occurs, the Cisco router will appear to pause indefinitely, and you will need to restart the router.
  Workaround: Enter the no crypto pregen-dh-pairs command.
  A Tag Forwarding Information Base (TFIB) table might show directly connected routes as being remote. There is no workaround.
  A Cisco router with the fras backup dlsw interface configuration command enabled will not be able to reconnect Frame Relay Access Support (FRAS) border access node (BAN) devices through data-link switching plus (DLSw+) after a failure on the main Frame Relay link that is used for the FRAS connection. In this situation, the router will reload while trying to reconnect the FRAS BAN devices through DLSw+. FRAS boundary network node (BNN) devices are not affected. There is no workaround.
  On a Cisco 3640 router with Digital Modems (DM) modules, modems in the Network Module (NM) turn into a BAD state after a few days of operation. There is no workaround.
  A Cisco 7200VXR series router resets after every 4 minutes because of a bus error "pc 0x64049A338 , address 0xFFFFFFFF." There is no workaround.
  A Cisco 7507 router might reload after you enter the configure {memory} privileged EXEC command.
  Workaround: Configure the router manually.
  When Cisco 4-port and 8-port Asynchronous/Synchronous Serial Network Modules (NM-4A/S, NM-8A/S) are configured with the physical-layer {async} interface configuration command, an asynchronous physical layer, the asynchronous interface does not set maximum transmission units (MTUs) correctly. You can observe this situation in the "buffer size" field in the output of the show controller command. The buffer size is always shown as 1500. There is no workaround.
  If you attempt to make interzone H.323 calls using gatekeepers and proxies through Context-Based Access Control (CBAC) inspection, calls are rejected. Calls using gatekeeper but no proxies connect correctly. There is no workaround.
  Very high values occur for the LostPacket counter when the show call history voice privileged EXEC command is entered. There is no workaround.
  Cisco Express Forwarding (CEF) output IP features are invoked twice for packets in the label imposition forwarding path. They are invoked once when they are IP packets and once after labels have been imposed.
  Refer to the following examples:
  Packets match both IP and Multiprotocol Label Switching (MPLS) rate-limit access lists. Packets also match non protocol-specific match criteria (such as implicit "match all" packets) twice.
  Rate-limit interface configuration command actions (including set-prec-transmit) might be invoked for packets even though the packets transmit as MPLS packets.
  The show policy-map interface command output might contain double counts for a class.
  There is no workaround.
  A Cisco multichannel T1/E1 port adapter does not support switching of Multiprotocol Label Switching (MPLS) packets. There is no workaround.
  This problem occurs when the modem startup-test global configuration command is configured and there are v120 calls.
  Workaround: Disable the modem startup-test command.
  An ATM virtual circuits (VCs) might stop sending traffic other than for Operation, Administration, and Maintenance (OAM) cells.
  Workaround: Reset the VCs by using the shut and no shut interface commands on the subinterfaces or by reconfiguring the subinterfaces.
  A Cisco 3640 router that is running Cisco IOS Release 12.0(7)T and that is configured with voice modules might experience poor performance on the serial interface.
  Workaround: Remove the voice cards.
  The H.245 open logical channel (OLC) audio capability "maximum frames per packet" value is statically defined and not dynamically updated to reflect the codec bytes configuration. The codec bytes (payload) can be configured under the dial peer and under the voice class codec configuration.
  Another related problem on the local gateway (GW) is that the GW does not consider the remote endpoints, H.245 terminal capability set (TCS) audio capability "maximum frames per packet" value when opening a logical channel to that remote endpoint. Local GW will OLC and then transmit packets larger than the remote endpoint advertised as its maximum receive size capability.
  Finally, upon codec negotiation, the digital signal processor (DSP) should be indicated to transmit/receive at the dynamically determined payload size, instead of the configured bytes. There is no workaround.
  While running PPP over ATM, accounting messages occasionally have a missing IP Address for start/stop records. There is no workaround.
  During router bootup, if Gigabit Ethernet Interface Processor (GEIP) is installed, GEIP might fail to be brought into service when the router reloads.
  Workaround: Replace the GEIP board.
  The difference between Real-Time Transport Protocol (RTP) timestamps and sequence numbers causes voice quality issues. There is no workaround.
  A Cisco 3600 series router with an ATM interface may unexpectedly reload due to an error as follows:
  %ALIGN-1-FATAL: Corrupted program counter pc=0x0, ra=0x6115D44C, sp=0x61BFF008
  There is no workaround.
  When an ISDN router receives a random IP address from the Internet service provider (ISP), disconnects due to idle time, then reconnects and gets a different IP address from the ISP, the IP Security (IPSec) tunnel will fail to build because the secure addresses already exist on the head end device. There is no workaround.
  When downstream physical units (PUs) connect into the SNASWITCH router and are requesting Dependent Logical Unit Requester (DLUR) support, the first connection from the downstream unit is successful. After disconnecting and then attempting to reconnect, the connection is rejected.
  Workaround: Create a new Switched Major node/PU definition in virtual telecommunications access method (VTAM) to reconnect. The SNASWITCH does not clean up the original PU connection, and rejects the reconnect, indicating that an invalid exchange identification (XID) was sent in.
  A Cisco 3600 series router with MICA modems used for Dial-Out does not allow the use of dial modifiers, such as Comma. There is no workaround.
  A route switch module (RSM) in a Cisco 5500 series router might reload because of memory corruption. The reload appears to be function of an encapsulation error and is preceded by memory allocation errors. There is no workaround.
  When a single T1 link in an Inverse Multiplexing over ATM (IMA) environment has surpassed the differential delay limit on a Cisco 3600 series router, the other "good" T1 links might be deactivated and the IMA group will run on the single T1 link with the excess delay.
  The expected behavior is that the "bad" T1 link with the added delay would be deactivated and the three "good" T1 links would remain operational. The problem behavior only occurs on the T1 links. E1 links behaved as expected. There is no workaround.
  When IP Security (IPSec) sessions are negotiated, the peer router may reload, on rare occasions. There is no workaround.
  A Cisco AS5300 series access server might reload when it attempts to access a RADIUS server. There is no workaround.
  About 25 percent of GSM faxes sent over a GSM cellular network between Canada and Prague fail. You do not see an "act_wait_ack_ack" on the fax that fails, indicating that the router does not get a "fax ack." The router goes to start fax_mode and the fax fails. There is no workaround.
  A new field named "Incomplete calls" has been added to the output of the show dial-peer voice privileged EXEC command.
  Outgoing call legs through a dial peer are classified as follows:
  Cause codes are as defined in Q.931. All values below are in decimal.
  A call is termed successful if the disconnect cause code is normal (value of 16). Calls with all other disconnect cause codes are termed failures. Incomplete calls are the subset of failed calls that have a disconnect cause of one of the following:
  17 (user busy)
  18 (no user response)
  19 (Ring no answer)
  Note that the sum of successful and failed calls still equals the total number of calls.
  Incomplete calls are a subset of failed calls in order to maintain compatibility with existing network management applications.
  The ATM interface pauses indefinitely on a Cisco 4500 series router.
  Workaround: Clear the ATM interface to temporarily resolve the issue. ATM interface locks up every 45 minutes (average).
  A Cisco 3640 router, running Cisco IOS 12.1(1.6), might get a memory corruption, upon entering "show traffic-shape queue seriql 1/6.11". There is no workaround.
  Under stress, a Cisco 3660 router reloads because there is no free event structure available from htsp_free_event_Q. There is no workaround.
  A Gigabit Ethernet Interface Processor (GEIP) on a Cisco 7500 series router may experience receive problems causing it to pause indefinitely. Workaround: Disable dCEF on the GE interface.
  When a variable bit rate (VBR) virtual path connection (VPC) is configured on a tag-switching ATM subinterface using the pvc interface configuration command, the traffic through the VPC is not shaped.
  Workaround: Configure the VBR parameters on the main interface using the atm pvp interface configuration command.
  When a generic routing encapsulation (GRE) tunnel in a simple setup is not encrypted but traffic flowing through it should be, no offers are accepted. There is no workaround.
  A Cisco 3660 router running the firewall feature set might experience a reload due to bus error if Real Time Streaming Protocol (RTSP) inspection is configured. There is no workaround.
  Analog E&M (recEive and transMit or ear and mouth) trunk ports on a Cisco 2600 series or 3600 series voice module may be in a nonoperational state during high call volume. This situation results in calls either not being routed to the port in question, or inbound telephony calls not receiving the proper call handling response from the port such as wink or dialtone.
  Use of the show voice port privileged EXEC command typically will show the port in an "Operation State" of UP with possibly an "active" seizure on the port when it should be in an idle state.
  To recover from this situation, reload the router. There is no workaround.
  An SA-ISA encryption card is not recognized on boot in a Cisco 7200 series router (with NPE-225 or NPE-300).
  Workaround: Use the boot image for 12.0(4)T and the ISA module loads and works (for the NPE-300). Use an NPE-200 instead of the NPE-225.
  The caller does not hear a busy tone when the called party is offline. Eventually, the ISDN switch disconnects the call and plays an invalid message back to the caller.
  This problem occurs when the originating gateway is connected to ISDN and the terminating gateway is connected to E&M or foreign exchange station (FXS) .
  Workaround: Use two-stage dialling.
  The statistics of the ifOutOctet on the Frame Relay subinterfaces are incorrect. The outoctets are displaying the same value for both subinterfaces. There is no workaround.
  On a Cisco 7500 router running Cisco IOS Release 12.0(7)T, packets destined for Information Resource Engineering (IRE) clients are dropped when an IP Security (IPSec) tunnel is terminated on an interface of a router or PIX firewall and the traffic that came through the tunnel is sent back out the same interface to another network. The traffic does not completely go through the device. There is no workaround.
  When a Route Switch Processor/Versatile Interface Processor (RSP/VIP) platform runs the Multilink PPP (MLPPP) feature, a certain scenario might cause the VIP to reload.
  There is no workaround.
  When calling from one MC3810 multiservice concentrator on the T1 to another over Frame Relay (or when hairpinning passing DSP), the called party will clip the calling party voice when the parties talk at the same time. There is no workaround.
  When the codec preference feature is used, the digital signal processor (DSP) might get into a faulty state and display error messages such as:
  %C542-1-NO_RING_DESCRIPTORS: No more ring descriptors available
  To recover, the router must be reloaded.
  This problem only happens when the codec preference lists on the voip dial-peers on the two gateways have the codecs ordered differently: for example, if the originating gateway has g711ulaw as codec preference 1, and the terminating gateway has g729 as codec preference 1.
  Workaround: Use similarly ordered codec preference lists on both sides.
  Alternate workaround: Use the codec dial-peer configuration command to enforce a particular type of codec.
  Krb5_telnet fails with the following error message:
  kerberos_server_auth: Couldn't authenticate client from xxx
Failed to generate authentication data!
  After attempting a "telnet router2 /encryption kerberos" (which fails), entering the show kerberos creds EXEC command prompts a "No kerberos credentials" message. The credentials have been lost. There is no workaround.
  When a Cisco 800 series router shares the S/T-bus with 4 or more other Terminal Equipments, it might fail to send certain Layer 2 messages due to D-channel collisions (that is, it might not send ID Check Reply in response to ID Check Request). As a result, the ISDN switch might drop Layer 3 calls connected to the 800 series router.
  Workaround: Enable some debugging on the 800 series router console (for example, debug bri and debug isdn q921). Doing so seems to help the situation when there are D-channel collisions.
  A software-forced reload might occur in Cisco IOS Release 12.0(7)T because of a memory corruption. There is no workaround.
  A Cisco 7200 series router might pause indefinitely when its PA-2FE-ISL interface receives cyclic redundancy check (CRC) error frames at a high rate. There is no workaround.
  Resolution of recursive routes by Cisco Express Forwarding (CEF) may add 1 to 15 seconds to the end-to-end route convergence time. There is no workaround.
  An FTP session fails for active data connections. Passive connections work fine, and both work if the access list is open for the data connection. There is no workaround.
  When you change the maximum transmission unit (MTU) size on an asynchronous interface, large packets are not coming through even though they are correctly negotiated by PPP. There is no workaround.
  When Frame Relay payload compression FRF.9 is enabled between a Cisco 1750 router and a Cisco 2610 router, audio is lost in one direction. FRF.9 must be removed, and the 1750 router must be reloaded. There is no workaround.
  On images 121-1.5 and 121-2a.T1 for a Cisco 5300 series router, an error message similar to the following appears when the gateway tries to register with a VocalTec gatekeeper:
  Gateway X failed to register with Gatekeeper Y even after 2 retries.
  There is no workaround.
  When a Cisco 7200 series router is used as a label switch controller (LSC) for a BPX and as a Label Edge Router (LER) (edge device) at the same time, all packets incoming on the XTAGAtm interface are being process-switched. There is no workaround.
  This problem occurs on a special release image (C827V-Y6-M) running with Cisco IOS Release 12.1(1)XB when Resource Reservation Protocol (RSVP) is enabled on the dialer interface with the req-qos controlled load configuration command.
  Workaround: Remove the req-qos controlled load command from the dialer interface.
  If the crypto ca trusted-root {label} hidden global configuration command is entered without at least one sub-command, the router might accept revoked certificates during Internet Key Exchange (IKE) negotiations which use the RSA-signature method of authentication.
  Workaround: If an incomplete crypto ca trusted-root {label} command is in your running or saved configuration, remove it by using the no crypto ca trusted-root {label} command or make it complete (if the command is used). Resave the configuration if necessary.
  Digital Signature Standard (DSS) keys generated in Cisco IOS Release 12.0(11) cannot be used or exchanged in Cisco IOS Release 12.1(2). There is no workaround.
  A router running a stress-test of Cisco Transaction Connection (CTRC) reloads after many of these error messages:
   2d01h: %TXCONN-3-TXEXCEPTION: Error occurred from transaction 6F of client 10.10.10.1 connected to server 1, exception type is 11
   2d01h: %CTRC-3-APPCVERB: Unexpected APPC error: opcode 000B, primary code 0300, secondary code 00004C08
   2d02h: %CTRC-4-APPCWARN: A conversation cannot be allocated because of a temporary condition, such as a link failure.
   2d02h: %CTRC-4-APPCWARN: A conversation cannot be allocated because of a temporary condition, such as a link failure.
   2d02h: %CTRC-4-APPCWARN: A conversation cannot be allocated because of a temporary condition, such as a link failure.
   2d02h: %SNASW-3-PS_LOG_4: PROBLEM - 5226 - Conversation ended by session outage
   2d02h: %CTRC-4-APPCWARN: A conversation cannot be allocated because of a temporary condition, such as a link failure.
   2d02h: %TXCONN-5-TARGETDOWN: CICS server QAMELBA has become unavailable. Recovery is in progress.
   2d02h: %TXCONN-3-TXEXCEPTION: Error occurred from transaction 5D5 of client 10.10.10.1 connected to server 1, exception type is 11
   2d02h: %TXCONN-3-TXEXCEPTION: Error occurred from transaction 627 of client 10.10.10.1 connected to server 1, exception type is 11
  There is no workaround.
  A Cisco 3640 router that is running Cisco IOS Release 12.0(7)XK1 or 12.1(2)T with an NM-2FE2W Network Module and a WIC-1B-S/T WAN interface card fails to make outgoing ISDN calls after working correctly for some time.
  The following message indicates the problem:
  BRI0/0: In state F7, Received code Power-up BRI0/0 : Unexpected indication (1C) in f7 code
  Workaround: Reload the router when the problem appears.
  Calls cannot be made on a Cisco 3600 series router configured for connection trunk using E&M voice modules after a few days. This situation affects one or two trunks. Performing a shut/no shut of the ports does not resolve this problem. The router must be reloaded. This problem is intermittent. There is no workaround.
  On a Cisco 7500 series router running Cisco IOS Release 12.0(3.1), the Versatile Interface Processor (VIP) might reload if encrypted connections are used for an extended period of time with feature acceleration enabled.
  Workaround: Disable feature acceleration globally or
  Alternate workaround: Turn off the ip route-cache flow command on the interfaces that are running crypto.
  A Cisco 7200 router may reload by bus error while trying to access 0x50000000 in ct3sw_rx_interrupt(). The bus errors always point to the same address.
  Reloads have been reported during internal testing on 7500 also.
  Workaround: Disable the ip cef global configuration command.
  With the ip pxf command configured or on by default, together with any of Cisco Express Forwarding flow accounting, weighted fair-queueing, or weighted random early detect, packet drops might occur at relatively light loads. The load may be as little as 25 percent of that at which packet drops would occur with the no ip pxf command configured. There is no workaround.
  A system might unexpectedly restart when the atm inarp command is disabled on an interface.
  A router also reloads when you configure a permanent virtual circuit (PVC) for "inarp" on an interface with an IP address and a non-zero inarp timeout value, and then enter just "inarp" for that PVC. There is no workaround.
  Error messages similar to the following text appear when a Fast Ethernet port adapter (FE PA) is being used in Versatile Interface Processor 2 (VIP2):
  %CBUS-3-CMDDROPPED:Cmd dropped, CCB 0x5800FF20, slot 0, cmd code 36
  The likelihood and frequency of these messages appears to increase as the number of subinterfaces on the FE interface is increased. The likelihood is also greater when a two-port FE PA (PA-2FEISL) than a one port FE-PA (PA-FE) is used. Traffic does appear to correctly traverse the FE port even when these messages are seen on the router console. There is no workaround.
  Cisco 7500 routers using Cisco Encryption Technology (CET) on Versatile Interface Processors (VIPs) may not be able to set up encrypted connections in networks with redundant links between encrypting peers.
  There is no workaround.
  After a router is reloaded, Multiprotocol Label Switching (MPLS)/VPN reachability might be broken.
  Workaround: Use the clear ip route vrf EXEC command after reloading the router.
  Mobile phone calls come into router 1. The caller can hear one ring tone cycle. Then, the call is disconnected and the caller hears a fast busy signal.
  Sometimes, this situation occurs with foreign exchange station (FXS) calls that are generated on the PBX. There is no workaround.
  When a router has integrated routing and bridging (IRB) enabled and is bridging between the Token Ring interface where the station is connected and an ethernet interface and is routing all other interfaces, including the Token Ring where the BOOTP server is attached, the bridge virtual interface (BVI) interface is doing the routing between the bridged side and the routed side for IP.
  When using BOOTP, the router sends the BOOTPREPLY to the station with the destination MAC address bit swapped. The destination station never gets the reply. There is no workaround.
  After adding some VPN/ OSPF-BGP configurations, a router reloaded. There is no workaround.
  A Cisco 7500 series router running rsp-jsv-mz.121-3 restarted with a software- forced reload. The reload was caused by memory corruption.
  The following text was seen in the log portion of the crashinfo file:
  %SYS-3-BADBLOCK: Bad block pointer 62E84A98 -Traceback= 602D7134 602D87D8 607460C0 60746F0C 602CACB4 602CACA0
  %SYS-6-MTRACE: mallocfree: addr, pc 62E84AC0,607460B8 62E84AC0,6073F244 62E84AC0,30000034 62E84AC0,607460B8 62E84AC0,6073F244 62E84AC0,30000034 62B394B8,60000020 62B39464,60765734
  %SYS-6-MTRACE: mallocfree: addr, pc 62C1D240,60000026 62C1D1F8,607656C0 62B7D1C4,60000022 62B7D16C,607655D0 62A526EC,6000000E 62A52694,607654C0 62455738,60790B68 62D5E7B0,607654C0
  %SYS-6-BLKINFO: Freespace does not end at end of the pool blk 62E84A98, words 52 , alloc 6073F244, InUse, dealloc 0, rfcnt 1 -Traceback= 602D4ABC 602D714C 602D87D8 607460C0 60746F0C 602CACB4 602CACA0
  %SYS-6-MEMDUMP: 0x62E84A98: 0xAB1234CD 0x63 0x625B2E3C 0x61607200 %SYS-6-MEMDUMP: 0x62E84AA8: 0x6073F244 0x62E84B28 0x62E84A4C 0x80000034 %SYS-6-MEMDUMP: 0x62E84AB8: 0x1 0x607460B8 0x0 0x0
  The stack trace decode reveals the following:
  0x602EEFF4:abort(0x602eefec)+0x8 0x602ED68C:crashdump(0x602ed67c)+0x10 0x602D7168:validblock(0x602d6cb0)+0x4b8 0x602D87D8:free(0x602d86d8)+0x100 0x607460C0:igrp2_send_hello(0x60745f90)+0x130 0x60746F0C:igrp2_hello_process(0x60746d60)+0x1ac 0x602CACB4:r4k_process_dispatch(0x602caca0)+0x14 0x602CACA0:r4k_process_dispatch(0x602caca0)+0x0
  There is no workaround.
  A Cisco 3640 router running Cisco IOS Release 12.1(2) or 12.1(3) reloads with a bus error because of a corrupted program counter (pc=0x0, address=0x0). No stack trace is recorded. This situation occurs when configuring an IP Security (IPSec) tunnel over a Layer 2 Tunneling Protocol (L2TP) tunnel. The reload occurs systematically. There is no workaround.
  Unconfiguring Cisco Express Forwarding (CEF) on a Cisco 3640 router running Cisco Applications and Services Architecture (CASA) might cause a traceback. There is no workaround.
  A Cisco 7513 router that is running Cisco IOS Release 12.1(1a) reloads at the checkheaps_process:
  %SYS-4-SNMP_WRITENET: SNMP WriteNet request. Writing current co nfiguration to 10.100.10.100
validblock_diagnose, code = 2
current memory block, bp = 0x62199718, memory pool type is Processor
data check, ptr = 0x62199740
next memory block, bp = 0x62199778, memory pool type is Processor
data check, ptr = 0x621997A0
previous memory block, bp = 0x62199674, memory pool type is Processor
data check, ptr = 0x6219969C
  ========== Stack Trace ========
  -Traceback= 602E6434 602E4B9C 602CE718 602D0FF0 602D20C0 602D2308 602C103C 602C1028
  Decode Result ==============
  "rsp-jsv-mz.121-1a.symbols.gz" Uncompressing (gzip fmt) and reading rsp-jsv-mz.121-1a.symbols.gz rsp-jsv-mz.121-1a.symbols.gz read in Enter hex value: 602E6434 602E4B9C 602CE718 602D0FF0 602D20C0 602D2308 602C103C 602C1028 0x602E6434:abort(0x602e642c)+0x8 0x602E4B9C:crashdump(0x602e4b8c)+0x10 0x602CE718:validblock(0x602ce260)+0x4b8 0x602D0FF0:validate_memory(0x602d0f70)+0x80 0x602D20C0:checkheaps(0x602d205c)+0x64 0x602D2308:checkheaps_process(0x602d2208)+0x100 0x602C103C:r4k_process_dispatch(0x602c1028)+0x14 0x602C1028:r4k_process_dispatch(0x602c1028)+0x0
  There is no workaround.
  When a Media Gateway Controller (MGC) places a call on a b-channel that is already in use on a Cisco AS5300 functioning as part of a Network Access Solution (NAS), the NAS might reload with a software-forced crash. This problem can, for example, occur after a failover. There is no workaround.
  The gatekeeper does not accept prefixes longer than 11 digits. This is currently a design limitation. There is no workaround.
  A Cisco AS5300 series access server fails to update its tag forwarding table (TFIB) for dialup clients. You can verify this condition by comparing the TFIB of the Cisco AS5300 access server to that of an adjacent tag router. Refer to the following example:
  as5300#sh ip route conn 172.16.1.0/24 is variably subnetted, 2 subnets, 2 masks C 172.16.1.254/32 is directly connected, Virtual-Access20
  as5300#sh tag for 172.16.1.254 Local Outgoing Prefix Bytes tag Outgoing Next Hop tag tag or VC or Tunnel Id switched interface
  (tag entry missing in Cisco AS5300)
  Router#sh tag for 172.16.1.254 Local Outgoing Prefix Bytes tag Outgoing Next Hop tag tag or VC or Tunnel Id switched interface 118 113 172.16.1.254/32 0 Fa0/0 172.16.0.18
  (tag present in adjacent router)
  This situation causes traffic from the router adjacent to the Cisco AS5300 access server for that host to fail. There is no workaround.
  Cisco Express Forwarding (CEF) does not work properly when traffic is load-shared across the subinterfaces of the receiving interface. There is no workaround.
  When a SC2200 Signalling Controller in a "Thunderdial 2.2 (TD 2.2) solution" sends a Continuity Test (COT) request to a Cisco AS5300 functioning as part of a NAS, the outgoing modem calls fail on the Cisco AS5300. Workaround:Do not set a COT percentage on any SC2200 that will be servicing modem calls of this type. Run COT tests manually on these trunks with the test-cot command. The command can also be scripted to run automatically as a UNIX chronologically started job.
  Virtual Private Network/Multiprotocol Label Switching (VPN/MPLS) causes a router to reload in c7100_ip_to_feature_tagswitch. There is no workaround.
  A Cisco router running Cisco IOS Release 12.1(3) or 12.0(7)T might not calculate a User Datagram Protocol (UDP) checksum for a Dynamic Host Configuration Protocol (DHCP) offer frame that it is forwarding. This situation seems to only occur when the offer does not contain an IP address for the client, which forces the router to broadcast the response. This problem does not affect normal DHCP traffic, since a real DHCP frame from a DHCP server offers an IP address. It will affect portable execution environment (PXE) clients that are using LCCM servers to gather their configurations files. There is no workaround.
  While running a Cisco 7206 VXR router, you might experience memory corruption while trying to dial out to multiple sites over PRI interfaces. There is no workaround.
  Removing the PA-MC-2T1 port adapter while passing traffic through it might trigger a hardware glitch and cause Peripheral Component Interconnect (PCI) parity errors. The user should avoid the above operation. There is no software workaround.
  A Cisco 4500 series router experiences a software-forced reload. There is no workaround.
  A Cisco AS5200 access server reloads when pinged while configured for Continuity Testing (COT).
  Workaround: Do not use COT when making outgoing modem calls.
  A Cisco AS5300 series access server might pause indefinitely under heavy traffic. There is no workaround.
  A Cisco 7200 or 7500 series router with ATM-PA3 might stop forwarding packets on one or more VCs. The packets would show up as output drops on those VCs. These VCs appear stuck.
  This problem occurs because of a newer version of ATM-PA3 microcode. Refer to the output of the show controllers atm privileged EXEC command to see this version.
  For a Cisco 3640 Access Concentrator-60 MICA / 2 x NM-1FE2CT1-CSU with one port each on the second controller card configured for PRI & chanT1, calls on the chanT1 fail at regular intervals with the following traceback (decoded below):
  ea1-d1.pointshare.wan 7531:
%DSX1-3-M32_AR_TIMEOUT: Munich 32 action request timeout: Ctrl = 0x3D930003, Channel = 12
sea1-d1.pointshare.wan 7532: -Process= "CSM background task", ipl= 3, pid= 26 sea1-d1.pointshare.wan 7533: -Traceback= 601187B8 60118C0C 601183E0 601182CC 60118608 60111DD0 6019FEF0 6019C9D8 601986C8 6030AC64 6030AC50
  Trace decoded ----------- 0x601187B8:m32_errors_msg(0x601186c0)+0xf8 0x60118C0C:m32_ar_poll(0x60118aa0)+0x16c 0x601183E0:m32_execute_ar_for_modem(0x6011831c)+0xc4 0x601182CC:m32_release_modem_channel(0x60118100)+0x1cc 0x60118608:m32_init_channel_for_cas(0x60118418)+0x1f0 0x60111DD0:dsx1_alloc_cas_channel(0x60111c8c)+0x144 0x6019FEF0:csm_proc_idle(0x6019fdf4)+0xfc 0x6019C9D8:csm_cas_event(0x6019c710)+0x2c8 0x601986C8:csm_background_process(0x601983e8)+0x2e0 0x6030AC64:r4k_process_dispatch(0x6030ac50)+0x14 0x6030AC50:r4k_process_dispatch(0x6030ac50)+0x0
  The controller card has been changed, but the problem reappears.
  Workaround: Reboot the router.
  When a user dials a wrong number, the user gets fast busy instead of NUMBER UNOBTAINABLE. There is no workaround.
  While removing a Border Gateway Protocol (BGP) process, a router might reload if you have BGP redistributed into Open Shortest Path First (OSPF). This situation occurs only if BGP and OSPF have network statements covering the same interface.
  Workaround: Remove the redistribution of BGP from OSPF, wait a few minutes, and then remove the BGP process.
  When you run three different (banking) applications on a Windows NT PC & access a Cisco 2600 series NM-8AM using Cisco dialout utility, the first and second applications use 8-N-1. The third application (called OutsideView32) uses 7-E-1. If you want dynamic allocation (8-N-1 or 7-E-1) depending on the application accessing the host computer, the characters you type might not be recognized or displayed on the screen correctly when you connect to the host computer with OutView32 and start entering information at the logon screen.
  Installing version 1.2.2 of the modem firmware does not resolve this problem. You must add the characters, "-ol" to the initialization string that OutView32 sends to the modem. You must also add the chat reset command to the chat-script {script-name} at&f command configuration.
  After adding those initialization strings, you can type characters on the OutView32 logon screen and they will show up correctly. You can logon to the remote host, but only once, before the modem stops responding. If you switch from 7-E-1 to 8-N-1, then you only need to clear the modem for every alternative call. If you are using only 8-N-1, then there is no need to clear the modem for subsequent calls. There is no workaround.
  A software-forced reload occurs in Cisco IOS Release 12.1 when using a t3 circuit. There is no workaround.
  Several Cisco 1750 routers fail to boot or only come up in ROM Monitor (ROMMON) mode right out of the router. This situation might be the result of a bad batch of memory. There is no workaround.
  Outbound calls on a Cisco AS5800 series access server with a Voice over IP (VoIP) application might have a lower than expected Call Success Rate (CSR) when configured with foreign exchange station (FXS) channel associated signaling (CAS).
  This can be seen by accessing the trunk card with the dsip console slave {slot#} command and then executing the show trunk state {T1 trunk #} command. Idle trunks should be in a g_idle state. The following text shows the trunks stopped in a g_ring_ground state:
  as5800-1#dsip console slave 4
Trying Dial shelf slot 4 ...
Entering CONSOLE for slot 4
Type "^C^C^C" to end this session
  DA-Slot4#sh trunk stat 1
  Trunk slot 4, T1 line 1, 24 channels
State=UP, Enabled, Loopback=1, NIP is running
RS Unit=117, channels start=24
Table=0x6069C984, instance=0x6069C940
  Timeslot TX RX Trunk state
         abcd abcd
  1 0000 1111 g_ring_ground
2 0000 1111 g_ring_ground
3         0000    1111     g_ring_ground
  Also, with the debug voip ccapi inout command enabled, the disconnect cause code might be 0x22, as shown below:
  ccCallDisconnect (callID=0x1239A, cause=0x22 tag=0x0)
  Workaround:Try to shut/no shut the T1 controller or reload the router.
  A Cisco 7206VXR series router with a PA-MC-T3 module has spurious memory accesses that cause the router to occasionally reload. There is no workaround.
  When the queueing strategy on an output non-VIP (Versatile Interface Processor) interface is changed from FIFO to fair-queue or something else, label forwarding information base (LFIB) entries for all prefixes that are switched out that interface are not automatically removed from the VIPs in the system. This situation causes packets to those prefixes to be distributed-switched, thereby bypassing the queueing mechanism.
  Workaround: Cycle shut the interface.
  Alternate workaround: Toggle distributed tag switching with the ip cef or ip cef distributed global configuration commands.
  Alternate workaround: Toggle Multiprotocol Label Switching (MPLS) forwarding on the router.
  All of these workarounds could potentially cause some disruption to the traffic. There is no non-intrusive workaround.
  A Cisco 7206VXR series router with a PA-MC-T3 module has spurious memory accesses that cause the router to occasionally reload. There is no workaround.
  When a Cisco router configured for Airline Product Set Universal Terminal System (ALPS UTS) receives an incoming terminal's message that includes intermediate SYN characters (that is, SYN characters between SOH and ETX), the router increases its cyclic redundancy check (CRC) error counter and the terminal session might pause indefinitely.
  Some terminals calculates the Block Check Character (BCC) without including those intermediate SYN characters. (Some terminals do include those characters). However, a Cisco router includes intermediate SYN characterss in the BCC calculation. Consequently, the mismatch occurs and the router interprets the incoming frame as having the wrong BCC value.
  Symptons of this problem include CRC errors incrementing on the router's serial interface, unstable terminal sessions (hangs), and the following output when you run debug on the router, at the time of the problem:
  ALPS ASCU: Rx UTS INCORRECT_MSG (131 bytes + BCC) from asc u 22 on i/f Serial3
  Finding intermediate SYN chars is rare, but if they are sent and the terminal did not include them in the BCC calculation, then the incoming frame is seen as a CRC error. There is no workaround.
  If a T1 line is removed an reinserted into a VWIC-2MFT-T1 that is set up as a channel group, the VWIC module loses synchronization with the T1. Continuous framing and line code violations occur. There is a 50 percent packet loss on simple end-to-end pings. If the router is reloaded, the VWIC module works without errors until the T1 is removed and reinserted.
  Workaround: Reload or powercycle the router with the T1 line plugged in.
  Configuration parsing error (loss) happens after a reload of a router running UDLR tunnels. The error is related exclusively to the configuration lines establishing which is the send-only/receive-only interface of the UDL (Uni Directional Link) and happens ONLY when the interface connected to the UDL is configured to be a subinterface (e.g. when using Frame Relay).
  A Route Switch Processor 2 (RSP2) running Cisco IOS Release 12.1(3) experiences %ALIGN-3-SPURIOUS messages with tracebacks. There is no workaround.
  An outgoing V.110 call originated from a Cisco AS5800 that is used as a Network Access Solution (NAS) as part of a Signalling System 7 (SS7), has an incorrect encoded intermediate rate of 0x0 in the Low Layer Compatibility IE. The value is incorrect for the speeds 600, 1200, 2400, 4800, 7200, 9600, 12000, 14400 and 19200. There is no workaround.
  A router might reload when logging certain system error messages internally. There is no workaround.
  After rebooting, the ISDN Layer 1 of the second controller of the Non-Facility Associated Signalling (NFAS) group shows the "activating" state.
  Workaround: Remove the pri-group from controller configuration mode and then reinsert it.
  A Cisco 7507 router running Cisco IOS Release 12.1.2(T) that attempts to establish a tunnel with a Cisco router running Release 12.0(9) is not able to establish the tunnel and will receive an "SA is still budding" error message. Phase 1 of the negotiation is not completed. The 7507 router has an overlapping access-list. There is no workaround.
  A Cisco uBR 7200 series router running Cisco IOS Release 12.1(3) reloads daily with memory coruption after adding Baseline Privacy (BPI). There is no workaround.
  Spurious memory accesses seen on a Cisco 7200 NPE 200 with the PA-2FEISL and Cisco Express Forwarding (CEF) switching in operation.
  Workaround: Disable CEF on the router.
  On a Cisco 7200 series running Cisco IOS Release 12.1(3), spurious aligments were seen before the following error message:
  %ALIGN-1-FATAL: Illegal access to a low address addr=0x0, pc=0x601245E8, ra=0x60124420, sp=0x6127D930
  The router then reloads with this bus error:
  System was restarted by bus error at PC 0x601245E8, address 0x0 at 7200 Software (C7200-P-M), Version 12.1(3), RELEASE SOFTWARE (fc1) Image text-base: 0x60008950, data-base: 0x60E8A000
  Stack trace from system failure: FP: 0x6127D930, RA: 0x601245E8 FP: 0x6127D978, RA: 0x601249C0 FP: 0x6127D9A0, RA: 0x60125F54 FP: 0x6127D9C0, RA: 0x601288BC FP: 0x6127D9E8, RA: 0x80007E90
  There is no workaround.
  After bootup, the fair-queue aggregate-limit and fair-queue individual-limit interface configuration commands might be lost from the configuration. The result is that the queue-limit values will be set to the default values, rather than the values specified by the fair-queue commands.
  Workaround: Re-issue the commands after boot-up.
  After prolonged use in some situations, a Cisco 800 series router running Cisco IOS Release 12.1(3)T gradually runs out of memory.
  Workaround: Reload the router.
  If writing to a SanDisk (diskx:) file system is interrupted before reaching completion by either a reload or power cycle, future attempts to read or delete from that SanDisk will cause a router to reload.
  Workaround: Formatting the SanDisk returns it to a usable state. There is no way to recover data on the corrupted SanDisk.
  Under certain circumstances, the VWIC-2MFT-T1-DI module may experience line code and path code violations.
  Workaround: Use a WIC-1DSU-T1.
  When interface bandwidth is set to 0 on Cisco 7200 and 3600 series routers, various problems might occur. These problems include reloads when using Resource Reservation Protocol (RSVP) and Enhanced IGRP (EIGRP) not redistributing routes over the interface that believes it has a bandwidth of 0. There is no workaround.
  A sensitivity problem with the current line interface unit (LIU) used on the PA-A3-T3 port adapter might cause the interface to flap with or without any line error condition reported under the show controllers atm privileged EXEC command.
  Workaround: Install 75 ohm fixed attenuators on the Rx side of the PA-A3-T3 port adapter.
  When a router acts as a tag switching edge router, it will sometimes stop running Tag Distribution Protocol (TDP) after a reboot. This situation has only been observed on a Cisco 6400 NRP running Cisco IOS Release 12.1(1)DC1 when the tag neighbor(s) are connected through the NRP1 Fast Ethernet interface. The router will restart and run TDP if rebooted again. There is no workaround.
  When a router is using integrated routing and bridging (IRB) and has a BOOTP client in a Token Ring on the bridging side and a BOOTP server in another Token Ring of the same router but on the routing side, the BVI interface of the router sends the BOOTPREPLY to the station with the destination MAC address bit swapped. The destination station never gets the reply. There is no workaround.

Novell IPX, XNS, and Apollo Domain

  When the ipx route-cache interface configuration command is enabled, session drops are noted on the IPX hosts. This situation occurs in sessions crossing the TR VLANs to some file servers on the Ethernet VLANS.
  Workaround: Remove the ipx route-cache interface configuration command.

TCP/IP Host-Mode Services

  A Cisco 7513 router might receive SYN flood attacks when you turn on TCP intercept. CPU utilization reaches 100 percent on a Cisco 7513 router when you enable TCP intercept. The output of the show tech command indicates that 73 percent of the CPU is taken by interrupt driven processes, and approximately 25 percent by IP input. There is no workaround.
  A Cisco 3660 router running IOS 12.0(7)T or 12.1(3a) with "ip tcp intercept" configured, might on occasion reload by a bus error.

Wide-Area Networking

  On a Cisco 7200 and 3600 series router, bridging over Frame Relay on a BRI interface will fail when static mapping is configured for the permanent virtual circuit (PVC).
  Workaround: Use routing instead of bridging, or use a point-to-point subinterface.
  A multilink interface might stop processing received packets from the peer if the peer multilink interface went down and came back up. The show ppp multilink command will indicate "received lost fragments" when this caveat is encountered.
  Workaround: Clear the multilink interface.
  Link control protocol (LCP) might fail on some Cisco routers that are running Cisco IOS Release 12.0(4)T if the maximum transmission unit (MTU) on the physical interface is less than 1500.
  Workaround: Enable PPP encapsulation on the physical interface, and then set the MTU to be less than or equal to 1500.
  On a Cisco 3640 router running Cisco IOS Release 12.0(2a)T1 with MICA modems, Cisco dialout fails when a local number is called. The router returns the "not end to end isdn" call-blocked group restriction error. Any call to a long distance number succeeds. There is no workaround.
  A Cisco 800 series router will not pass incoming voice bearer cap calls to plain old telephone service (POTS) ports unless the isdn incoming-voice modem interface configuration command is entered on the BRI interface. There is no workaround.
  ISDN Layer 2 might not come up if you have enabled Link Access Procedure, Balanced (LAPB) encapsulation.
  Workaround: Enter the no fair-queue interface configuration command on the BRI interface.
  On the ISDN Primary Rate Interface (PRI) switchtype for a 4ESS, primary-4ess might not be fully compliant with AT&T products. There is no workaround.
  Cisco Layer 2 Tunneling Protocol (L2TP) will not process packets that do not include a checksum in the User Datagram Protocol (UDP) header. This condition will not occur with other Cisco implementations, but might with the systems of other vendors.
  Workaround: Enter the vpdn ip udp ignore checksum configuration command.
  Incoming calls are not being answered or acknowledged. You might receive a SETUP message followed by a DISCONNECT message, and the router only sends a response to the DISCONNECT message. This situation is rare. There is no workaround.
  Under certain conditions, the input queue count of a multilink bundle interface will increase until the limit is reached. The bundle interface will then discard any further packets received, and the bundle becomes unusable because it will not accept inbound traffic. There is no workaround.
  The input queue on a dialer interface might not work properly. There is no workaround.
  Under a load, a Cisco AS5300 series access server might reload with the following console debug:
  ALIGN-1-FATAL: Corrupted Program Counter pc=0x0 ra=0x601d51b0 sp=0x6
  or
  ALIGN-1-FATAL: Corrupted program counter pc=0x0, ra=0x6017E4A8, sp=0x61384B70
  There is no workaround.
  Packets might be lost when links congest when you use first-in, first-out (FIFO) queueing in any release of Cisco IOS software. Multilink PPP (MLP) might lose more streaming traffic than single-link PPP. The best way to minimize the effects of fragment loss and packet loss is to place weighted fair queueing (WFQ) on the MLP bundle. There is no workaround.
  The load on a bri0:0 interface (D channel) never increases and always remains at 1/255. There is no workaround.
  When PPP over ATM is configured on an ATM Interface Processor (AIP), the following error messages might appear and result in a Cbus reset:
  %RSP-3-IP_PANIC: Panic: ATM9/0 00000006 9FC0333C 000013F0 9FC0333C %RSP-3-IP_PANIC: Panic: ATM9/0 00000007 A000FDB0 10000014 A000FDB0 %RSP-3-IP_PANIC: Panic: ATM9/0 00000008 A000FD50 10000014 A000FD50 %CBUS-3-CMDTIMEOUT: Cmd timed out, CCB 0x5800FFB0, slot 9, cmd code 2 -Traceback= 6033A4F8 6033A9EC 60332E64 60330EE4 60298308 60298538 602B94BC 602B94A8
  The ATM interface will no longer come up until the AIP is physcally reinserted or the router is reloaded.
  Workaround: Use a PA-A3 port adapter.
  During scalability testing of the Annex G (X.25 over Frame Relay) feature, we have identified a problem with recovery from interface failures: with large numbers of Annex G circuits defined, loss of the interface may cause the router to reload. There is no known workaround, but the problem has only been observed with extreme configurations.
  With the Annex G (X.25 over Frame Relay) feature, the router might reload when unconfiguring Annex G on a Frame Relay circuit while the circuit is being actively used to forward data. This situation is most likely to occur in large Annex G configurations. The reload does not occur if the circuit is idle. There is no workaround.
  ISDN callback fails for the second call if two calls are coming in simultaneously, or the second call comes in within the configured enable timeout for the first call, and a certain vendor's callback is negotiated by the caller.
  The problem can be mitigated for isdn callers by configuring a very small enable-timeout value, say, 1 second. For non-isdn callers, there is no workaround (but configuring enable-timeout as small as possible will help). The enable-timeout value is used as a timer for initiating the callback.
  Please contact the Technical Assistance Center (TAC) for more information.
  A router terminating PPP sessions using STAC compression may reload in the LZS_Compress process.
  Workaround: Disable STAC compression.
  A router experiences a software-forced reload with a bus error every 5 to 10 minutes after enabling the broadcast trigger command on a map-list associated with an active Frame Relay switched virtual circuit (SVC).
  Workaround: Do not enter the class subcommand under the map-list global configuration command.
  A precloned virtual access might not renegotiate Internet Protocol Control Protocol (IPCP) when it is reused in a new call.
  Workaround: Do not use the preclone command.
  When a Cisco router is configured for an outbound X.25 permanent virtual connection (PVC) using protocol translation, the router ignores a received Reset, resulting in a PVC (once recreated) for which the two stations have a conflicting state. This situation is unrecoverable without a restart.
  Workaround: Initiate an X.25 interface restart.
  An X.25 connection over a D channel might fail with DECnet Phase IV. There is no workaround.
  The PA-A1 port adapter stops receiving traffic under stress on Bay 1 of a Versatile Interface Processor 4-80 (VIP4-80). A shut/no shut sequence recovers the condition temporarily.
  Workaround: UseVIP2-40 or VIP2-50.
  When RADIUS Dialed Number Identification Service (DNIS) preauthentication is used, calls might be rejected even when the screening works and there are enough resources. Debug resource will show RM_DNIS_RES_ALLOC_FAIL. There is no workaround.
  A Cisco AS5300 router might reload due to memory corruption. There is no workaround.
  Always on Dynamic ISDN (AODI) stops functioning if traffic traversing the D channel fails with any component along the path. There is no workaround.
  Layer 2 Forwarding (L2F) might reload the router if Virtual Private Dialup Network (VPDN) and PPP are unconfigured during the lifetime of an L2F call. There is no workaround.
  An X.25 ISDN bridging connection over a D channel might fail. There is no workaround.
  After a reload, a Cisco router will wait for a random period of time before activating Layer 2 and exchanging service profile identifier (SPID) information. If during this time a call is received, it will trigger Layer 2 and SPID exchange activation, but the call and any subsequent call might be ignored until the interface is reset. This problem occurs when the router is connected to the network interface (NI) switch of some vendors.
  Workaround: Enter the isdn twait-disable interface configuration command under the BRI0 interface to eliminate the vulnerable twait period.
  A Cisco router running Cisco IOS Release 12.1(1a) might experience a software-forced reload while making outgoing IDSN calls. There is no workaround.
  If you observe a modem that shows low-average connection times, lookat the int async input queue; it might be wedged.
  Workaround: Reload the router. Increase the input queue by entering the hold-queue 200 in interface configuration command.
   Alternate workaround: Remove IP header compression.
  A Cisco 7200 series router with an ATM interface might unexpectedly reload because of a corrupted program counter. There is no workaround.
  A Cisco 3640 router returns to ROM by bus error at PC 0x602DF148, address 0xEF4321D1. There is no workaround.
  The virtual access interface is formed by a combination of the parameters configured on the virtual template and the RADIUS attributes if the virtual-profile aaa and virtual-profile virtual-template global configuration commands are placed in the configuration.
  When the no peer default ip address interface configuration command is placed on the virtual template, the router and host fail PPP negotiation. The host does not receive an IP address.
  Workaround: Remove the virtual template from the configuration and reboot the router. Readd the virtual template and then remove the no peer default ip address command from the virtual-template.
  A Cisco 7500 series router running rsp-jsv-mz.121-1a experienced a software-forced reload because of a software-forced memory corruption.
  A decode of the stack reveals that ATM processes were involved:
  0x602E6434:abort(0x602e642c)+0x8 0x602E4B9C:crashdump(0x602e4b8c)+0x10 0x602CFF08:free(0x602cfc88)+0x280 0x6002151C:atmSig_freeSvcTosMemory(0x6002147c)+0xa0 0x60020FCC:atmSig_freeSvcP2PMemory(0x60020efc)+0xd0 0x60020ED8:atmSig_freeSvcMemory(0x60020ea8)+0x30 0x6001E4B8:atmsig_background_free(0x6001e48c)+0x2c 0x6001CF40:atmsig_proc_sig2DriverApi_msg(0x6001c9f8)+0x548 0x6001CFF4:atmsig_sigDriverApi_process(0x6001cf6c)+0x88 0x602C103C:r4k_process_dispatch(0x602c1028)+0x14 0x602C1028:r4k_process_dispatch(0x602c1028)+0x0
  In the crashinfo file log, the following text was recorded:
  %SYS-2-FREEFREE: Attempted to free unassigned memory at 62569700, alloc 603FDF18, dealloc 603FE128 -Traceback= 602CFED8 6002151C 60020FCC 60020ED8 6001E4B8 6001CF40 6001CFF4 602C103C 602C1028
  This traceback message decodes to:
  0x6002151C:atmSig_freeSvcTosMemory(0x6002147c)+0xa0 0x60020FCC:atmSig_freeSvcP2PMemory(0x60020efc)+0xd0 0x60020ED8:atmSig_freeSvcMemory(0x60020ea8)+0x30 0x6001E4B8:atmsig_background_free(0x6001e48c)+0x2c 0x6001CF40:atmsig_proc_sig2DriverApi_msg(0x6001c9f8)+0x548 0x6001CFF4:atmsig_sigDriverApi_process(0x6001cf6c)+0x88 0x602C103C:r4k_process_dispatch(0x602c1028)+0x14 0x602C1028:r4k_process_dispatch(0x602c1028)+0x0
  There is no workaround.
  Cisco IOS software fails in X.25 over TCP (XOT) negotiation when a virtual circuit (VC) is connected between the terminal of which Packet_size(PS)/Window_size(WS) is 256/5 and the terminal of which PS/WS is 128/2. There is no workaround.
  Authentication failure on the Cisco L2TP Network Server (LNS) reboots the ECI Telecom/Telematics Nevada L2TP Access Concentrator (LAC). There is no workaround.
  An incoming packet assembler/disassembler (PAD) call with a destination that matches the primary interface subaddress is cleared if the line has no rotary configured for the subaddress. In releases earlier than Cisco IOS Release 11.3, the call was switched instead of cleared. There is no workaround.
  The X.25 for Dialer interface feature is still unusable because X.25 packet assembler/disassembler (PAD) calls are correctly routed to the dialer interface and bound to the BRI, but the dialer idle-timeout interface configuration command is not reset by the flow of X.25 data. Any traffic defined by a dialer list, however, does reset the dialer idle timer. X.25 PAD data is not definable by a dialer list. There is no workaround.
  With a Logical Link Control, type 2 (LLC2) session between a Cisco router and a Tandem ICE, Frame rejects (FRMRs) might occur if there is a delay in a POLL/FINAL sequence. This situation results from a slight variance between the IEEE 802.2 1985 LLC2 state machine and the IBM Token Ring LLC2 state machine.
  Workaround: Set the llc2 tpf-time {milliseconds} interface configuration command to a value larger than the round-trip delay of the POLL/FINAL sequence.
  The Always on Dynamic ISDN (AODI) connection works only for the client configured in the first X.25 PPP map that is configured under the X.25 interface. There is no workaround.
  A router using multilink with configuration profiles (dialer profiles, virtual templates, and so on) might encounter a bus error with "ppp_notify_cb_configured" in the call stack. Failure(delayed authorization response) occurs when a significant delay in multilink link up processing is encountered and there are non-default multilink configuration values ("no ppp multilink fragmentation") present in the profile. There is no workaround.
  Incoming ISDN calls are no longer working on a Cisco router that uses ISDN switch type VN3. There is no workaround.
  When changing from a TS014 switch type to another Primary Switch type, channels that were excluded through the pri-group configuration might become active. TS014 switch numbers the channels 1-30, without reserving a timeslot for the D channel. The show isdn service privileged EXEC command displays the D channel as channel 31. (To eliminate confusion, the show isdn service command displays channel 31 as "D", for the TS014 switch).
  Workaround: Use another switch-type in the global isdn switch-type (for example, primary-ni). Then, speciying the TS014 switch-type on the relevant serialx:15 configuration(s) will prevent the problem.
  Using Dynamic Host Configuration Protocol (DHCP) as the global default pool mechanism allocates duplicate addresses in Cisco IOS Release 12.0T and Release 12.1. There is no workaround.
  When configuring Async dial-on-demand routing (DDR), the dialer idle-timeout is ignored.
  Workaround: Add an absolute-timeout under the line.
  When the Cisco SC2200 Signaling Controller "switch" connected to a router is configured as "passive," it will not send set asynchronous balanced mode extended (SABME) sequences to initiate link recovery after a link failure. The router perodically does not send SABMEs to initiate recovery, resulting in a "deadlock" where both the SC2200 and the router are waiting for the other to initiate recovery.
  Workaround: Change the configuration on the "switch," or reload the router to recover the link.
  A Cisco AS5300 access server has 1 PRI connected to a public switch with "isdn switch-type primary-net5." V.110 calls arriving on that interface fail when the bearer capability does not indicate a V.110 call. The AS5300 is trying a High-Level Data Link Control (HDLC) call and fails. This situation is common with many ISDN adapters.
  The second problem occurs when a call comes in over a Cisco VSC3000 Virtual Switch Controller. Redundant Link Manager (RLM) is used and switch-type is set to "primary-ni" to make it work with RLM. The call from the ISDN adapter fails in the same manner as above, and a call from a GSM phone fails, too.
  The Bearer Capability (BC) flags cause the failed call. The flags are 6 bytes long and indicate a correct V.110 call. However, if switch-type is set to "primary-ni," an Invalid Element message appears because the BC is longer than 4 bytes. This condition might be a check in the Cisco IOS release that only applies to switch-type primary-ni and not to primary-net5. There is no workaround.
  For LSDO, the default gateway wins the Stack Group Bidding Protocol (SGBP) bid even if the PRI or BRI controller is down. There is no workaround.
  On a Cisco 3600 series router running Cisco IOS Release 12.1(3) and 12.1(3a), a High-Level Data Link Control (HDLC)-configured serial interface does not become "line protocol down" when it does not receive keepalives. There is no workaround.
  IP traffic stops when you have a second B-channel up while running Cisco IOS Release 12.1.
  Workaround: Clear the dialer interface.
  On a Cisco 7500 series router that is running Multiprotocol Label Switching (MPLS), distributed CEF, and Packet Over SONET (POS) under normal traffic load, the POS PA goes into the output-stuck, output-frozen, and cbus complex restart states. All interfaces are then reset, resulting in a loss of established routing protocol neighborships, as well as other established traffic flows.
  Workaround: Disable dCEF globally on the 7500 series router.

Resolved Caveats—Release 12.1(4)

All the caveats listed in this section are resolved in Release 12.1(4). This section describes severity 1 and 2 caveats and select severity 3 caveats.

Basic System Services

  A router might experience high CPU utilization at the interrupt level with flow switching because of spurious accesses by flow switching code.
  Workaround: Turn flow switching off, using the no ip route-cache flow interface configuration command.
  A Cisco Route Switch Processor (RSP) might periodically display the following traceback:
  IPC-5-INVALID: Sequence Structure port index=0x0 appears on the console.
  When you enable the debug ipc errors EXEC command, the RSP might display the following traceback:
  IPC: SEQ_ERR ACK ... source seat 0x1000000 port 0x0
  If you enter the show ipc stat command, the RSP produces "messages dropped on input" and "no local port" errors.
  This situation might occur when distributed Cisco Express Forwarding (dCEF) is enabled on line cards. There is no workaround.
  If you use hardware compression and fancy queueing with two simultaneous traffic streams (for example, FTP and Telnet), packets are not compressed. Entering the show compress EXEC command will not provide any useful information. There is no workaround.
  After being operational for a week, calls can no longer be processed by a Cisco MC3810 multiservice concentrator. "Fatal Error" messages appear on the console, and the router must be reloaded to function correctly. There is no workaround.
  A Cisco AS5300 access server reloads with a bus error at PC 0x6022A3C8 when you try to Telnet to a named-host. There is no workaround.
  High capacity counters in the IF-MIB ifXTable display 0 for all interfaces not supporting them. There is no workaround.
  A Cisco 5300 series router that is using Multiservice IOS Channel Aggregation (MICA) modems might experience high CPU utilization with Cisco IOS Release 12.1(1a)T1, because of alignment and spurious errors. This situation occurs primarily when Multilink PPP and fast-flow switching is enabled on an interface (but might also occur when fast switching is enabled).
  Workaround: Disable flow switching or Multilink PPP.
  A vty password configured through the CLI interface available via http or html works correctly. If this same interface is used to save the configuration (as opposed to the console or Telnet), the vty password is not nvgened. A subsequent reload of the device will not have the vty password.
  Workaround: Reapply the vty pasword through the http or html interface (or any other method) and then save the configuration through a Telnet session to the device. This will cause the vty password to be nvgened properly.
  You can determine if your vty password will be nvgened properly by using a show run command through the same interface with which you will be saving the configuration. If the configured vty password does not appear in the output of the show run command, then Telnet to the device to save the configuration.
  A Cisco MC3810-V3 multiservice concentrator running Cisco IOS Release 12.0(7)XK1 or later with a X.21 serial cable might have trouble bringing up a Frame Relay circuit because of timing problems. The serial interface is unstable, causing the Local Management Interface (LMI) protocol to flap.
  Workaround: Use other types of serial cables, such as RS232, V.35, or RS449.
  When multiple Cisco MC3810 multiservice concentrators are connected through a T1 / Channel Associated Signaling (CAS) to a different PBX, the MC3810 claims that an extra digit (a "1") is leading the digits coming from the PBX. This new destination number does not match any dial-peers in the configuration, so calls are not established. When you place a T1 analyzer in between the router and the PBX, the PBX claims there is no leading digit of "1" being sent. To clear this condition, reload the router. This problem affects Cisco IOS Release 12.1(2).
  Workaround: Increase the time during which you ignore any digits received after you detect off-hook to 220 msecs.
  A Cisco MC3810 multiservice access concentrator running Cisco IOS Release 12.1(2.1) reloads while disconnecting a call. There is no workaround.
  When a Cisco MC3810 multiservice concentrator makes a call out to be tandem-switched through a Cisco 7206VXR router, the MC3810 does not roll over to the next 7206VXR if the first 7206VXR it hits has no resources available. Instead, it drops the call. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0 might unexpectedly reload while it is performing Simple Network Management Protocol (SNMP) functions related to routing table information (specifically ipRouteNextHop). The reload occurs only in cases where an IP route configuration of the form "ip route <network address> <netmask> 0.0.0.0" exists. There is no workaround.
  When Response Time Reporter (RTR) Dynamic Host Configuration Protocol (DHCP) operation is used with certain DHCP servers, all DHCP IP address leases may be used up on the server.
  Workaround: Change the frequency of the operation from the default 60 seconds to every 5 minutes. Also, you should avoid configuring many DHCP operations on the same subnet. Both of these changes reduce your chance of the problem, because the situation depends on how the DHCP server is configured. If the DHCP server has long lease times, then the problem is more likely to happen.
  RADIUS tunnel attribute Tunnel-Assignment-ID is misinterpreted for authentication purposes.
  Workaround: Do not use Tunnel-Assignment-ID.
  After a router reload in Cisco IOS Releases 12.0(7)T, 12.1(1)E, and 12.1(2), the router will automatically insert Simple Network Management Protocol (SNMP) Read-Only communities into the running configuration even though these communities were explicitly removed from the startup configuration.
  In Cisco IOS Releases 12.0(7)T, 12.1(1)E, and 12.1(2), the router automatically creates Read-Only community entries based on the SNMP community string specified in the snmp-server host configuration when snmp-server host global configuration commands are issued. This feature supports SNMPv2 informs messages, which require an SNMP management station to acknowledge messages sent to it by the SNMP agent (router) using the same community string that was sent by the agent in the informs message.
  When you use the "public" community string in snmp-server host configurations, the router will automatically create an snmp-server read-only "public" community. When you manually remove the snmp-server "public" community, the router will recreate it after the next reload. The effect is that you might not be aware that the router has "public" SNMP read-access. There is no workaround.
  Environmental alarms are triggered spontaneously due to wrong hardware revisions. Only AC power supply alarms are affected. There is no workaround.
  Any product that uses an RM7000 CPU—including the NPE-300 network processing engine, a Cisco 7140 router, a Cisco RSP8, and a VIP4—might cause the router to execute instructions incorrectly or not at all. This situation might result in memory corruption or reload. There is no workaround.
  The Cisco MC3810 multiservice concentrator does not play a progress tone when a call is routed over the ATM or Frame Relay network for a number that cannot be terminated. The user does not hear anything excpet dead air.
  This is more prevalent when destination-patterns such as ".T" are used since they match any dial-string.
  Workaround: Use only dial-peers and destination-patterns for valid numbers, as a reorder tone is played if a matching dial-peer cannot be found locally.
  When an online insertion and removal (OIR) is performed, the IF-MIB ifTable, ifStackTable, and ifNumber retains the old entries even though the associated interface layers have been removed. There is no workaround.
  CSCdr86700 breaks compilation in a Cisco 1700 series router.
   Although "if_ipm_isr_pquicc.c" calls to "reg_invoke_hwcomp_compress," it does not support hardware compression so changes for CSCdm91180 do not apply there. There is no workaround.
  Depending on the platform, if you use the operation ID of 2147483647 for Service Assurance Agent (SAA) operation and use the show running-config, show rtr operational-state, or show rtr collection-statistics EXEC commands, the platform could reload or pause indefinitely.
  Workaround: Do not use 2147483647 for an SAA operation.

DECnet

  DECnet ping packets might be dropped when using X.25 encapsulation over an ISDN BRI D channel. This situation only occurs when DECnet is sending large routing updates. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(8) might experience an unexpected system reload when you use the clear decnet accounting EXEC command or the no decnet accounting interface configuration command.
  This situation will only occur if DECnet accounting has been enabled on at least one interface and DECnet traffic is being forwarded by the router at that point in time. This situation is more likely to occur if the following error message has been logged:
  %SCHED-2-EDISMSCRIT:Critical/high priority process DECnet Input may not dismiss. -Process= "DECnet Input", ipl= 0, pid= 74
  Workaround: Do not enable DECnet accounting on any interface.
  A Cisco router forwarding DECnet packets, that is configured with a DECnet interface static route, might occasionally reload unexpectedly.
  An unexpected reload occurs when a DECnet packet is forwarded to the DECnet node number specified in the DECnet interface static route.
  A DECnet interface static route is configured by the decnet route decnet-address interface-type interface-number global configuration command.
  Workaround: Replace the interface static route with a specific static route that has a next-hop address rather than an interface.

IBM Connectivity

  A Cisco router running data-link switching (DLSw) with TCP encapsulation might restart with the following error message when the TCP connectino to the peer router is congested:
  System restarted by bus error at PC 0x60735548, address 0xD0D0D21
  Workaround: Avoid congested TCP connections to the DLSw peer router.
  Receiving Intermediate Transmission Block (ITB) characters in bisync on some Cisco routers will cause memory corruption and cause a software-forced reload. There is no workaround.
  A Cisco router might reload if there are a large number of Native Client Interface Architecture (NCIA) clients with heavy NCIA traffic because of a watchdog timer. There is no workaround.
  A router pauses indefinitely when performing the no encapsulation sdlc-primary interface configuration command. There is no workaround.
  A Cisco 2500 router or a Cisco 4000 series router that is running LAN Network Manager (LNM) with data-link switching (DLSw) or Remote Source Route Bridging (RSRB) enabled might reload when the LNM tries to remove a port adapter.
  Workaround: Disable LAN Manager in the interfaces and bridging configuration.
  When remote source route bridging (RSRB) is configured on two Cisco 4500 series routers that are connected by an ATM network, and RSRB is configured with an encapsulation type of Fast Sequenced Transport (FST), some frames might not be forwarded. This situation might result in a data-link connection (DLC) session not being established through the RSRB peers.
  Workaround: Either change the RSRB encapsulation from FST to TCP, or configure data-link switching (DLSw) as the transport.
  Frame Relay Access Support (FRAS) boundary access node (BAN) works over the primary connection and switches to the backup connection. However, when it switches to the primary connection from the backup connection, it does not work. There is no workaround.
  When the dlsw explorer-wait-time and lf 1500 commands are configured, it requires two X.25 call setups to bring up a data-link switching (DLSw) circuit.
  The DLSw circuit connects with one X.25 call after one of these two commands is configured.
  The debug shows that the message "CSM: Reach index error CSM_NO_REACH_ENTRY" is displayed with the first X.25 call. The second call completes circuit setup, using the debug dlsw reach verbose sna and debug x25 all commands.
  There is no workaround if the two commands are required.
  After making changes to the dlsw local-peer global configuration command (adding or removing init/max pacing values), a router reloads when a write memory command is issued. The router configuration includes data-link switching (DLSw) directly in Frame Relay with backup peers defined.
  Workaround: Configure the dlsw disable global configuration command before issuing the write memory command.
  When locally administered address (LAA) addresses are used on the Ethernet interfaces of the data-link switching (DLSw) routers running the DLSw Ethernet redundancy feature, some test frames might leak. There is no workaround.
  A Cisco 7206VXR router that is running Cisco IOS Release 12.1(1a)T1 might display several MALLOCFAIL messages and reload with a bus error because of an ALIGN-1-FATAL error. The router will also not generate a core dump. There is no workaround.
  A Cisco router running Cisco IOS Release 12.0 with data-link switching (DLSw) configured might reload if the DLSw peers are misconfigured in a way that one DLSw peer is configured for priority and the other end is not. Also, when the router tries to bring up the priority peer, the DLSw peer with no priority might reload.
  Workaround: Fix the configuration so that the DLSw priority peers are properly configured.

Interfaces and Bridging

  Enabling Weighted Fair Queuing (WFQ) on an interface that belongs to a (transparent bridging) bridge-group might cause packets that are egressing that interface to be sent out-of-order. This situation causes failure in terminated and bridged Logical Link Control 2 (LLC2) sessions.
  Workaround: Disable WFQ with the no fair-queue interface configuration command.
  A Cisco PA-POS-OC3 Packet-Over-SONET port adapter does not automatically switch to line clock after clearing Loss of Signal (LOS), Loss of Frame (LOF), or Line Alarm Indication Signal (LAIS) defects that cause the clock to automatically switch to internal.
  A Cisco PA-POS-OC3 port adapter that detects the LOS, LOF, LAIS, and Path Alarm Indication Signal (PAIS) alarms might report additional nonexistent alarms. When LOS is present, the port adapter might also detect LOF, LAIS, path remote defect indication (PRDI), or B1-TCA threshold crossing alert alarms. When LOF is present, the port adapter might also detect LAIS, or PRDI alarms. When LAIS is present, the port adapter might also detect PAIS or PRDI alarms. When PAIS is present, the port adapter might also detect PRDI alarms. There is no workaround.
  With Constrained Multicast Flooding (CMF) configured, a Cisco router might reload while displaying the output of the show bridge multicast EXEC command. This situation is more likely to happen as the number of multicast groups being displayed increases. There is no workaround.
  A single Packet-over-SONET (POS) port adapter in a Cisco 7500/RSP series router with a Versatile Interface Processor 4 (VIP4) might stop transmitting and cause an "output stuck" condition. A POS port adapter that is a coresident with another port adapter in a Versatile Interface Processor 2 (VIP2) might also cause an "output stuck" condition for itself or the coresident port adapter. There is no workaround, but disabling distributed Cisco Express Forwarding (dCEF) globally, or in some conditions on the POS interface, will stop this condition from appearing.
  Clients cannot ping the Bridge-Group Virtual Interface (BVI) of a remote router when using Multilink PPP to connect two serial interfaces. There is no workaround.

IP Routing Protocols

  A Cisco 4700 series router that is running Cisco IOS Release 11.2(16)P might experience a situation where the E0 interface shows "INTERFACE UP, LINE PROTOCOL DOWN," but the data-link switching (DLSW) peers are still shown as connected.
  Workaround: Shut down the interface.
  A Cisco Catalyst 6000 family switch with a Multilayer Switch Feature Card (MSFC) might reload when the show ip pim neighbor EXEC command is issued. There is no workaround.
  The BGP Site of Origin (SOO) attribute is not added to some prefixes.
  Workaround: Perform a Border Gateway Protocol (BGP) soft reset, using the clear ip bgp soft EXEC command (if soft-reconfiguration is configured).
  A Cisco nondesignated router on a multiaccess interface creates state for (S, G) with the Olist populated when receiving a multicast packet. This situation causes a periodic assert every three minutes. There is no workaround.
  A Cisco 3640 router with Cisco IOS Release 12.1(2) reloads with the system returned to ROM by bus error at PC 0x601F9A28, address 0xFFFFFFF9.
  Workaround: Turn off default IP fast-switching if it is on.
  Cisco IOS Network Address Translation (NAT) might stop working if an overload mapping is configured and an extensive File Transfer Protocol (FTP) data session is generated across the NAT router.
  Workaround: Reload the router.
  The default originate command is deprecated for the VPNv4 address family.
  Workaround: Do not use this command under address-family vpnv4 mode.
  If you remove the last redistribute router configuration command from Open Shortest Path First (OSPF) that used type-1 metric, other routers might not remove the routes from the routing table even though the OSPF database entries are successfully removed.
  Workaround: Keep the router from which you removed the redistribute router configuration command as an autonomous system boundary router (ASBR) by removing one redistribution completely and keeping another dummy redistribution.
  Alternate workaround: Do not use the no redistribute router configuration command. Flush all the link-state advertisements (LSAs) by using the redistribute [route-map foo] router configuration command. Under route-map foo, disallow all routes. Enter the clear ip ospf redist command. Now you can safely remove the redistribute router configuration command completely.
  The hash value used for rendezvous point (RP) selection process in a bootstrap router (BSR) is calculated with a modulo of 2 to the power of 32. This modulo might cause inconsistency in the selected RP address between Cisco and other vendors implementing BSR.
  Workaround: Follow these rules to avoid inconsistency between Cisco routers when upgrading Cisco IOS software:
  If you plan to upgrade routers in your network that use BSR, avoid running Cisco IOS Release 12.0(10)S, Release 12.0(11)S, Release 12.0(11), or Release 12.1(2).
  When upgrading your routers, always upgrade the candidate BSR routers first to a release later than Cisco IOS Release 12.0(10)S, Release 12.0(11)S, Release 12.0(11), or Release 12.1(2). You do not need to upgrade the rest of you network immediately unless you are running the images in CSCdp95116 that might cause the routers to choose the RP on the basis of an incorrect hash value.
  When using distributed policy routing, the set interface route-map configuration command on subinterfaces does not work. The set interface route-map configuration command does not work on an interface that is on a different line card either.
  Workaround: Use the set ip next-hop route-map configuration command.
  A Cisco router that is running Open Shortest Path First (OSPF) on Cisco IOS Release 12.0(11) does not populate the routing table properly. When redundant paths exist, half of the paths are missing from the routing table. However, the missing networks appear in the OSPF database.
  Workaround: Run a full shortest path first (SPF) by doing clear ip route * EXEC command.
  Network Address Translation (NAT) might be corrupting NetBIOS packets if they have the source port "TCP 139" and the payload starts with 0x84 (SESSION TARGET RESPONSE packet type, cf.RFC 1002) but they are not SESSION TARGET RESPONSE packets. There is no workaround.
  An Address Resolution Protocol (ARP) alias configuration is lost after a Cisco router reloads or after you use the no shut command on an interface.
  Workaround: Explicitly reconfigure the arp-alias entries for the IP addresses of the interface after using the no shut command.
  Static routes that are within the range of a network statement where the gateway is an interface will not be distributed into OSPF by the redistribute static command.
  Workaround: Do not cover static to the interface by the network statement under OSPF.

Miscellaneous

  A router might reload when bad packets, whose datagram size is smaller than the IP total length carried in the ip header, are received by the router and also classified for encryption. This situation affects Cisco Encryption Technology (CET) only, and does not affect IP Security (IPSec).
  Workaround: Upgrade to Cisco IOS Release 12.1(4).
  If you run a hub-and-spoke Frame Relay configuration, and the hub router is set to be a multipoint interface, Dynamic Host Configuration Protocol (DHCP) requests will fail.
  Workaround: Configure both the hub and the spoke to use point-to-point subinterfaces. Alternatively, configure the DHCP server address on the spoke router instead of specifying the network address on which the DHCP server resides. This DHCP server address turns into a unicast address instead of a directed broadcast.
  The clock rate interface configuration command does not support 256 kbps or 768 kbps on a single port serial WAN Interface Card (WIC-1T/2T) for a Cisco 2600 series router or a 3600 series router.
  Workaround: Upgrade to the image that contains the fix for this problem.
  A router reloads after the show queue privileged EXEC command is issued multiple times when IP packets are fragmented in the fast switching path in a congested interface for a particle system and the ip cef global configuration command is disabled.
  Workaround: Enable the ip cef global configuration command.
  Alternate workaround: Disable fragmentation by not configuring an MTU size or by using a large enough value to avoid fragmentation at the queuing permanent virtual circuit (PVC).
  In a network where there are multiple paths between two networks and the paths travel through different sets of routers between these networks, per-destination load balancing will not be effective in any router after the first router where the load balancing paths diverge.
  Workaround: Use per-packet load balancing.
  A Cisco 7200 series router that is configured with a T1 multichannel T1 card might not show any errors while a remote Cisco 7000 series router shows cyclic redundancy check (CRC) errors, input frame errors, or overrun errors. Errors occur but the counter remains at zero. There is no workaround.
  In some circumstances, the adjacency associated with a static Address Resolution Protocol (ARP) entry might be deleted if the interface is shutdown. There is no workaround.
  A Cisco 7500 series Route Switch Processor (RSP) that is running Cisco IOS image rsp-jsv-mz.112-18 might reload with a segV exception error while performing encryption processing. There is no workaround.
  Voice calls might fail on dial-peers matching the Non-Facility Associated Signaling (NFAS) trunks of a Cisco AS5300 access server running with a mix of Regular PRI and an NFAS configuration. There is no workaround.
  There is an unexpected disconnect of Layer 2 on a Cisco 800 series router that shares an S-bus with four or more devices only.
  Q.921 debugging points to the ID being removed by the switch when other devices are sharing the S-bus.
  When multiple devices on the S-bus try to reply to an IDCKRQ message from the switch by doing an IDCKRSP, there are D-channel collisions. There was a problem in the D-channel collision-handling mechanism in the driver software because it resides in software in the Cisco 800 series router using the mc145574 Motorola ISDN S/T transceiver. There is no workaround.
  Macs cannot authenticate to AppleTalk servers across WAN links when compression is enabled on a Cisco 3660 router running Cisco IOS Release 12.0(5)XK1. There is no workaround.
  A Cisco 1720 router experiences a software-forced reload when you load the c1700-bno3r2sy56i-mz, c1700-bno3r2sv3y56i-mz, c1700-bk2no3r2sv3y-mz, and c1700-bk2no3r2sy-mz images. There is no workaround.
  Packets coming from a Cisco TN3270 Telnet server onto ATM are corrupted. If a number of packets arrive, the leading packet is overwritten by the contents of the following packets. There is no workaround.
  A Cisco 7200 series router might reload because of memory corruption and exhibit the following error message:
  %SYS-2-MALLOCFAIL: Memory allocation of 520 bytes failed fr om 0x60395028, pool Processor, alignment 0 -Process= "Pool Manager", ipl= 4, pid= 4 -Traceback= 603C84DC 603C9F88 60395030 603D4578 603C2C34 603C2C20
  There is no workaround.
  Ciscoview 4.2 might not work properly on a Cisco 7513 router that is running Cisco IOS Release 12.0(9). This situation occurs when slots 0, 2, 5, and 9 are populated with VIP2 Versatile Interface Processors and PA-8T-V35 port adapter cards. CiscoView 4.2 might not show the PA-8T-V35 port adapter on slot 9. Other slots might also report incorrect information. This situation is caused by Simple Network Management Protocol (SNMP) agent, not CiscoView 4.2. The output of the snmpwalk command shows the wrong cardType with a note indicating that indices 19 and 24 are "unknown." There is no workaround.
  When a PA-A3 port adapter is used on a Cisco 7200 series router, input drops might occur if slow output interfaces are not returning buffers quickly enough. This problem is more visible when the NPE-150 is used in a 7200 router that has a high number of slow output interfaces.
  Workaround: Limit the number of buffers that can be held by the egress interfaces by using First In, First Out (FIFO) queueing with limited queue sizes. Refer to the following example, using the interface serial global configuration command:
  interface Serial1/0 no fair-queue hold-queue {x} out
  This problem has been seen with FastEthernet and FDDI interfaces when network connectivity is disrupted, or when an HSRP router with higher priority is added to a network.
  When the HSRP state changes from Active to Speaking, a link-state change occurs. The switch detects this link state change on the interface and a Spanning Tree protocol transition takes place. The Spanning Tree takes 30 seconds (twice the default Forward Delay time of 15 seconds) to transition the port into the Forwarding state.
  At the same time the Speaking router will transition to the Standby state after 10 seconds (HSRP Holdtime). The Spanning Tree is not Forwarding yet, so no HSRP Hello messages will be received from the Active router. This causes the the Standby router to become Active after a further 10 seconds.
  Both routers are now Active. When the Spanning Tree ports become Forwarding the lower priority router will change from Active to Speaking and the whole process will repeat.
  It has also been seen on interfaces where a host on the network was causing an IPX broadcast storm. This causes HSRP to loose Hello packets, which then caused the HSRP state to flap. The symptom of this was a duplicate HSRP packets.
  Workaround: Perform one of the following actions: 1. Ensure that there are no packet storms on the network. 2. Change the duplex setting of the switch; set it to "auto." 3. Configure "set spantree portfast enable" on the switchports. (See http://www.cisco.com/warp/public/784/packet/oct99/cat5000.html Troubleshooting the Catalyst 5000) 4. Configure "standby usebia."
  Snmpwalk on a Cisco 7500/RSP series router with an ATM interface might cause a CPUHOG situation and affect router performance. There is no workaround.
  You cannot create variable bit rate non-real time (VBR-NRT) switched virtual circuits (SVCs) if Resource Reservation Protocol (RSVP) is enabled. Requests for SVC creation will fail for VBR-NRT class of service. There is no workaround.
  Under a large Virtual Private Network (VPN) configuration (and possibly other configurations), the Versatile Interface Processor (VIP) might spend an unusual amount of time processing interprocess communication (IPC) messages. If this amount of time becomes especially long, a FIB_DISABLE might occur on the Route Processor (RP). There is no workaround.
  In a large Virtual Private Network (VPN) configuration, a Versatile Interface Processor (VIP) might spend several minutes processing interprocess communication (IPC) messages causing a FIB_DISABLE on the Route Processor (RP). There is no workaround.
  Cisco Express Forwarding (CEF) might not process an interface up event, so that the show interface command might show the interface up while the show cef interface EXEC command has it down. This situation can also result in missing prefixes in the CEF table.
  Workaround: Repeat the no shutdown interface configuration command on the interface. It is not neccesary to shut down the interface first (with the shutdown command).
  If input MAC accounting is configured on an interface, misaligned read accesses might be encountered on the Cisco 7500 series route switch processor (RSP). The show alignment command should be entered to determine if misaligned accesses are occurring.
  Workaround: Disable input MAC accounting on the interface.
  A digit string might only be recognized if the user dials 0 followed by the string. Any digits before the 0 are not recognized.
  Workaround: Enter the pots rx-loss -6dB configuration command.
  Interface counters will give incorrect values for tunnels on a serial interface when Cisco Express Forwarding (CEF) and IP Security (IPSec) are in use. There is no workaround.
  A memory leak might occur on a Route Switch Processor (RSP) when it is used with Versatile Interface Processors (VIPs) while running Open Shortest Path First (OSPF). The problem results due to repetitive reloading and downloading of VIP linecards that are disabling Cisco Express Forwarding (CEF) due to a lack of memory.
  Use the show processes memory [inc OSPF R] EXEC command and the show memory summary [inc OSPF R] EXEC command on the RSP to determine if increasingly large amounts of memory are being held by the OSPF process. Use the show cef linecard EXEC command to determine the number of reloads that a VIP has encountered.
  Workaround: Ensure that the VIPs have sufficient memory for their configuration and environment, such that CEF is not disabled on the VIP.
  A router might reload with crypto enabled on a Fiber Distributed Data Interface (FDDI) interface processor (IP). There is no workaround.
  When you configure an IP address on any intf or subinterface on a VIP2 Versatile Interface Processor using a Fast Ethernet PA-2FEISL-TX port adapter, the IP network configured on the subinterface does not show up as a directly connected interface in the IP routing table. As a result, there is no IP connectivity across the network.
  Workaround: Configure a static interface route for the directly connected network.
  When Cisco Express Forwarding (CEF) switching and Virtual Access interfaces are used, a Cisco 7200VXR router experiences serious problems with many alignment errors causing the CPU to reach 100 percent utilization. With an increasing number of users (virtual access interfaces), a production router will cease functioning instantly. There is no workaround.
  With Cisco IOS image c5300-js-mz.dp77759_del_5, the R2-R2 hairpinning calls are fine except that some of them stay connected, even when the called party and the caller hang up the phone, because the gateway does not recognize that the calls have ended and the circuit remains connected.
  Workaround: Reload the router.
  If you run Cisco IOS Release 12.0.7 T, a router might get in the E_DSP_ALARM_IND in the SETUP_REQ_FAIL state. This error occurs only occasionally.
  Workaround: Perform a digital signal processor (DSP) reset or reload.
  Cisco routers configured to use Stack Group Bidding Protocol (SGBP) might experience a buffer leak in large buffers.
  The memory leak might occur if a fairly large number of Multilink PPP (MLP) bundles are terminated on the SGBP member at the time the member enters or reenters the stack group.
  Evidence of the leak can be determined by monitoring the SGBP connection hello messages. These messages are enabled with the debug sgbp hellos command. If you see the following message and the size value is greater than 1360, a memory buffer will be leaked:
  SGBP:Send Info, count 1 size 7
  There is no workaround. Reload the router to recover I/O memory.
  When multiple, load-shared paths exist between provider edge (PE) routers, a PE router might reload if all paths are lost simultaneously while Virtual Private Network (VPN) traffic is being forwarded. There is no workaround.
  Any multiflex trunk module (MFT) T1/E1 card, when plugged in alongside a BRI WAN interface card (WIC) in either a Cisco 2600 WIC slot or a Multicast Address Resolution Server (MARS) series router combo card (1e1r2w,1e2w,2e2w), does not operate if the BRI is not connected to the network interface provided by the ISP.
  This situation only occurs when BRIs and any type of MFT WIC card are installed together in either a Cisco 2600 router's built-in WIC slots or a MARS combo card.
  Workaround: Connect the BRI to its WAN network interface at all times. If this cannot be done, you must remove the BRI from its WIC slot to get the MFT to work.
  Open Shortest Path First (OSPF) updates might be corrupted on a Cisco 7500 series router using Multiprotocol Label Switching (MPLS) switching with Cisco Express Forwarding (CEF) output features enabled (including "service policy output"). IP routes are temporarily deleted from the IP routing table and a loss of connectivity might occur.
  Workaround: Configure the ip cef global configuration command. Then, execute the copy running start command, and reload.
  Alternate workaround: Enter the memory cache-policy io uncached command. However, entering this command might sacrifice packet switching performance.
  Windows 2000 configured to encrypt Layer 2 Tunneling Protocol (L2TP) traffic with IP Security (IPSec) reloads Cisco IOS software.
  This situation occurs only when the Windows 2000 user initiates IPSec with the default IPSec settings.
  Workaround: Use IPSec settings with only the transforms you need. By default, Windows 2000 proposes a large number of proposals with all possible combinations of esp and ah transforms. This large number of transforms is causing a buffer overflow and thus memory corruption that causes a reload.
  The Five minute output rate counters do not match the Five minute input rate of the directly connected serial interface. There is no workaround.
  When the ip unnumbered interface configuration command is configured, an ATM subinterface does not respond correctly with Inverse Address Resolution Protocol (InARP) requests. There is no workaround.
  A Cisco 1605-R router might experience a bus error restart when it is booted with the Ethernet interfaces connected and is running Cisco Express Forwarding (CEF) and Distance Vector Multicast Routing Protocol (DVMRP). Replacing the hardware does not solve the problem.
  Enabling Protocol Independent Multicast (PIM) and the ip verify unicast reverse-path command on the same interface might cause the bus error at boot.
  Workaround: Disable either PIM or the Reverse-Path Forwarding (RPF) check.
  An NM-2E2W, NM-1E2W, NM-1E1R2W network module with a BRI WAN interface card (WIC) and a WIC-T1CSU on a Cisco 3600 series router might function incorrectly on rare occasions. When this problem happens, ISDN layer 1 on the BRI WIC is persistently deactivated and the WIC-T1CSU continues to work corrrectly (if no event requires the integrated CSU/DSU to signal the main IOS, in which case it ceases to pass date) except that the show service-module serial privileged EXEC command returns an error message.
  Workaround: Power cycle the router.
  Alternate workaround: Reload software.
  The cookie value in the chassis of a Cisco AS5800 access server with integrated modems is programmed wrong. This error causes instability in environmental monitoring. There is no workaround.
  You cannot terminate a Cisco virtual private network (VPN) client that requires Mode Configuration (Mode Config)/Extended Authentication (Xauth), and a Windows 2000 IP Security (IPSec) client on the same interface, which has a crypto map applied to it, to do Mode Config/Xauth.
  Workaround: Use two separate interfaces with two separate crypto maps. The Windows 2000 client does not implement Mode Config/Xauth; rather it attempts to achieve the remote access through Layer 2 Tunneling Protocol (L2TP).
  The PA-MC-8T1 port adapter reports that Firmware paused indefinitely and then reloads the router. There is no workaround.
  When IP Security (IPSec) hardware encryption is used, packets are not forwarded if Frame Relay or a generic routing encapsulation (GRE) tunnel is used.
  Workaround: Use the Cisco IOS software crypto engine.
  A Gigabit Ethernet interface might remain in an up/up state with no cable attached when running Cisco IOS Release 12.1(2). This condition can cause problems when running Hot Standby Routing Protocol (HSRP). The result is that if the active router fails, the backup router will take over; however, any traffic destined for the local segment from the original active router will be dropped. There is no workaround.
  The "long pound" feature does not work since the codec bytes were set to an incorrect value for a G.711 codec.
  There is no workaround.
  If a new recursive prefix is learnt by CEF, and a less specific prefix already exists, traffic that would have been forwarded using the less specific prefix will be dropped for up to 15 seconds, while the new recursive prefix is resolved.
  There is no workaround.
  Multiprotocol Label Switching (MPLS) labeled packets that are larger than 1500 bytes cannot be sent out FastEthernet even if "tag mtu" is configured to be larger than 1500 on the FastEthernet interface. There is no workaround.
  A Cisco router that is configured with Multiprotocol over ATM (MPOA) and Cisco Express Forwarding (CEF) might create an ingress cache and an egress cache with the same IP address, and the egress MPOA Client (MPC) address of the ingress cache points to itself. This situation might create a routing loop that causes the router to function improperly. This situation might occur during the route transitions in the network caused by topology changes.
  Workaround: Clear the CEF adjacencies using the clear adjacency EXEC command. Clear the corresponding ingress and egress caches using the clear mpoa client cache [ip-address ip-address] EXEC command.
  A Cisco 7507 router running Cisco IOS Release 12.0(7)T might reload by bus error at hdlc_address. The following message appears in the crashinfo file just before the reload:
  %ALIGN-1-FATAL: Illegal access to a low address addr=0x198, pc=0x600D1138, ra=0x60297D18, sp=0x624FC7C8
  There is no workaround.
  An SNA Switching Services (SNASw) router does not turn on the Command/Response (C/R) bit on the source service access point (SSAP) when replying to an exchange identification (XID) request.
  The trace between the SNASw router and the Point of Sale (POS) device shows that the C/R bit on the SSAP is not set when the SNASw router replies to the XID request. Thus, the POS device fails to begin the XID process. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1(2.1)T with authentication, authorization, and accounting (AAA) configured might reload with a bus error. There is no workaround.
  A Cisco AS5300 access server using Feature Group-D-Exchange Access North American (FGD-EANA) as the T1/channel-associated signaling (CAS) type does not strip off the two digits info-digits in front of the Digital Number Identification Service (DNIS) from the switch. This condition might result in reporting the wrong DNIS or failure of authentication.
  Workaround: Use the Cisco IOS image "c5300-is-ms.CSCdr59154."
  The tx-ring-limit command is deleted after reloading if it is configured with the service-policy command. If the service-policy command is not configured, this problem does not occur.
  Workaround: Re-issue the tx-ring-limit command after reloading the router.
  When you use the async mode interactive interface configuration command and autoselect during-login line command on a Cisco AS5800 series access server, autoselect might not work. With approximately 300 users online, the probability of this problem is about 20 percent, or 2 out of 10 calls. There is no workaround.
  When connecting an LU6.2 session to a low-entry networking (LEN) device with the Systems Network Architecture (SNA) Switch, snasw location statements do not take effect properly when that LEN device is connected over a link that is a defined link from the SNA Switch perspective (hence an uplink). There is no workaround.
  TXCONN is causing SNA sessions to unbind when it terminates conversations.
  Impact SNASw must rebind the sessions again before another conversation can be allocated. This condition is inefficient and might, in certain configurations, cause allocation failures. There is no workaround.
  Removing the last Virtual Private Network (VPN) route advertised by a provider edge (PE) router might result in all VPN forwarding information to remote PEs being lost.
  Workaround: After bringing the interface back up, enter the clear ip route vrf [name] * command for all configured VRFs.
  Alternate Workaround: Have at least one loopback address on the PE that is advertised into a VPN.
  On rare occasion, POTS ports might not ring when telephone devices are attached to a Cisco 800 series router. There is no workaround.
  Address family information that is specific to Border Gateway Protocol (BGP) does not appear in the configuration, even though it is applied by the router when the only configuration present for the address family is a redistribute router configuration command.
  Workaround: Add a network (router configuration command) statement in the BGP configuration for this address family so that the correct configuration is visible.
  A Cisco AS5300 access server might experience a bus error in csm_vtsp_dialing_done. This situation occurs on systems using channel-associated signaling (CAS) types. There is no workaround.
  Upon system initialization, ATM permanent virtual circuits (PVCs) are left in the inactive state and do not change to the active state unless the user issues a shutdown/no shutdown of the associated ATM interface in configuration mode.
  Workaround: Reinitialize the interfaces manually.
  A router might reload when using a TCP server on a unique port, which might happen when an internal ping runs to test connectivity with the host.
  Workaround: Configuring another DBConn TCP server on the same port. Two separate DBConn servers configured on the same port disables the internal ping mechanism.
  The OFF_TIME between digits for Modify/Channel Associated Signalling (MF/CAS) is currently 30 msecs when it should be 60 msecs. This condition might cause problems in digit detection when using MF/CAS. There is no workaround.
  Configuring a Cisco Route Switch Processor RSP with Multiprotocol over ATM (MPOA) and Cisco Express Forwarding (CEF) might lead to data loss when CEF uses the virtual channel connection (VCC) that was created by MPOA Client for data transfer in the following cases:
  Absence of MPOA client ingress cache
  Absence of the VCC created by MPOA Client
  Workaround: Clear the CEF adjacencies using the clear adjacency EXEC command.
  On Frame Relay permanent virtual circuits (PVC)s, Cisco Express Forwarding (CEF) might not restore the adjacency when a Frame Relay PVC goes down.
  Workaround: Clear the adjacency using the clear adjacency EXEC command.
  A Cisco 1750, 2600 series, or 3600 series router reloads when you use the no ip inspect command before the session information structure (SIS) is removed.
  Workaround: Wait until all sessions are removed, and then use the no ip inspect command to clean up the inspect configuration.
  A Versatile Interface Processor (VIP) or a Cisco 7200 series router might reload after fragmenting packets while CEF switching (distributed CEF switching in the case of the VIP) is enabled. There is no workaround.
  A Cisco router might reload if a large Certificate Revocation List (CRL) is retrieved during certificate processing. If the CRL is larger than 8000 bytes, the router might reload. Testing the crl request command (manual CRL retrieval) might cause the router to reload if the CRL is larger than 8000 bytes.
  Workaround: Decrease the size of the corresponding CRL on the CA server to smaller than 8000 bytes.
  Alternate workaround: Add the crl optional configuration subcommand to the crypto ca identity {name} global configuration command in the router configuration and remove the query url {url} configuration subcommand from the same crypto ca identity {name} command.
  When you have many Rumba clients configured for a low-entry networking (LEN) connection, exchange identification (XID) negotiation fails with sense 1016001A if you attempt to connect to a SNASwitch port that does not specify CONNTYPE "len." The adjacent node is inconsistent in its support of parallel trunk groups (TGs). A DLCTRACE shows that the first prenegotiation XID inbound from the Rhumba client contains the x'80' bit at displacement x'0F'. This trace indicates support for multi-link TGs (MLTGs). The next inbound XID has the x'80' bit off at displacement x'0F', indicating that it does not support MLTGs. SNASwitch properly fails the XID exchange and sends an XID with a CV22 appended with the sense code.
  This situation previously worked on the Advanced Peer-to-Peer Networking (APPN) code, so checking in the SNASwitch code is more stringent, but correct. There is no workaround.
  When e&m-immediate-start is used on a Cisco AS5300 series access server, dialout operation might be effected.
  The AS5300 may be unable to make outgoing calls after a period of operation.
  Workaround: Reboot the router.
  When two user sessions are using the show cable flap command at the same time and and the first user is stopped at the "-- More --" prompt, the Universal Broadband Router (UBR) will reload when the first user tries to continue passing the "-- More --" prompt while the second user goes through the show cable flap command output and enters the clear cable flap all command.
  Workaround: Limit user sessions with the show cable flap command to one.
  Alternate workaround: Disable the "-- More --"prompt by setting "terminal length 0."
  If a user opens a session, console or telnet, and quits from the session (ie "Ctrl ], quit", or unplugs the console cable) at the "-- More --" prompt, the user has to go back in to go through all the output of the show cable flap command or abort the CLI command.
  Otherwise, the clear cable flap all and the cable flap list aging commands will not be able to change the cable flap list command.
  In a configuration where dial-peers form a hunt group, the hunt-group feature might not work in all cases. For example, when a router fails to place a call through one dial-peer, it will not roll over to the next dial-peer in the hunt-group. There is no workaround.
  When a SNASwitch router has two ports defined to use one or two connection networks, and a CN link is established to a Virtual Telecommunications Access Method (VTAM) host, the transmission group (TG) negotiation fails when SNASwitch attempts to use the same tg#1 as the other real link (which is rejected with cv22 from the host).
  There is no workaround if multiple connection-network ports are required with parallel CN links to same real host.
  A Cisco7200 or Cisco7500 series router having ATM-PA3 in them, might reload due to bus error at address 0x50000000 or at address 0x08000000 in atmdx_rx_interrupt. The bus error always points to one of the ablve addresses.
  Reloads occur after the atm-dlx driver has received a packet properly and tries to process it.
  Workaround : Disable the ip cef (in Cisco7200 series) or ip cef distributed (in Cisco7500 series) global configuration command.
  Cisco Distributed Director will produce an incorrect URL when operating in HTTP redirect mode in Cisco IOS Release12.1. This problem does not occur in Cisco IOS Release11.1 IA. Domain Naming System (DNS)-based server selection is not affected by this problem. There is no workaround.
  In Cisco IOS Release 12.1(1.3), the show controllers e1 call-counters privileged EXEC command gives out the wrong call duration value for some channels on which active calls are carried. The call duration value returns to the correct value after the active call ends. Voice over IP (VoIP) calls go through without a problem. There is no workaround.
  When a gatekeeper configured with a static gateway entry has been reloaded, the gatekeeper can forward calls to the static gateway. However, approximately 30 minutes later, calls can no longer be forwarded, and a reload is required to resume functionality. There is no workaround.
  A Cisco 7500 series router that is running encryption in Cisco IOS Release 11.2 might experience spurious access and a software-forced reload after displaying the following log messages:
  Crypto engine 4: sign message using crypto engine: %SYS-2-NOBLOCK: idle with blocking disabled -Process= "Crypto SM"
  There is no workaround.
  Tunnel End-Point Discovery (TED) fails when TED probe is sent out on a multiaccess interface (such as Frame Relay or X.25) if the packet destination is not the next hop. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1(2.3) might reload because of a race condition between destruction of a physical unit (PU) control bank (CB) and a DSPU_ACTIVE message when the node is under stress. There is no workaround.
  A route switch processor (RSP) reloads with a bus error. A versatile Interface Processor (VIP) reloads when RSP reports a bus error exception. The slave RSP appears but does not pass traffic.
  The problem is related to the Duplicate Ring Protocol (DRiP), which is enabled when the Token-Ring Inter-Switch Link (TR-ISL) protocol is in use. Further, the problem occurs only when the I/O memory pool on the router runs dangerously low or is fragmented in such a way that buffers cannot be allocated for packet duplication. There is no workaround.
  Entering the resetcd {active CC} R command on the BPX node causes permanent loss of Tag Distribution Protocol (TDP) neighborship. This symptom results in service disruption between label edge routers (LERs).
  Other commands (such as switchcc) that force all the BPX slaves to resync with the label switch controller (LSC) might also also cause this problem.
  Another symptom is that the output for the show xtagatm cross-connect traffic interface xtagATM xxx command shows that both rx-cell-cnts and tx-cell-cnts are "n/a" for the affected cross-connects.
  Workaround: Perform a shut/no shut sequence on the Virtual Switch Interface (VSI) control interface on the LSC.
  Alernative workaround: Enter the resetcd command against the BXM connected to LSC.
  Alternate workaround: Perform a full rebuild.
  The CCH323 call control block (CCB) leaks when calls are terminated to the gateway (GW) and the GW receives an admission rejection (ARJ) from the gatekeeper.
  The memory held by CCH323_CT process will grow over time. The show process mem {process ID of CCH323_CT} command should be used to determine where memory is located. There is no workaround.
  Fragmented Encapsulating Security Payload (ESP) packets arriving on a router do not seem to be passed to the crypto engine because decryption counters are not increasing. No encrypted packets pass through the router. There is no workaround.
  A Manufacurer Serial Number is appended with 0x when it should be displayed as alphanumeric string. There is no workaround.
  When using class-based weighted fair queueing (CBWFQ) and low latency queueing (LLQ) on the physical interface of a Frame Relay connection (that is, the configuration in which the service-policy command is attached to the physical interface and the IP addresss given to the subinterface), the service-policy command is not activated after the router is reloaded. The following message appears during the boot process:
  Service policy policyTST is already attached
  After the reload completes, CBWFQ and LLQ are not active, even though the service-policy command is present on the interface in the configuration.
  Workaround: Remove the service-policy command from the interface, and then reapply it.
  When polling Cisco 2600 series routers in the network for Hot Standby Router Protocol (HSRP) statistics, the routers reloads.
  Workaround: Disable Simple Network Management Protocol (SNMP).
  In Cisco IOS Release 12.1(3), if an incoming dial call cannot get a dial modem to proceed under stress, this call will be processed as a voice call (that is, not as a modem or ISDN dialup call) if there is a voice modem card.
  In this case, Cisco IOS software is not able to find an incoming voice dial peer associated with the incoming call (this is always the case; otherwise, this call would have been treated as voice call in the first place). The memory allocated for the voice active call record and the voice call history record of this call will leak.
  This situation most likely occurs in a dial/voice combined configuration in which you have both the dial modem card and the voice modem card, a group of PRI lines for dial, and another group of PRI lines for Voice over IP (VoIP). In this case, if the incoming dial calls cannot find modem resources because of heavy call load, they are treated as voice calls. Usually, you do not have a dial peer for the call, so you will have memory leak.
  To avoid this case, ensure that every incoming dial call always gets dial modem resources. For example, make sure that you have enough digital signal processor (DSP) cards. There is no workaround.
  Weighted Random Early Detection (WRED) classifies all Multiprotocol Label Switching (MPLS) packets as precedence 0 in the MPLS->MPLS and MPLS->IP paths, regardless of their actual MPLS Experimental field values. There is no workaround.
  As a Cisco router boots Cisco IOS, it attempts to read the CPU EEPROM into a data structure that is not big enough, causing the router to reload. The crash dump typically looks like this:
  Nested r4k_return_to_monitor call (2 times) -Traceback= 0 6038EC00 6038EB14 6038D524 603768D4 6042C86C 6042BF88 6043866C 604379CC 6042EB18 Nested r4k_return_to_monitor call (3 times) *** System received a Bus Error exception *** signal= 0xa, code= 0x4008, context= 0x61869b70 PC = 0x60431200, Cause = 0x4020, Status Reg = 0x34008002 rommon 2 >
  There is no workaround.
  When you configure Class of Service (CoS) on an extended Tag (XTag) interface, you can only configure it on operational XTag interfaces. If the XTag interface is down, the CoS changes are not written to NVRAM or they become part of the running configuration.
  When you configure CoS on operational XTag interfaces, you might get false information when you display the running configuration.
  The data structure needed to store the value for CoS is dynamically allocated and does not exist when the XTag interface is down. The values for CoS supplied from the command line have no place to be stored, and the values will not apply.
  When the show run command is entered, the code loops through each XTag interface to obtain the CoS values from the appropriate data structure. The CoS percent bandwidth values are then stored in a global array that acts as a temporary placeholder and are written to NVRAM when the show run command output is displayed. If the next XTag interface is not operational, no structure exists that contains the CoS data, so the data cannot be stored in the global array. However, the values in the global array are still written to NVRAM. It is likely that the values are from the previous interface because the global array was never cleared.
  Workaround: If you want to apply CoS to XTag interfaces, you should only issue the tag-switching atm cos global configuration command if the XTag interface is up.
  You can verify the settings by using the show xtagatm cos-bandwidth-allocation xTagATM {0-2147483647} command.
  If you issue a show run command, then you should not pay attention to the CoS values displayed on the XTag interfaces.
  You should not perform a write memory command because it could cause the startup configuration to put corrupted CoS information on the XTag interfaces.
  CSCdm94333, which was integrated in Cisco IOS Release 12.0(9.6)S, introduced a new version of the automatic protection switching (APS) protect group protocol that required both working and protect routers to be upgraded simultaneously. The fix for CSCdr75997 relaxes this restriction by supporting interoperability between systems that are running different versions of the APS protocol so that working and protect can be upgraded independently.
  The VLAN0 of a Route Switch Module (RSM) participates in Open Shortest Path First (OSPF) election. There is no workaround.
  Full-duplex is not found and is not a valid function when a Cisco 3660 router is running Cisco IOS Release 12.0(7)T. However, the NM-4E and NM-1E Ethernet network modules should support full-duplex in all releases later than Release 12.0(4)T. There is no workaround.
  If a VPN routing/forwarding instance (VRF) static route points to a next-hop that is also resolved by a VRF static route, a tag is incorrectly imposed over the provider edge-customer edge (PE-CE) link. Consequently, traffic does not pass.
  Workaround: Change your VRF static routes so that one is not dependant on the other.
  Some versions of a Cisco uBR7200 series VXR/NPE300 with I/O control, Fast Ethernet, and the MC16C or MC16E cable modem cards will not boot. They will reboot to ROM by watchdog hard reset. There is no workaround.
  When using a Cisco AS5300 access server with Cisco IOS Release 12.1(2a)T1 to run two-stage Voice over IP (VoIP), the TFTP-based interactive voice response (IVR) loading will cause the input queue of Fast Ethernet to build up. There is no workaround.
  When a named access list is configured in a class map, a router running Cisco IOS Release 12.1(3) restarts by a bus error exception after the show class-map command in entered.
  Workaround: Avoid the show class-map command with named access lists, or use a numbered access list.
  When Cisco Express Forwarding (CEF) switching and Virtual Access interfaces are being used, a Cisco 7200VXR router running Cisco IOS Release 12.1(3) might experience high CPU usage because many alignment errors are occurring. There is no workaround.
  When the Cisco uBR924 cable access router in a Data Over Cable Service Interface Specification (DOCSIS) bridge mode is being operated, voice calls originating from the Cisco uBR924 do not work. There is no workaround.
  When a protected client station opens a new connection using the same port numbers (source and destination) too quickly after the previous connection was closed, the old structure is used by Cisco IOS to let the beginning of the session go through. However, the structure will be freed after timewait has elapsed. This causes the connection to be discontinued. There is no workaround.
  Windows Common Object Module Transaction Interface (COMTI) pauses indefinitely if your COMMAREA data size is less than the configurable COMMAREA size.
  Client connections pause indefinitely. There is no workaround.
  Encryption (IPSec) does not work on a Cisco 7200 series router with desktop and 3DES images when fast switching (using the ip route-cache interface configuration command) is turned on. There is no workaround.
  After you try to configure a secondary standby Hot Standby Router Protocol (HSRP) group, a Multilayer Switch Feature Card (MSFC) running Cisco IOS Release 12.1(1)E1 (or E2) experiences a bus error exception if a user attempts to remove the VLAN interface. Queued messages might resemble the following text:
  *** System received a Bus Error exception *** signal= 0xa, code= 0x10, context= 0x6160c240 PC = 0x602137a8, Cause = 0x2420, Status Reg = 0x34008002
  Workaround: Remove the secondary HSRP groups before removing the VLAN interface.
  H.323 gateway might reload when it experiences a time-out on admissions requests (ARQs) for call admission at the terminating gateway. There is no workaround.
  The ignore-dcd interface configuration command tells a router to look at Data Set Ready (DSR) to determine if a link is up. However, this command does not work unless the interface is shut down and then brought back up again. Then, the igonore-dcd command will take effect. This situation is undesirable for unattended sites.
  Workaround: For attended setups, shut/no shut the affected serial interface after the router boots.
  When a port-channel subinterface on a Cisco 7500/RSP series router that is running Cisco IOS Release 12.0 T is deleted, the entire interface (both the main and subinterfaces) is deleted instead of just the selected interface. There is no workaround.
  Hot Standby Router Protocol (HSRP) does not work with distributed Cisco Express Forwarding (CEF). There is a 50 percent ping rate for the HSRP virtual IP address, and the following message occurs on the versatile interface processor (VIP) when you enter the ip cef distributed global configuration command:
  VIP-Slot1# 20:45:17: %FIB-4-FIBXDRINV: Invalid format. virtual prefix with invalid if_number
  Workaround: Upgrade to Cisco IOS Release 12.1(4) or Release 12.1(4)T.
  When a user advances the system clock (using the clock set EXEC command) on a Cisco uBR7200 series router to pass the Baseline Privacy Interface (BPI) key expiration period, the system expects the cable modem (CM) to renew BPI keys. Since the CM does not know about the system time change, it does not renew its BPI keys.
  As a result, the CM stays in an online (pk) or reject (pk) state until it renews its BPI keys. The CM cannot pass traffic in these states.
  The time stamp is used as an index to BPI key tables. The granularity of the time stamp is 4 msecs. If two or more CMs obtain or renew their BPI keys within the same 4 msecs interval, only the first BPI key information will be kept. The CM stays in the online (pk) or reject (pk) state until it renews its BPI key.
  Workaround: Set the system clock before enabling the cable interface.
  Making a V.110 call from a Cisco AS5300 access server that is used as a network access server (NAS) connected to an SC2200 Signalling Controller in a special solution results in the call being rejected with the following message:
  Invalid information element contents
  There is no workaround.
  ISDN does not come up when using T1 or E1 on a Cisco AS5800 series access server. This situation occurs only in Cisco IOS Release 12.1(4.2), 12.1(4.3), and 12.1(4.2)T images. There is no workaround.

Novell IPX, XNS, and Apollo Domain

  The ipx routing {mac-address} command does not work properly when a DECnet address is used as the {mac-address} option. If you use DECnet and WANs such as ATM you will not be able to use a mac-address of your choice in the map statement for the WAN. There is no workaround.
  In Cisco IOS Release 12.0 and 12.1, but not in earlier releases, a Cisco router that is running Internetwork Packet Exchange-Enhanced Interior Gateway Routing Protocol (IPX-EIGRP) Service Advertising Protocol (SAP) on unstable WAN links might cause IPX EIGRP SAP packets to remain in memory, eventually exhausting processor memory.
  Workaround: Run Routing Information Protocol (RIP)/SAP instead of EIGRP/SAP on the WAN links.
  The host part of the ipx nhrp map ipx-address interface configuration command is not saved and, consequently, is incorrectly parsed when the show run command is issued. There is no workaround.
  A router reloads when handling Internetwork Packet Exchange (IPX) access- logging messages. This situation might occur even if access-logging is not enabled.
  You might exceed the 200 IPX network limit when hardware is activated after being removed. When the hardware is reactivated, interfaces that were not counted while the hardware was removed cause the limit to be exceeded. There is no workaround.

Protocol Translation

  A Cisco 3640 router reloads into ROM Monitor mode when it is routing background traffic and running several X.25 translation sessions. There is no workaround.

TCP/IP Host-Mode Services

  A Cisco router that is running Cisco IOS Release 12.1(1) or Cisco IOS Release 12.0(7)T might experience a stopped input queue on a serial interface while under normal operation. There is no workaround.

Wide-Area Networking

  The National German BRI switch type 1TR6 is not supported for voice calls. There is no workaround.
  On a Cisco router, the PPP WAN layer might call idb->reset whenever link control protocol (LCP) _close state is reached and whenever the router fails to get a confirmation acknowledgment after repeatedly sending configuration request messages to the peer. Since the idb->reset in the ISDN driver resets the physical layer, you see physical layer going up and down whenever the protocol layer with the peer is broken. There is no workaround.
  A Cisco 3620 that is running Cisco IOS Release 12.0(7)T might reload on l2f_ip_udp_input process when a single virtual private dialup network (VPDN) group is used for L2F dial-in and L2TP dial-out.
  Workaround: Use separate VPDN groups for these two services.
  A Cisco AS5300 series router might reload under normal traffic operation with Cisco IOS Release 12.0(7)T. There is no workaround.
  On a Cisco router that is running Cisco IOS Release 12.0(7)T and that is configured to perform X.25 switching with local acknowledgement, some of the switched X.25 virtual circuits (VCs) might pause indefinitely. There is no workaround.
  A Cisco 2610 router that is configured with dialer profiles and that is running Cisco IOS Release 12.0(7)T or Release 12.1(1) might experience a low and variable success rate in ping tests.
  Workaround: Configure the hidden dialer disable multiencaps command.
  On an access platform, when dialer profiles and virtual profiles are used together with a certain vendor's callback, the virtual profile is not applied to the callback call. There is no workaround.
  Please contact the Technical Assistance Center (TAC) for more information.
  A router running Cisco IOS Release 12.0(7.3) might experience the following error message when the show x25 EXEC command is used:
  System returned to ROM by bus error at PC 0x6042D870, address 0xD0D0D0
  There is no workaround.
  Two CAUSE information elements are sent in a STATUS message. There is no workaround.
  The show x25 xot EXEC command might trigger a reload. The system returns to ROM by an Illegal Instruction error. There is no workaround.
  Versatile Interface Processor (VIP) might reload with the following message:
  DMA-1-DRQ_STALLED: DRQ stalled. Dumping DRQ
  There is no workaround.
  Layer 2 Tunneling Protocol (L2TP) with Multilink Protocol (MLP) configured on a L2TP Network Server (LNS) and PPP client causes packets to be dropped in L2TP dialout, because pak fields are not initialized correctly after switching and there is a routing loop. There is no workaround.
  If Dial-out is configured, the Virtual Private Network (VPN) tunnel is freed because the dialer in network access server (NAS) fails to dial the PPP client. This situation occurs when the dialer in the NAS is unconfigured and then reconfigured. There is no workaround.
  A router reloads when downloading large (10MB) HTTP files using Multilink PPP (MLP) with a Layer 2 Forwarding (L2F) tunnel. There is no workaround.
  Cisco IOS software installs bogus dynamic dialer maps for the Large-Scale Dial-Out feature when the primary RADIUS server fails, and the network access server (NAS) fails over to the secondary. Here is a sample output for the show dialer map privileged EXEC command:
  Dynamic dialer map ip 13.13.13.13 name dun0751995 (3602218413) on Di0 Dynamic dialer map ip 0.0.0.1 name dun0755205 (3606946292) on Di0 Dynamic dialer map ip 0.0.0.9 name dun0807006 (3347353827) on Di0 Dynamic dialer map ip 171.18.52.205 name dun0807015 (8504214328) on Di0 Dynamic dialer map ip 220.86.120.187 name dun0807016 (8505470577) on Di0 Dynamic dialer map ip 0.15.11.72 name dun0807031 (8509841046) on Di0 Dynamic dialer map ip 172.26.178.253 name dun0807180 (9125576877) on Di0 Dynamic dialer map ip 171.18.52.205 name dun0807293 (8137198235) on Di0 Dynamic dialer map ip 172.25.24.253 name dun0807607 (8136648661) on Di0 Dynamic dialer map ip 172.25.90.245 name dun0807751 (9124364312) on Di0
  All the above entries except for 172.x.x.x are bogus, such as "13.13.13.13."
  Workaround: Reload the router to clear these dialer maps once they are installed.
  The convenience CLI command, show vpdn group, does not produce the desired output. Use show running-config as a workaround and look for the vpdn-group configuration.
  A Cisco 7513 router with RSP4 may reload intermittently when running Annex G. There is no workaround.
  A Cisco 802 router is not responding to Embedded Operations Channel (EOC) loop code on lease line ISDN. There is no workaround.
  Any Cisco router running Cisco IOS Release 12.1(2.1) fails to register with ATM Address Resolution Protocol (ARP) server if the User-to-Network Interface (UNI) was configured to a UNI 3.0 version. Other ARP clients fail to resolve the logical IP subnet (LIS) of neighboring clients through the ATM ARP server. There is no workaround.
  You may see the message "%ALIGN-3-SPURIOUS: Spurious memory access ..." when a certain vendor's callback is being used and a callback fails to complete. The stack trace indicates the alignment error occurred in the dial_if() function, called from dialer_enable_timeout().
  There is no workaround. Please contact the Technical Assistance Center (TAC) for more information.
  When the primary D channel is shut in PRI Non-Facility Associated Signalling (NFAS) operation, the D-channel layer 1 might still be in the ACTIVATE state and layer 2 might still be in the AWAITING_ESTABLISHMENT state. There is no workaround.
  On a Cisco router running Cisco IOS Release 12.1(3), the LAN Emulation Clients (LECs) fail to send LE_ARP responses if the LAN Emulation (LANE) interfaces form part of a bridge group with routing protocols enabled and are in the blocking state. Routed packets addressed to the router's own interfaces might be dropped. There is no workaround.
  A Cisco router running Cisco IOS Release 12.0 or Release 12.1 might reload if it is running low in memory and a new Logical Link Control (LLC) session is added. There is no workaround.
  A reload might occur during heavy usage at the Layer 2 Tunneling Protocol (L2TP) network server (LNS). There is no workaround.
  The ISDN Switchtype is not set on BRI S/T and U interfaces on BRI WAN interface cards (WICs). There is no workaround.
  On a Non-Facility Associated Signalling (NFAS) configuration, the ISDN subsystem might hold a large amount of processor memory which causes the system to run out of memory.
  Workaround: Reload the system.
  When a Non-Facility Associated Signaling (NFAS) outgoing setup specifies a preferred B channel and interface and the "switch" negotiates for the same B channel on another interface, the B channel on the interface specified in the outgoing setup will remain in the proposed service state. B channels in the proposed service state are not useable for outgoing calls.
  A change in the digital subscriber lines (DSLs) state through configuration (using the shutdown and no shutdown commands) will not restore the B channel(s) to IN_SERVICE.
  Workaround: Reload.
  The following messages are generated for every call disconnect with switchtype primary-ni configured:
  ISDN ERROR: Module-l3_sdl_u Function-Ux_BadMsg Error-Source ID = 400 Event = AB
  There is no workaround.
  Packets that cannot be switched using Cisco Express Forwarding (CEF) through a virtual private dialup network home gateway/L2TP Network Server (VPDN HGW/LNS) might cause alignment errors.
  Workaround: Ensure that all packets can be CEF-switched.
  Alternate workaround: Disable CEF switching on the HGW/LNS virtual template.
  On a Cisco 803 router running Cisco IOS Release 12.1(2), a Memory Protection Violation might appear.
  Memory Protection Violation: epc =0x004F6C8C, location of fault eva =0x00000???, read fault address error=0x00000082, illegal address pti =0x00000000 pte =0x02102100
  CPU Register Context: PC = 0x004F6C8C MSR = 0x00009032 CR = 0x33500053 LR = 0x00275374 CTR = 0x00125DCC XER = 0xE000AF7F R0 = 0x00000000 R1 = 0x0061BBEC R2 = 0x0A010901 R3 = 0x0061BBF4 R4 = 0x00000000 R5 = 0x00000000 R6 = 0x0061BD10 R7 = 0x0271934C R8 = 0x0282E624 R9 = 0x0061BBF4 R10 = 0x00000000 R11 = 0x02738FF4 R12 = 0x55000053 R13 = 0x00010001 R14 = 0x00000000 R15 = 0x00000001 R16 = 0x02737F38 R17 = 0x00000001 R18 = 0x00000001 R19 = 0x0282E624 R20 = 0x00000002 R21 = 0x0271934C R22 = 0x00000000 R23 = 0x00600000 R24 = 0x0061BD10 R25 = 0x00000000 R26 = 0x00000000 R27 = 0x0061BD54 R28 = 0x02730498 R29 = 0x02742984 R30 = 0x027448A4 R31 = 0x0273C398
  Traceback: 0x004CC50C 0x004ABE00 0x004A0EF8 0x004c6E4E4 0x0046E4B4 0x00000000
  There is no workaround.
  A router reloads if the following interface configuration command is entered on a Frame Relay multipoint subinterface:
  frame-relay map ip ip-address dlci rtp header-compression
  Workaround: Specify the number of compression connections:
  frame-relay map ip ip-address dlci rtp header-compression connections 256
  Calls might be disconnected on some switchtypes during "overlap sending" due to the processing of T304_EXPIRY (expiration of timer T304). The time (in a Q.931 debug trace) between the INFORMATION transmission and the DISCONNECT transmission is extremely short. There is no workaround (using Primary-NI switchtype may work for some users).
  A Cisco 7206VXR router that is running Cisco IOS Release 12.1(1)E reboots with bus errors approximately once per day, with no obvious changes or reasons. There is no workaround.
  A Cisco router running a Cisco IOS 12.1 release that is configured for X.25 might experience a memory leak in X.25 background. There is no workaround.
  If you enter the show isdn memory EXEC command on a Cisco AS5300 access server, you can see that the line interface (LIF) timer blocks reach the maximum number and no more calls can be accepted by the router. A reload is required in order to clear the blocks. There is no workaround.
  According to an Italian Homologation test requirement, a router should not respond with a STATUS message when it receives a NOTIFY message in the U3 and U4 states (for example, following the receipt of a Call Proceeding and Alerting message, respectively).
  NOTIFY should be ignored in U3 and U4 states for all switch types. There is no workaround.
  A Cisco 7500 router reloads because of a bus error. The reload appears to be triggered when Open Shortest Path First (OSPF) link state advertisement (LSA) updates reach the router. Updates includes an LSA, a routing table update, or a Tag Forwarding Information Base (TFIB) update. There is no workaround.
  When initializing a Cisco router, the PRI receives the following message:
  Channel State (0=Idle 1=Propose 2=Busy 3=Reserved 4=Restart 5=Maint_Pend) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Service State (0=Inservice 1=Maint 2=Outofservice) 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
  E1 and primary-net5 will not work.
  Workaround: Enter the isdn service dsl0 b_channel 0 state 0 interface configuration command. Though doing so clears the problem, the initial situation recurs after reloading. E1 is not accessible unless you perform this workaround or the alternate workaround (below). If you reload the router, E1 will be inaccessible again.
  Alternate workaround: Perform a shut/no shut sequence of the controller. Though doing so clears the problem, the initial situation recurs after reloading.
  A Cisco 7500 series router configured with LAN Extender may reload.
  Workaround: Turn off fastswitching.
  On a Cisco 2600 or 3600 series router, no T1 B-channel is available, even if the T1 ISDN layer is active.
  Workaround: Perform a shut/no shut sequence.
  Challenge Handshake Authentication Protocol (CHAP) authentication does not work on a Layer 2 Tunneling Protocol (L2TP) network server running Cisco IOS Release 12.1(1.5) or later releases. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)T might reload when Domain Name System (DNS)-based X.25 routing is used. There is no workaround.
  A router running Cisco IOS Release 12.(3.0.3) and image c800-nsy6-mw reloads when you reconfigure the isdn switch-type global and interface configuration command at the "Config Term" level (and isdn switch-type basic-5ess configured previously) and initialize interface bri0 (shut/no shut). The following sequence of commands illustrates this situation:
  uut1(config)#isdn switch-type basic-5ess
uut1(config)#interface bri0 uut1(config-if)#shut
uut1(config-if)#no shut
  There is no workaround.
  A Cisco 2600 series router running Cisco IOS Release 12.0(11.6) reloads when a call connects with Multilink PPP configured on Dialer Interface. There is no workaround.
  After several hours of operation, a router reports the following error message:
  %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004C59C handle -1
  The show isdn memory EXEC command shows that the line interface (LIF) timer blocks has reached the limit of 30911.
  Workaround: Always use interface 0 as the primary Non-Facility Associated Signaling (NFAS) D channel for Signaling System 7 (SS7) interconnect and PRI ISDN line connection.
  A link error occurs because of a function in Cisco IOS core file "linkdown_event.c" that calls a PPP function that might not be present in some builds. There is no workaround.
  Generating a running-config after configuring resource pooling and Virtual Private Dialup Network (VPDN) might cause a reload. There is no workaround.

Resolved Caveats—Release 12.1(3)

  All the caveats listed in this section are resolved in Release 12.1(3). This section describes only severity 1 and 2 caveats.

Basic System Services

  If you send a "stream" (reverse telnet) print job through an asynchronous line, and the print job is very small, the job might fail to print. In releases earlier than Cisco IOS Release 12.0(7)T or Cisco IOS Release 12.0(8), this situation occurs if the flush-at-activation command is active on the line. In later releases, the situation occurs if either flush-at-activation command is active on the line, or if the aaa new-model global configuration command is configured on the system.
  Workaround: Use line printer daemon (LPD) or upgrade to Cisco IOS Release 12.0(7)T, Cisco IOS Release 12.0(8) or a later Cisco 12.0 release, or Cisco IOS Release12.1 and either use the no flush-at-activation command on the asynchronous lines or unconfigure the aaa new-model global configuration command on the system.
  IP Security (IPSec) configured with FastSwitch works on the serial interface but fails on other interfaces. There is no workaround.
  A Cisco AS5300 access server reloads with a bus error at PC 0x6022A3C8 when you try to Telnet to a named-host. There is no workaround.
  Block Serial Tunneling (BSTUN) char-set ascii mode on Cisco 1600 serial interfaces does not interoperate with some Binary Synchronous Communications protocol (Bisync) implementations. The Bisync code will reject frames that have trailing Pad (0x7F) characters. You might see the following message while the debug bsc event command is enabled:
  BSC: Serial0:POLLER-FSM event:E_RxInvalid old_state:TCU_Polled. new_state:TCU_Idle.
  There is no workaround.
  A Cisco router might experience degraded performance when several duplicate accounting records are sent. There is no workaround.
  A Cisco access server might unexpectedly reload when there is a high number of Cisco DialOut calls.
  Workaround: Temporarily disable DialOut.
  A Cisco 3660 router might experience a spurious memory access when booting up. There is no workaround.
  For Large-Scale Dial-out, there are static routes in the following format:
  ip route 50.0.0.1 255.255.255.255 Dialer0 200 name LSDOtest
  This format gets inappropriately modified to:
  ip route 50.0.0.1 255.255.255.255 Dialer0 200
  The wrong username is extracted (formats ip address-out instead of username-out). There is no workaround.
  When the network access server (NAS) and PPP client attempt to negotiate both Internetwork Packet Exchange (IPX) and IP for a session, attribute 8 (IP address) is dropped from the network accounting records. This situation does not occur if IP is the only protocol negotiated. There is no workaround.
  A defect in multiple versions of Cisco IOS software will cause a Cisco router or switch to stop and reload if the Cisco IOS http service is enabled and an attempt is made to browse to http://<router-ip>/%%. This defect can be exploited to produce a denial of service (DoS) attack. This defect has been discussed on public mailing lists and should be considered public information.
  The vulnerability, identified as Cisco bug ID CSCdr36952, affects virtually all mainstream Cisco routers and switches running Cisco IOS Release 11.1 through Release 12.1. The vulnerability has been corrected and Cisco is making fixed versions available to replace all affected Cisco IOS releases. Customers are urged to upgrade to releases that are not vulnerable to this defect.
  Workaround: Nullify the vulnerability by disabling the Cisco IOS HTTP server, by preventing access to the port in use by the HTTP server on the affected router or switch, or by applying an access-class option to the service itself. The IOS HTTP server is not enabled by default except on a small number of router models in specific circumstances.
  Please see http://www.cisco.com/warp/public/707/ioshttpserver-pub.shtml for the latest complete version of this security advisory.
  The system reloads during authentication, while accessing already freed memory.
  This failure occurs in Password Authentication Protocol/Challenge Handshake Authentication Protocol (PAP/CHAP) authentication.
  There is no workaround.
  There is a severe performance degradation for the NPE300 processor, specifically with the Packet Over SONET (POS) and serial port adapters (PAs).
  Workaround: Upgrade to Cisco IOS Release 12.1, 12.1 E, or 12.1 T.
  The Service Assurance agent (SAA) responder can cause a memory leak if it is responding to SAA packets with different target IP addresses but the same target User Datagram Protocol (UDP) port. There is no workaround.
  The Cisco Service Assurance Agent (CSAA; formerly known as Response Time Reporter (RTR)) granularity for the icmpecho is 4 milliseconds (msecs), even in Cisco IOS Release 12.0 T. Granularity should be 1 msec. There is no workaround.
  The chassis interface and various interface processors on a Cisco 7500 series router does not come up after a reload. There is no workaround.
  If a router has access to more than one TFTP server when it attempts to netboot, the netboot can fail if one TFTP server has the file being netbooted, while other TFTP servers do not have the file.
  Workaround: Use the boot filename ip-address ROM monitor command in conjunction with the IP address of the TFTP server that has the image.
  Alternative workaround: Disable the TFTP server on the systems that do not have the file.
  Alternative workaround: Put the file on all TFTP servers that are accessible from the router being netbooted.
  A Cisco MC3810 multiservice access concentrator running Cisco IOS Release 12.1(2.1) reloads while disconnecting a call. There is no workaround.
  Environmental alarms are triggered spontaneously due to wrong hardware revisions. Only AC power supply alarms are affected. There is no workaround.

IBM Connectivity

  A Cisco router that is configured for Synchronous Data Link Control (SDLC) and Qualified Logical Link Control (QLLC) through data-link switching (DLSw) local switching might reload with a bus error at:
  PC 0x60F03F1C, address 0x10D.
  There is no workaround.
  In a Blocked Serial Tunnel (BSTUN) network, the BSTUN state might be open on the headend and closed on the remote router after the headend router is reloaded.
  Workaround: Enter the no bstun route interface configuration command, followed by the bstun route command. The tunnel then opens up on both sides.

Interfaces and Bridging

  A Cisco router that is running Cisco IOS Release 12.0(04a) might receive the following alignment error messages on dot1q over Internetwork Packet Exchange service access point (IPX SAP):
  %ALIGN-3-CORRECT: Alignment correction made at 0x605B8E70 reading 0x63E5A6F1 %ALIGN-3-TRACE: -Traceback= 605B8E70 60095714 60095700 00000000 00000000 00000000 00000000 00000000
  IPX SAP is affected only when dot1q encapsulation is involved (such as when IPX SAP is configured on a dot1q interface).
  These alignment error messages do not affect connectivity or cause packet loss. There is no workaround.
  If the input line to a PA-A3 port adapter in a Cisco 7200 or 7500 series router is not functioning properly and the port adapter is experiencing many alarms or errors on the line, link-up messages may appear without link-down messages on the console logs.
  Confirm this occurrence by entering the show controllers atm privileged EXEC command. Check the cables and the interface on the other end of the link for problems. There is no workaround.
  In Constrained Multicast Flooding (CMF), when a bridge group member joins and leaves a multicast group, multicast traffic still goes out that bridge member.
  Work around: Enter the clear bridge multicast EXEC command. Traffic stops going out of the bridge member.
  A Cisco Packet OC-3 Interface Processor (POSIP) might reload with a bus error. There is no workaround.
  A Cisco 7500 series router with VIP2 Versatile Interface Processor boards containing a PA-ATM Enhanced port adapter and a PA-POS port adapter will experience an interface reload and will not be able to send or receive traffic.
  Workaround: Configure these two high-speed PAs on different VIPs.
  With IP Cisco Express Forwarding (CEF) and VLANs configured, the router might reload with a bus error when you use the configure memory privileged EXEC command. There is no workaround.
  A packet that is 1500bytes or longer (ip/tcp/payload) that is bridged from a Bridge Group Virtual Interface (BVI) to a Token Ring and then moves out to the bridged environment receives 8 bytes of extraneous characters inserted between the Token Ring MAC header and the IP header.
  Workaround: Disable fast switching on the BVI interface.

IP Routing Protocols

  A Cisco 2500 or 4000 series router might reload when you unconfigure virtual subinterfaces on X.25 or Frame Relay networks from a partial mesh topology for Enhanced Interior Gateway Routing Protocol (EIGRP) after you use the no ip routing global configuration command for some images such as the js image.
  Workaround: Remove the subinterfaces before you use the no ip routing global configuration command.
  Alternate workaround: Remove the EIGRP process with the no router eigrp global configuration command before using the no ip routing global configuration command.
  If you move interfaces from a normal area to a new Not-So-Stubby Area (NSSA), the internal NSSA autonomous system boundary router (ASBR) might generate type-7 link-state advertisements (LSAs) into the NSSA area without the propagate bit (P-bit) set, so NSSA Area Border Router (ABRs) will not translate these LSAs into type-5s. The P-bit is not set because the router treats the P-bit as an NSSA ABR as well as an NSSA ASBR which, per Request For Comments (RFC) 1587, should not set the P-bit in type-7 LSAs.
  Workaround: Once the interfaces have been moved to the new NSSA area, perform the following steps:
  no area x (x is the old area)
  no area y nssa (y is the new NSSA area)
  area y nssa
  The above sequence will remove the old area and reoriginate type-7 LSAs with the P-bit set.
  Alternate workaround: Instead of moving the interfaces to a new NSSA area, change the existing normal area into an NSSA area by using the area area-id nssa router configuration command.
  Under rare circumstances, a link-state advertisement (LSA) on a neighboring router might get stuck in MAXAGE state and not be deleted. In this situation, the LSA cannot be originated again on this router, which might cause the route to become unavailable or cause packets to take another route that is less than optimal. This situation has been seen to occur when an OSPF neighbor runs out of memory and OSPF tables are corrrupted.
  Workaround: Restart the OSPF process by entering the clear ip ospf proc command.
  Interoperability between NetMeeting 2.1 and NetMeeting 3.0 might not work properly when used with Cisco IOS Release 12.1(1) Network Address Translation (NAT). Some static NAT configurations do not work for NetMeeting 2.X when used with Cisco IOS Release 12.1 Network Address Translation (NAT). There is no workaround.
  If you configure Protocol Independent Multicast (PIM) or Hot Standby Router Protocol (HSRP) on an ATM-LANE interface, the CPU of the Route Switch Processor (RSP) might reach 99 percent. This situation only occurs when Open Shortest Path First (OSPF) is enabled on more than 12 interfaces in combination with ATM-LANE. This situation does not occur on an RSP that is running Cisco IOS Release 12.0 S or Release 11.2 GS. There is no workaround.
  In Cisco IOS Release 12.1 and Release 12.1 T, the IP policy cache is not cleared when you use the no ip route-cache policy interface configuration command.
  Workaround: Issue the clear ip cache {prefix | mask} command for a specific cache.
  When a reservation for a label switched path (LSP) is torn down, Resource Reservation Protocol (RSVP) sends the upstream neighbor a ResvTear message containing a RESV_CONFIRM object. RSVP maintains the reservation and keeps sending this message periodically until the upstream neighbor responds with a ResvTearConf message or the reservation times out.
  If the downstream neighbor continues to send RESV refreshes for the LSP, the reservation never times out. In this situation, if the upstream neighbor never sends a ResvTearConf, the reservation remains stuck in this state indefinitely. The ResvTearConf mechanism is no longer defined in the RSVP-TE IETF draft, and is not used by some non-Cisco IOS implementations.
  Workaround: Shut down the LSP at the headend. Wait for the reservation to time out (2 minutes) and then reestablish the LSP.
  If you enter the show ip mds summary EXEC command or the show ip mds forwarding EXEC command at the Versatile Inter face Processor/line card (VIP/LC) console, then a counter to track the number of multicast routes might be incorrectly decremented and underflow to a huge number. The VIP/LC console might display a syslog message similar to the following:
  %MDS-4-ROUTELIMIT: 4294967237 routes exceeded multicast route-limit of 2147483647
  If this situation occurs, no additional multicast routes can be created and the router must be restarted.
  Workaround: Do not use the show ip mds summary EXEC command or the show ip mds forwarding EXEC command at the VIP/LC console.
  This problem occurs on a Cisco AS5300 access server with the Simple Gateway Control Protocol (SGCP) call model only. A Cisco AS5300 access server reloads when you enable a (specifically requested) hidden feature that is only applicable to a SGCP call model. There is no workaround.
  When virtual private network (VPN) is configured and Border Gateway Protocol (BGP) is used to exchange prefixes between the provider edge (PE) router and the customer edge (CE) router in Cisco IOS Release 12.1(02)T, the following tracebacks might occur when the PE router needs to withdraw a route from the CE router:
  %ALIGN-3-SPURIOUS: Spurious memory access made at 0x6062A6CC[bgp4_format_unreachable(0x6062a5bc)+0x110] reading 0x41 %ALIGN-3-TRACE: -Traceback= 6062A6CC[bgp4_format_unreachable+0x110] 6062C28C[bgp_send_update+0xa84] 6062CC40[bgp_update_walker+0x178] 610020E0[rn_walktree_version+0x158] 6062D080[bgp_updates+0x2e0] 6062D87C[bgp_update_af+0x698] 60621BA8[bgp_router+0xa68] 602C9074[r4k_process_dispatch+0x14]
  These occurences might include neighbors going down and removal of interfaces. There is no workaround.

ISO CLNS

  On Cisco routers, if a DECnet packet is received on an ethernet interface and is converted to a CLNS packet, in the unlikely event that it cannot be routed to its destination, an unexpected reload may occur in ether_extract_addr.
  There is no workaround.

Miscellaneous

  After sending packets at a very high rate through a PA-A3 port adapter on a Cisco 7200 series router, the router may reload if you remove the card and reinsert it to a different slot with a new configuration on the same interface at the new slot. The following error patterns occur on the console:
  %SYS-2-LINKED: Bad enqueue of 60DFD980 in queue 60CCFB30
-Process= "<interrupt level>", ipl= 1
-Traceback= 602398F0 601C1370 602000C8 60203958 601C4408 601C81B0
%SYS-2-BADSHARE: Bad refcount in retparticle, ptr=14AF, count=0
-Traceback= 601C14FC 602000C8 60203958 601C4408 601C81B0
  There is no workaround.
  Cisco Encryption Technology (CET) connection is established but encryption and decryption fails. There is no workaround.
  There is an erratic disconnect of Layer 2. The debug isdn q921 command shows bad sequencing and a PRI seems to have a Layer 2 looped. There is no workaround.
  If there is a large network delay and Asynchronous Serial Protocol (ASP) does not support local acknowledgements (ACKs), you need to increase the poll interval on the end device to accommodate the delay. There is no workaround.
  When a crypto map is deleted for a subinterface, the system reloads. There is no workaround.
  The no ip route-cache cef interface configuration command is missing from the configuration after system reload.
  Workaround: Turn off the ip route-cache cef interface configuration command when you reload the system, if it is in the configuration file.
  When some PC fax modem software in conjunction with fax relay is used, proper operation is not accomplished unless the software is operating in class 1 mode when sending and in class 2 mode when receiving. There is no workaround.
  In a Cisco AS5800 access server using channel-associated Signalling (CAS) loop start signalling, ds0s might stop in l_wait_connect0 state while the trunk state shows that the far end is onhook. The state machine is not handling a host disconnect properly in this state. There is no workaround.
  When you configure several hundred PPP over ATM virtual circuits (VCs) on a single subinterface using VC classes, some of the VCs might end up with a virtual access interface number of 0.
  Workaround: Delete and reconfigure the VC to obtain a valid VA number. To prevent this problem from occurring, do not use VC-class statements. Another method is to netboot the configuration and ensure that the physical interface is shut down before configuring the interface and then bringing it back up.
  In some circumstances, the adjacency associated with a static Address Resolution Protocol (ARP) entry might be deleted if the interface is shutdown. There is no workaround.
  A Cisco 3600 series router with a PRI T1 or E1 card might start dropping almost all traffic if weighted fair queueing (WFQ) is used on any interface. The show buffers EXEC command will show buffer element misses, but it will also show many free buffer elements. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)S or Cisco IOS Release 12.0(8)S will disable Cisco Express Forwarding (CEF) with a FIB-3-NOMEM failure even though there appears to be plenty of memory. There is no workaround.
  When there is a high volume of traffic moving through IP Security Protocol (IPSec), you might see the following message:
  Memory allocation of 18196 bytes failed.
  There is no workaround.
  In countries where low powered mode is supported (ISDN physical layer is deactivated when there is no traffic), the user may notice that sometimes data traffic will not bring up the ISDN line out of low-powered mode. In this case, only an incoming call would correct the situation and bring up the line.
  Workaround: Enter the clear int bri0 EXEC command .
  When you initiate remote copy protocol (rcp) from a remote system, the system will not check the available space on Flash memory before initiating the copy. For files that are too large, all the memory is consumed and the connection is reset without sending an error message.
  Workaround: Use the copy rcp flash command on the system or check the available Flash memory before copying from the remote system.
  A Cisco 3640 router that is running Cisco IOS Release 12.0(7)T might experience problems when the T1 controller is shut down. When the T1 controller is shut down, a Simple Network Management Protocol (SNMP) trap is generated, but when the T1 connection is pulled from the router, SNMP trap is not generated. There is no workaround.
  When Hot Standby Router Protocol (HSRP) is configured between two NM-1FE-TX network modules that are connected to any Cisco switch, both switches become active, and the interface on the switch with the lower HSRP priority will flap. There is no workaround.
  The TI1575 ATM driver reloads with bus error, due to double freeing buffers. There is no workaround.
  A Cisco MC3810 multiservice access concentrator configured for circuit emulation service (CES) requires Data Terminal Ready (DTR) to be set high in order to avoid initial video freezes.
  When interacting with a non-MC3810 device, some of those devices toggle DTR and cause a traffic problem (for example, video freezes). There is no workaround.
  On a Cisco AS5800 series access server that is running Cisco IOS Release 12.0(5)T or portware 2.7 or higher, once a modem is used for handling V110 calls, it cannot be reused to handle asynchronous calls. The asynchronous calls will fail and the user will be disconnected. There is no workaround.
  PPP over Internetwork Packet Exchange (IPX) over ATM might not work properly on a Cisco 4500 series router that is running Cisco IOS Release 12.1(1.0.4)PI. This situation does not occur in a router that is running Cisco IOS Release 12.1(0.13) or Cisco IOS Release 12.1(0.9)T4. There is no workaround.
  A low memory condition might provoke a reload in Integrated File System (IFS). There is no workaround.
  In Signaling System 7 (SS7) applications, a Cisco router might not clear calls properly when the T1/E1 cables are unplugged from the interfaces. The show isdn {active} EXEC command shows the calls are still active. There is no workaround.
  If Resource Reservation Protocol (RSVP) receives path messages for the same label switched path (LSP) session on two different interfaces, it sends RSVP Reservation (Resv) messages alternately on those interfaces. Only the messages sent on one of the interfaces contain a valid label. As a result, some packets might be forwarded with no label. There is no workaround.
  Multilink PPP calls might not work properly when you dial in to a Stack Group Bidding Protocol (SGBP) group formed by two routers connected through a split dial shelf. There is no workaround.
  File Transfer Protocol (FTP) fails with a large packet. The maximum transmission unit (MTU) calculated by the router during the MTU discovery (MTUD) process is incorrect. There is no workaround.
  When you use Cisco Express Forwarding (CEF), generic routing encapsulation (GRE) tunnels, and tunnel checksums, the packet loss rate is abnormally high.
  Workaround: Disable tunnel checksums or disable CEF on the tunnel interface.
  A Route Switch Processor (RSP) that is configured with Multiprotocol over ATM (MPOA) clients might take a large number of packets (approximately 500) to transfer the data through the MPOA shortcut. There is no workaround.
  A Cisco Route Switch Processor (RSP) might reload when Generic Traffic Shaping (GTS) is enabled on an interface. This situation only affects RSPs. The RSP will reload when you configure the GTS with the traffic-shape group interface configuration command with an access list that has a deny statement in it. There is no workaround.
  A Cisco router configured to redirect Multiprotocol Label Switching (MPLS) and Web Cache Communication Protocol (WCCP) on the same outgoing interface does not redirect HTTP packets to the cache engine, although the WCCP statistics are being updated. The HTTP traffic goes directly to the destination HTTP server, and cache is not being used. There is no workaround.
  A Cisco router (800 series through 4700 series) might silently drop large packets. This situation occurs when both Cisco Express Forwarding (CEF) and output rate-limiting are enabled on an interface, and a packet to be switched out of an interface is larger than the maximum transmission unit (MTU).
  Workaround: Disable CEF or disable the Committed Access Rate (CAR) on interfaces that will need to fragment packets.
  Tag switching cannot be used on the 4/8 port T1/E1 IMA network modules for Cisco 2600, 3620, 3640, or 3660 series routers. Tag control packets are not sent out the interface, but instead are queued and dropped once the virtual circuit (VC) output queue is filled. There is no workaround.
  When a router has only Intrusion Detection System (IDS) configured and Context-Based Access Control (CBAC) is disabled, the router sends a TCP Reset (RST) packet for some sessions, after about 11 hours.
  Workaround: Configure the ip inspect name global configuration command as follows:
  ip inspect name try tcp
  Then, apply the rule to any interface. For example:
  interface Ethernet0/0 ip inspect try in
  This configuration will activate the session rate timer to temporarily fix the problem.
  A missing H.323 ID configuration might cause the gateway to access (read) using a NULL pointer and might cause reloads on platforms where the service alignment global configuration command is a default, as it is on a Cisco 2600, 3600, 5300, and 7200 series router or 5800 series access server. A spurious read access to low memory will generate a syslog message instead of reloading the router. This situation can occur while the gateway is registering or unregistering from the gatekeeper while there is no H.323-ID configured.
  Workaround: Configure H.323-ID and ensure that the no gateway configuration command is entered first. Disabling the gateway prevents H.323 software from using a NULL pointer while the configuration process is not complete.
  CSCdr31528 A memory leak might occur when the interactive voice request (IVR) script calls for prompt/audio files that do not exist.
  Workaround: Ensure that all prompt/audio files are valid (and spelled correctly).
  A Cisco router might reload when interactive voice response (IVR) in a voice gateway requests authorization.
  Workaround: Enter the aaa authorization exec h323 group radius global configuration command when authorization is needed for voice gateway.
  Alternate Workaround: Upgrade to Cisco IOS Release 12.1(2) or Release 12.1(2)T.
  When you import routes to a Virtual Private Network (VPN) from another VPN at the same Provider Edge (PE), and the routes learned by Interior Gateway Protocol (IGP) at the originating VPN, and PE and Customer Edge (CE) are connected to each other by an unnumbered interface at the originating VPN, the router will fail to locate the next hop. As a result, routes do not appear in the VPN routing and forwarding tables (VRFs).
  Workaround: Use IP address on the interface between the CE and the PE.
  When you use a Cisco Encryption Service Adapter (ESA) to encrypt or decrypt, Cisco Express Forwarding (CEF) or fast switching might fail when configured on a subinterface. There is no workaround.
  Distributed Multilink (DML) PPP does not come up on a VIP4-80. VIP4 reloads and the controller goes into the shutdown state. There is no workaround.
  In some circumstances, Dual-tone Multifrequency (DTM) F relay might fail for the pound (#) and the star (*) symbols. There is no workaround.
  Configuring a policy map whose length is a multiple of 4 will cause the router to reload because of redzone corruption.
  Workaround: Do not configure policy maps with names that are 4, 8, 12, 16, 20,... characters long.
  On a Cisco router that is running Cisco IOS Release 12.1, dynamic crypto maps will refuse to negotiate IP Security (IPSec) Security Associations (SAs). An "! Incomplete" message will appear in the output of the show running-config command after each dynamic crypto map in the configuration. There is no workaround.
  A Cisco AS5300 access server store and forward fax might fail during off-ramp at Phase D where no pages are sent. There is no workaround.
  HTTP packets arriving over a Switched Multimegabit Data Service (SMDS) interface will not be correctly redirected by Web Cache Communication Protocol (WCCP), which will prevent web access by SMDS clients when WCCP is enabled.
  Workaround: Disable WCCP.
  When you configure an IP address on an Inter-Switch Link (ISL) encapsulated subinterface on a VIP2-50 Versatile Interface Processor using a Fast Ethernet PA-2FEISL-TX port adapter, the IP network configured on the subinterface does not show up as a directly connected interface in the IP routing table. As a result, there is no IP connectivity across the network.
  Workaround: Configure a static interface route for the directly connected network.
  Cisco Express Forwarding (CEF) is not building the subnet broadcast address receive adjacencies for locally attached broadcast networks. Hosts on the segment that use the "all the host bits set" (for example, if the network address is 1.2.3.0/24, then the "all the host bits set" is 1.2.3.255) or network address forms of IP broadcasts might fail to reach the router or to have broadcast packets forwarded (helpered) off the segment. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1(1) might reload if you enter the show snasw session detail command. There is no workaround.
  When a Hot Standby Router Protocol (HSRP) group on a Catalyst 5000 Route Switch Feature Card (RSFC) VLAN with the standby use-bia interface configuration command configured becomes active, the Address Resolution Protocol (ARP) entry for the HSRP virtual IP addresses is not updated with the local burned-in MAC address (BIA). It is overwritten by the BIA of the router which was previously in the "Listen" state. This situation does not occur if only two routers are participating in the HSRP group.
  This also affects other platforms when HSRP is configured on an encapsulated (ISL, dot1q etc.) interface.
  Workaround: Do not configure the standby use-bia interface configuration command.
  An SNA Switching Services (SNASw) router might run into memory fragmentation issues:
  R103PI02#sh mem Head Total(b) Used(b) Free(b) Lowest(b) Largest(b) Processor 61698980 43415168 38989864 4425304 2992172 9804 I/O 40000000 16777216 2894160 13883056 13758116 13802144
  The router might have access problems or might reload. There is no workaround.
  Open Shortest Path First (OSPF) does not work on the port channel interface.
  Workaround: Configure the ip ospf network broadcast interface configuration command under the port channel interface.
  When Cisco Express Forwarding (CEF) switching and Virtual Access interfaces are used, a Cisco 7200VXR router experiences serious problems with many alignment errors causing the CPU to reach 100 percent utilization. With an increasing number of users (virtual access interfaces), a production router will cease functioning instantly. There is no workaround.
  When dialing in with a modem on a Cisco AS5300 access server with a PRI, the performance of the AS5300 slows greatly. Using the show processes cpu EXEC command shows that the 5-minute average reaches 100 percent. The call management process listed in the show processes cpu command output is utilizing over 60 to 70 percent of the CPU. Reloading does not drop the call, nor does it drop the CPU utilization. Clear the utilization by using the shut and no shut commands on the group-async1 interface.
  Workaround: Configure the no async address dynamic command.
  A Cisco AS5200 access server that is running Cisco IOS Release 12.0(7)T and mica-modem-pw.2.7.1.0. is not able to dial out by reverse Telnetting into one of the MICA technologies modems. Dial-in works. There is no workaround.
  Cisco routers configured to use Stack Group Bidding Protocol (SGBP) might experience a buffer leak in large buffers.
  The memory leak might occur if a fairly large number of Multilink PPP (MLP) bundles are terminated on the SGBP member at the time the member enters or reenters the stack group.
  Evidence of the leak can be determined by monitoring the SGBP connection hello messages. These messages are enabled with the debug sgbp hellos command. If you see the following message and the size value is greater than 1360, a memory buffer will be leaked:
   SGBP:Send Info, count 1 size 7
  There is no workaround. Reload the router to recover I/O memory.
  Release-note Under sever stress the DBConn subsystem can cause the router to crash when the back-end host connection is reset unexpectedly. There is no workaround.
  When multiple, load-shared paths exist between provider edge (PE) routers, a PE router might reload if all paths are lost simultaneously while Virtual Private Network (VPN) traffic is being forwarded. There is no workaround.
  The clear cable flap-list command might cause the router to pause indefinitely.
  When a previous show cable flap-list command shows the flap list to be empty, the router will pause indefinitely after subsequently issuing the clear cable flap-list all command.
  When the output of the show cable flap-list command pauses due to automore processing (when "-- more --" is displayed) and another command line interface (CLI) EXEC session attempts to clear cable flap-list all, the router pauses indefinitely until the show cable flap-list command completes.
  Multiple concurrent CLI sessions that might modify data in the router are not supported by the system. The router might reload, pause indefinitely, or operate incorrectly.
  Workaround: Avoid using the clear cable flap-list all command.
  After the gateway resource threshold command is added, the router reloads.
  After entering the show version EXEC command, you might see the following error message:
  PH3-01C53.420284001008 uptime is 0 minutes System returned to ROM by bus error at PC 0x60B44944, address 0xD0D0D11 at 05:14:58 UTC Tue Mar 14 2000 System restarted at 05:15:32 UTC Tue Mar 14 2000 System image file is "flash:c5300-is-mz_120-7_T.bin
  Workaround: Remove the gateway resource threshold.
  The cookie value in the chassis of a Cisco AS5800 access server with integrated modems is programmed wrong. This error causes instability in environmental monitoring. There is no workaround.
  The upstream ports of a linecard might be poorly initialized and not receive upstream bursts from the cable modems. Modems are not able to complete initial ranging and power cycle infinitely.
  This situation only occurs if the upstream had nondefault configuration parameters stored in the startup configuration.
  Workaround: Configure the following sequence of cable interface configuration commands so that the upstream port can start receiving data if you wish to use nondefault upstream port physical layer configuration parameters:
  cable upstream {port} shutdown
  no cable upstream {port} shutdown
  Alternative workaround: Use default upstream physical layer configuration parameters.
  The PA-MC-8T1 port adapter reports that Firmware paused indefinitely and then reloads the router. There is no workaround.
  A gigabit interface may remain up/up with no cable attached in Cisco IOS Release 12.1(2). There is no workaround.
  Multiprotocol Label Switching (MPLS) labeled packets that are larger than 1500 bytes cannot be sent out FastEthernet even if "tag mtu" is configured to be larger than 1500 on the FastEthernet interface. There is no workaround.
  An SNA Switching Services (SNASw) router does not turn on the Command/Response (C/R) bit on the source service access point (SSAP) when replying to an exchange identification (XID) request.
  The trace between the SNASw router and the Point of Sale (POS) device shows that the C/R bit on the SSAP is not set when the SNASw router replies to the XID request. Thus, the POS device fails to begin the XID process. There is no workaround.
  Virtual Switch Interface (VSI)-controlled permanent virtual circuits (PVCs) supported by a PA-A3 port adapter on a Cisco 7200 series router do not come up after reload.
  Workaround: Clear (reset) the ATM interface.
  When connecting an LU6.2 session to a low-entry networking (LEN) device with the Systems Network Architecture (SNA) Switch, snasw location statements do not take effect properly when that LEN device is connected over a link that is a defined link from the SNA Switch perspective (hence an uplink). There is no workaround.
  TXCONN is causing SNA sessions to unbind when it terminates conversations.
  Impact SNA Switching Services (SNASw) must rebind the sessions again before another conversation can be allocated. This condition is inefficient and might, in certain configurations, cause allocation failures. There is no workaround.
  Upon system initialization, ATM permanent virtual circuits (PVCs) are left in the inactive state and do not change to the active state unless the user issues a shutdown/no shutdown of the associated ATM interface in configuration mode.
  Workaround: Reinitialize the interfaces manually.
  A router might reload when using a TCP server on a unique port, which might happen when an internal ping runs to test connectivity with the host.
  Workaround: Configuring another DBConn TCP server on the same port. Two separate DBConn servers configured on the same port disables the internal ping mechanism.
  On a Cisco 2600 or 3600 series router or a Cisco MC3810 multiservice access concentrator, CAS-CAS (Channel-Associated Signalling) and CAS- ISDN calls using images prior to Cisco IOS Release 12.1(3)XI might have wrong PI values in the ch323_h225_send_setup function. Instead of PI being equal to 3, PI is equal to 0. There is no workaround.
  When polling Cisco 2600 series routers in the network for Hot Standby Router Protocol (HSRP) statistics, the routers reloads.
  Workaround: Disable Simple Network Management Protocol (SNMP).
  Voice calls fail when going to multiple cable modem termination systems (CMTSs) on a Cisco uBR924 cable access router that is in DOCSIS-bridge mode instead of IOS-router mode.
  Workaround: Configure the Cisco uBR924 cable access router as a simple router, using static routes.

TCP/IP Host-Mode Services

  Remote source-route bridging (RSRB) data gets corrupted when the ip tcp path-mtu-discovery global configuration command is configured and the IP data path changes from a larger maximum transmission unit (MTU) interface to a smaller MTU interface because of IP routing changes.
  Workaround: Disable the Path MTU discovery algorithm by using the no ip tcp path-mtu-discovery command.
  Under rare circumstances, a Cisco router that is running Cisco IOS Release 12.1(1.4) might reload because of a bus error while processing a TCP timeout.
  Workaround: Disable selective-ack by using the no ip tcp selective-ack global configuration command.
  TCP header compression on a Cisco 7200 series router that is acting as an Layer 2 Tunneling Protocol (L2TP) Network Server (LNS) causes substantial performance degradation.
  Workaround: Disable TCP header-compression.

Wide-Area Networking

  A Cisco AS5300 access server using Group-Async Interfaces and Virtual Templates might terminate dialup links when Link Quality Monitoring (LQM) Link Quality Reports (LQRs) are received. This situation occurs when the local router is configured for LQM, but the remote side is not. There is no workaround.
  Frame Relay Local Management Interface (LMI) Status Enquiry (StEnq) packets are sent out on a D channel and are received by the data communications equipment (DCE) as unnumbered information (UI) packets. The DCE does not respond to the packets and records LMI timeouts. Eventually, the DCE indicates that it is down. The pings succeed until Frame Relay marks the line down because of the LMI timeouts. There is no workaround.
  Dialer profiles do not enforce the use of the correct dialer remote name for outgoing calls and for incoming calls binding before PPP authentication. However, once dialing and authentication occurs, the router does not terminate the connection as it normally does if the authenticated name is different than that on the dialer profile.
  Cisco IOS software does not enforce that the host name presented by a called peer must actually match the remote name configured on the selected dialer profile. Similarly, incoming calls that bound by a means other than the presented host name (such as dialer caller or dialer called) will not be checked for consistency of the presented name to the remote name configured on the selected dialer profile. This situation is not a security issue because the presented host name and password are authenticated in the normal manner. The presented username is not checked for consistency with the remote name configured on the bound dialer profile. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)T with a Dialer interface configured with PPP encapsulation might experience a memory leak in the output dialer interface. The output of the show buffers old command will have dialer in the output. A mallocfail error in input/output (I/O) memory might appear.
  Workaround: Use Multilink PPP.
  Alternate Workaround: Ensure that fast switching is occurring.
  In violation of Q931 of the International Telecommunication Union Telecommunication Standardization Sector (ITU-T), a Cisco router might not respond to RESTART messages from a basic-net3 routing switch. This situation might result in calls not being placed or accepted over the channel or interface by the originator of the RESTART message. There is no workaround.
  On a Cisco 800 series router that is running Cisco IOS Release 12.0(7)T, plain old telephone service (POTS) flash functionality does not work for NI switchtype, although it does work for 5ESS switchtype. There is no workaround.
  When you use the debug isdn event EXEC command, the ISDN debugging messages use all CPU resources, and the Double-Density Modem (DMM) card and the Hex Modem Module (HMM) card become inaccessible. This situation only occurs when you use the debug isdn event EXEC command, which can be easily disabled. There is no workaround.
  A Cisco router might reload with the following error when the frame-relay qos-autosense and frame-relay traffic-shaping interface configuration commands are configured:
  System was restarted by error - an arithmetic exception.
  Workaround: Disable frame-relay qos-autosense and configure the shaping parameters manually instead.
  The D channel to B channel conversion will not properly set the initial service state. There is no workaround.
  A Cisco 3660 router running Cisco IOS Release 12.1 might experience %ALIGN-3-SPURIOUS messages while booting. There is no workaround.
  A Cisco 5300 series router that is running Cisco IOS Release 12.1(1)T and that is configured with Octal PRI cards that are running multiple Non-Facility Associated Signaling (NFAS) groups experiences an ISDN memory leak (the line interface (LIF) timers constantly increment before reloading). There is no workaround.
  In-depth technical analysis:
  The memory leak occures when stanby D channel is in AWAITING_ESTABLISH state and there is SETUP request on active D channel and the assigned B channel is one of B channels in standby D channel controller.
  In normal operation, both active and standby D channels should be in MULTIPLE_FRAME_ESTABLISHMENT state.
  The ppp callback [accept | permit] noverify interface configuration command breaks a per-user configuration when you perform callback. There is no workaround.
  Two CAUSE information elements are sent in a STATUS message. There is no workaround.
  The show x25 xot EXEC command might trigger a reload. The system returns to ROM by an Illegal Instruction error. There is no workaround.
  The Catalyst 5000 2-port ATM module with OC-12 support or 4-port module with OC-3 or OC-12 support running Cisco IOS Release 12.1(1.1) will reload whenever the card is configured to have a LAN Emulation Configuration Server (LECS) and is placed in an environment that has another LECS of lower priority operational.
  Workaround: If the LECS of lower priority is shut down or removed, the reload does not occur.
  If a router is running Cisco IOS Release 12.1(2) and a Cisco IOS Release earlier than 12.1(2), the two versions of Layer 2 Tunneling Protocol (L2TP) might drop tunnels because of a control message parse error:
  Tnl 5 L2TP: Parse AVP 1, len 8, flag 0x0x8000 (M) Tnl 5 L2TP: Result Code AVP too short, expect min 10, got 8 Tnl 5 L2TP: Error processing Mandatory AVP 1
  There is also an incompatibility with non-Cisco IOS L2TP implementations that may send an 8-byte Result Code.
  This problem occurs when running Cisco IOS releases which have CSCdp68265 or CSCdr42451 with those that do not have either CSCdp68255 or CSCdr42451 present. The following table indicates when these two items were checked into IOS.
  DDTS              Interim Build     Maintenance Release
  CSCdp68265       12.1(1.1)                    12.1(2)
  CSCdr42451        12.1(2.1)                    12.1(3)
  Thus, this problem occurs when running Cisco IOS release 12.1(2) with releases prior to 12.1(2). The problem will not occur with Cisco IOS releases 12.1(2) and itself or later releases. There is no workaround.
  When an inactive ("not currently active") D channel of a Non-Facility Associated Signaling (NFAS) group fails or is shutdown, attempts to restart the digital subscriber line (DSL) might cause high (98 percent or more) CPU utilization and "CPUHOG" messages. Reestablishment of the D channel corrects the problem. There is no workaround.
  Using PPP or SLIP EXEC might cause the router to reload when SLIP or PPP are entered on a line that does not support it. Entering PPP at the router prompt also causes the router to reload. There is no workaround.
  On a Cisco router running Cisco IOS Release 12.1(3), the LAN Emulation Clients (LECs) fail to send LE_ARP responses if the LAN Emulation (LANE) interfaces form part of a bridge group with routing protocols enabled and are in the blocking state. Routed packets addressed to the router's own interfaces might be dropped. There is no workaround.
  The ISDN Switchtype is not set on BRI S/T and U interfaces on BRI WAN interface cards (WICs). There is no workaround.
  The following messages are generated for every call disconnect with switchtype primary-ni configured:
  ISDN ERROR: Module-l3_sdl_u Function-Ux_BadMsg Error-Source ID = 400 Event = AB
  There is no workaround.

Resolved Caveats—Release 12.1(2)

All the caveats listed in this section are resolved in Release 12.1(2). This section describes only severity 1 and 2 caveats.

Access Server

  A Cisco AS5300 series access server might report Call Fail on successful calls when modem allocation is VDEV_ALLOCATE_ALMOST_READY. There is no workaround.

AppleTalk

  A Cisco AS5300 access server that is running Cisco IOS Release 12.0(4.4)T with both IP and AppleTalk might experience difficulties with AppleShare. If you use a laptop that is configured for remote access through AppleTalk Remote Access Protocol (ARAP) using MAC-IP protocol, the access server will authenticate access, but will fail to find a file server in any zone when you select the "chooser" in AppleShare, even though you will be able to see all the zones. There is no workaround.

Basic System Services

  Under a sustained heavy load of calls, authentication, authorization, and accounting (AAA) requests backup, using up memory until memory runs out and the network access server reloads.
  Workaround: Upgrade to a Cisco IOS release that contains the fix for this caveat.
  A Cisco 7200 series router with a compression service adapter (CSA) that is configured with Frame Relay Traffic Shaping (FRTS), and using either Cisco Express Forwarding (CEF) or fast switching will not compress data when the amount of data on the interface is greater than the shaping value (for example, CSA will stop compression when traffic shaping becomes active).
  Workaround: Use process switching.
  If you enable IP route-caching (IP fast-switching) on a PPP serial interface that is part of a multilink bundle, traffic that is destined for that bundle might stop.
  Workaround: Shut down the serial interfaces before entering the ip route-cache command, and then enable the interfaces.
  A Cisco 7200 series router with an NPE-300 network processing engine installed might not boot up when certain Cisco IOS Release 12.0(5)XE3 subset images are installed. The router will pause indefinitely in the early stage of booting up, and a power cycle is required to resume. For systems set for auto boot, you will need to enter the break command to abort the boot process and break out to the ROM monitor before the 12.0(5)XE3 image is launched for execution. You will then need to either modify the software configuration register to revert to a manual boot of some other known good image, or you will need to switch the PCMCIA flash card with a known good image in case the system is set for a default image boot from the slot0: PCMCIA card. There is no workaround.
  While trying to get a core dump, if there is another reload, a Cisco router might only partially generate the core file. This situation occurs when there are double reloads. There is no workaround.
  HC (64-bit) counters need more accuracy over short polling cycles.
   Workaround: Shorten the polling interval for HC counters to achieve the required accuracy. Poll-on-demand can be used, but is resource-intensive.
  Establishing an autocommand telnet session to the second host in the autocommand telnet generates a Carriage Return and LineFeed (CR/LF) after receiving a connection reset (RST) from the first host that is down on vty. This occurs if the global configuration command service hide-telnet-addresses is configured on the router, together with the command busy-message.
  Workaround: To avoid getting the extra CR/LF, remove the global configuration command service hide-telnet-addresses.
  A Cisco AS5800 series access server, a Cisco 7200 series router, a Cisco UBr7200 universal broadband router, or a Cisco 6400 series router might experience a memory leak. If an IP packet needs fragmentation in the fast-switching path and the packet has additional data bytes tagged on beyond the length indicated in the IP header, there might be a leak in the input/output (I/O) memory of the system. The memory leak will depend on the size of the original unfragmented packet and on the size of the extraneous data.
  Workaround: Disable fragmentation in the fast-switching path by using the service disable-ip-fast-frag global configuration command.
  A Cisco router might reload because of memory corruption or stack overflow if you use the show ip bgp regexp pattern privileged EXEC command.
  Workaround: Simplify the regexp pattern.
  A voice gateway might reload when a voice call disconnects if you use the gw-accounting [h323 | syslog] global configuration command and you do not properly configure the authentication, authorization, and accounting (AAA) accounting method list for voice.
  Workaround: Use the aaa accounting connection h323 {start-stop} [radius | tacacs+] global configuration command.
  On a Cisco 3810 router that is running Cisco IOS Release 12.(07)Twith a private branch exchange (PBX), if you place a call to a Cisco 3810 router with Voice over Frame Relay (VoFR), the called phone rings immediately but the ringback is heard about 10 seconds later on the calling phone. Phone calls placed in the other direction work properly. There is no workaround.
  Configuring a large number of Virtual Private Networks (VPNs) on a Cisco 7500 series Route Switch Processor (RSP) with a large number of channelized interfaces might result in a FIBDISABLE message. This message indicates that the RSP has not received a FIB keepalive from the line card within the expected length of time. When this situation occurs, the RSP functions as if the interprocess communication (IPC) mechanism has failed and disables Cisco Express Forwarding (CEF) on that line card.
  Workaround: Disable distributed switching.
  A Cisco 7500 series router that is configured with a High-Speed Serial Interface (HSSI) port adapter will experience a Virtual Interface Processor (VIP) reload when distributed software compression or hardware compression is configured. When you update the image, the bootloader also needs to be updated, or router will experience the same reload at boot time.
  Workaround: Disable compression or use software compression.
  Snmpboots, a boot counter for SNMP Version 3, is incremented and saved during bootup, which might cause a noticeable bootup delay. This bootup delay will only occur when SNMP Version 3 is configured. There is no workaround.
  A Cisco 3810 router that is running Cisco IOS Release 12.0(4) or a later release might generate the following error message while sending a call back in the direction that it came from or while making phone calls on the router:
  %VOICE_FSM-3-ERROR: Fatal Error: st(10)/ev(2) out of range, ../voice/eecm_fsm.c, 2357
  Functionality is not affected. There is no workaround.
  When a Cisco 5300 series network access server (NAS) is configured with Resource Pool Manager Server (RPMS), sustained call traffic might cause the NAS to deplete usable memory.
  Workaround: Upgrade to Cisco IOS Release 12.1(2) or 12.1(2)T..
  When you upgrade a Cisco AS5300 series router using Simple Network Management Protocol (SNMP), SNMP will use a program that erases Flash instead of boot flash memory and copies the file to boot flash memory. There is no workaround.

EXEC and Configuration Parser

  On a Cisco router that is running Cisco IOS Release 12.1(0.8), priority queueing cannot be configured. If you try to enter the priority group list-number interface configuration command on an interface, the router will exhibit the following error message:
  % Ambiguous command
  There is no workaround.

IBM Connectivity

  A list element that should be displayed in the output of the show lnm station privileged EXEC command or the show lnm ring privileged EXEC command might have been deleted from the list because it no longer exists in the ring. If you leave the command waiting on a "-- More --" prompt for a long time, this list might change to such an extent that an attempt to print an element that existed previously but now has a NULL pointer might cause the router to reload. This situation rarely occurs and the following workarounds are recommended to reduce the possibility of the router reloading.
  Workaround: Set the term len to 0 before using the show lnm station privileged EXEC command or the show lnm ring privileged EXEC command.
  Alternate Workaround: Use the show lnm station privileged EXEC command or the show lnm ring privileged EXEC command after the ring has stabilized.
  A Cisco 2500 series router might suddenly stop receiving or not forward packet frames while transferring files over data-link switching plus (DLSw+) link using Fast Sequenced Transport (FST) encapsulation. This situation causes the NetBios session to disconnect. There is no workaround.
  A Cisco 3620 router might pause indefinitely if it is hit with an excessively large number of Layer 2 packets while transparent bridging is enabled. There is no workaround.
  A Cisco router that is running Cisco IOS Release 11.2(15a) or Cisco IOS Release 11.2(15a)P might reload with a bus error. There is no workaround.
  A Cisco router might reload with a SegV error if you attempt to configure the SDLC interface with the sdlc address hexbyte [xid-passthru] interface configuration command.
  Workaround: Enter the shutdown interface configuration command on the interface before making the change, and then enter the no shutdown interface configuration command after you have changed the configuration.
  LAN Net Manager might cause memory leaks on a Cisco 2600 series router. In this situation, several buffers are held by LAN Net Manager in LLC-CIRCUIT-DYNAMIC buffer types. Telnet might not be able to reach the router after 24 hours, and the router might reload after three to four days.
  Workaround: Disable LAN Net Manager in the configuration, and reload the router.
  In a data-link switching (DLSw) environment with DLSw Version 2 configured between a Cisco router and an IBM router, the Cisco router might have difficulty reestablishing a DLSw peer on demand after the peer has timed out because the Cisco router flushes all reachablity information for a peer when the peer disconnects. When a device reestablishes a connection over the same DLSw path, the IBM router sends a tcp sync message to bring up the DLSw peer on demand. The Cisco router rejects this connection because it has no information on how to reach the peer or the circuit. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)T might experience a memory leak with "IP Input" appearing as the process that continues to hold memory. This situation appears to be related to having data-link switching (DLSw) configured. There is no workaround.
  Data-link switching (DLSw) circuits for NetBIOS traffic will not be established, which will break all NetBIOS connectivity using data-link switching plus (DLSw+). This situation only occurs in Cisco IOS Release 12.1(1.1) or a later release. There is no workaround.

Interfaces and Bridging

  CT3/CE3 port adapters on a Cisco 7200 series router might drop TX packets under bursts of heavy traffic instead of putting them in a hold queue if the number of outstanding transmit packets temporarily exceeds the number specified by the TX limit. There is no workaround.
  Distributed switching from a Packet-Over-SONET (POS) port adapter to a Channelized T3 port adapter might fail.
  Workaround: Use Route/Switch Processor (RSP) Cisco Express Forwarding (CEF) switching.
  When integrated routing and bridging (IRB) is enabled and a Bridge-Group Virtual Interface (BVI) is configured on the IRB, Banyan VINES packets generated by a serial interface get dropped by the receiving routers because of the encapsulation for Banyan VINES packets that are sent by the router configured for IRB. There is no workaround.
  A Cisco 2600 router that is configured with a token ring interface or a Cisco 7200 or 7500 series router that is configured with a PA-4R-DTR port adapter token ring interface that is running Advanced Peer-to-Peer Networking (APPN) with Cisco Link Services (CLS)-managed LLC2 logical link control connections might fail host link activation with sense codes 10160007 or 10160022. A router with PA-4R or PA-4R-FDX port adapters is not affected.
  Workaround: Enable source route bridging on the affected token ring interface.
  Currently when the color of an encapsulation is changed on a sub-interface the sub-interface does not transition through the spanning tree states. In some situations this could result in spanning tree loops. In order to prevent this in case the subinterface color needs to be changed, delete the subinterface and recreate it with the new encapsulation color.
  A Cisco AS5800 universal access server with E1 PR1 lines that is running Cisco IOS Release 12.0(4)XJ2 or Cisco IOS Release 12.0(5)T1 might drop Serial Line Internet Protocol (SLIP) connections after Cisco IOS installs a route for a remote user. PPP users (both asynchronous and ISDN) can connect, but the SLIP users cannot connect with MICA modems that are running 2620 portware. There is no workaround.
  A Cisco PA-A1 port adapter does not automatically switch to the internal clock during LOS (Loss of Signal) condition. There is no workaround.
  In Cisco 7100 and 7200 series routers, transparent bridging using the PA-DTR Token Ring Port Adapter does not operate properly. The show bridge command fails to indicate any of the end-stations in the bridge table.
  Workaround: On the Cisco 7200 series routers, use alternative token ring port adapters. There are no known workarounds for the Cisco 7100 series routers.
  A Cisco 1601 router that is running Cisco IOS Release 12.0(7) with Frame Relay FRF.9 compression and weighted fair queueing (WFQ) might experience a failure of the fair_enqueue function that results in fatal reentrant error messages and a software-forced reload.
  Workaround: Remove WFQ from Frame Relay interfaces.
  When a channel on a CT3/CE3 port adapter is continually overstressed by traffic, other unstressed channels might experience some transmit packet drops. There is no workaround.
  On a Cisco router, access lists in the process path of the bridging code that handles output pattern lists might not work. There is no workaround.
  When a bad transmit packet is generated and sent to the Channelized T3 (CT3) interface, the packet might cause the address of the transmit queue accumulator (txacc) value to not increment correctly for the CT3 interface. In this situation, the output eventually becomes stuck when the txacc value reaches zero.
  Workaround: Configure the CT3 interface with the tx-queue-limit 5 interface configuration command to restore the txacc value for the effected CT3 interface.
  Configuring an Async interface in any Cisco 7500 series Route Switch Processor (RSP) will prevent the proper parsing of interface name for CT3/CE3 port adapters.
  Workaround: Deconfigure any Async interface, and then write the configuration to NVRAM and reload the router. Or, you can move the VIP from slot 0 to another slot.
  With IP Cisco Express Forwarding (CEF) and VLANs configured, the router might reload with a bus error when you use the configure memory privileged EXEC command. There is no workaround.

IP Routing Protocols

  IP access lists always permit IP fragments. There is no workaround.
  A Cisco router may experience a bus error and reload when you enter the ip accounting output-packets command. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(5)T and later releases, Cisco IOS Release 12.1, or Cisco IOS Release 12.0 ST might reload if the clear ip bgp {*} EXEC command is entered on a peer provider edge (PE) router. The following stack trace is exhibited:
  bgp_fwdentry_info
  bgp_v4class_update_fwdtable_walker
  rn_walktree_version
  bgp_update_fwdtable bgp_router
  The reload occurs on the local PE when a PE Internal Border Gateway Protocol (IBGP) session is cleared on the remote box. If the PE is importing routes from other PE devices, clearing BGP session on the remote PE will cause the local PE to reload. There is no workaround.
  Multiprotocol Label Switching (MPLS) traffic engineering (TE) tunnels might bounce under certain conditions if there is a very heavy traffic load on a Gigabit Switch Router (GSR) line card. This condition has been observed when a routing loop is present and the card is generating Internet Control Message Protocol (ICMP) "TTL expired" messages, and when the card is used as the data sink for ICMP ECHO requests from a traffic generator. There is no workaround.
  When the Protocol Independent Multicast (PIM) designated router (DR) changes, the F flag is not changed. This situation might then cause header registers either not to be sent to the Route Processor (RP) or to be duplicated. There is no workaround.
  When the Incoming Interface List (IIL) changes for a (S,G) state, the F flag is not updated. This situation might cause registers to be sent for sources that are not directly connected anymore. There is no workaround.
  A Cisco router that is acting as an area border router (ABR) connected to a stub area might place an Open Shortest Path First (OSPF) external route into its routing table whose outgoing interface is in the stub area. This situation can cause a routing loop because the stub area routers do not know about the external route and will forward packets along their default route, which might be the ABR.
  This problem has been found to occur only when there are 2 paths between the stub area ABR and the autonomous system border router (ASBR) that is originating the external route, where one path is through the stub area and the second is through a nonstub area(s).
  Workaround: Set the link costs within the stub area high enough so that the stub area ABR routers do not see the path to the ASBR as being as good as the path(s) through the nonstub areas.
  When Network Address Translation (NAT) is configured, a Cisco router might assign the same inside global address for multiple inside local addresses.
  Workaround: Enter the clear ip nat translation {*} EXEC command.
  When a Multicast Routing Monitor (MRM) Test Sender is instructed by an MRM Manager to send test packets out of all interfaces that are configured for multicast routing, which is the default option, the MRM might experience a leak in the small buffers.
  Workaround: Configure the MRM Manager with the senders {access-list-number | access-list-name} [target-only] command.
  You cannot specify a Bridge Group Virtual Interface (BVI) in a Network Address Translation (NAT) inside source list. There is no workaround.
  In a Protocol Independent Multicast (PIM) Version 2 router configuration, if a hash mask is not configured on a Cisco bootstrap router (BSR), the router takes the first rendezvous point (RP) address from the local RP-mapping cache. The RP-mapping cache is not sorted, so when RPs are added and deleted from the cache, there might be inconsistency through the BSR domain that causes routers to choose different RPs for the same group.
  Workaround: Create a hash mask length of 1 on the BSR router by entering the ip pim bsr-candidate Ethernet1/2 1 global configuration command.
  Border Gateway Protocol (BGP) might cause a Cisco router to reload when using MED for best path selection. There is no workaround.
  A Cisco router that is running Cisco IOS Release 11.2(21)P or Cisco IOS Release 12.0(7)T might experience a memory leak in the dead process when you configure IP Network Address Translation (NAT) by Hypertext Transfer Protocol (HTTP) server using the static ip nat inside source global configuration command, if the global IP address does not belong to any directly attached network. You can verify this situation by using the show memory dead command.
  Workaround: Configure an interface (such as a loopback interface) with a network IP address including the global IP address.
  A Cisco RSP4 Route Switch Processor might reload if the managed timer does not work properly. This situation rarely occurs. There is no workaround.
  When you change the Reverse Path Forwarding (RPF) interface for (*, G) and (S, G), the interface that was in the outgoing interface list (O-list) becomes the new RPF interface. At this point, the new RPF interface will be deleted from the O-list, which becomes NULL. When the router switches back to the original RPF interface, the O-list remains NULL until the next Internet Group Management Protocol (IGMP) report. While the O-list is NULL, and the original RPF interface has directly connected members, the CONNECTED flags are deleted and the router switches back to the shared tree by sending (S, G, RPT) Join messages. This situation causes a delay in the convergence time. There is no workaround.
  A Cisco router might reload because of a bus error. There is no workaround.
  If there is a link flap somewhere in the network between the area border router (ABR) and an autonomous system boundary router (ASBR), the ABR might not generate a type 4 summary ASBR link-state advertisement (LSA) to other areas after the link is restored. The net effect is that routes being redistributed by the ASBR into Open Shortest Path First (OSPF) will not be installed in the routing tables in the affected areas.
  Workaround: Restart OSPF on the ABR by using the clear ip ospf proc command.
  Alternate Workaround: On the ABR, restart OSPF for the affected areas only by removing and restoring the network statements under the router ospf global configuration command for the impacted areas.
  Alternate Workaround: For this workaround, perform the action only after the subject ASBR LSA has been removed from the database of the affected areas (no longer seen in the show ip ospf database EXEC command).
  On the affected OSPF routers (that are not seeing the routes and the ASBR LSA) adjacent to the ABR, reestablish adjacencies with the ABR. One way to temporarily change the hello-interval to some other value. After the adjacency is taken down, change the hello-interval back to the original value to reestablish the adjacency. This action causes the ABR to regenerate and re-send the LSAs. On the ABR, create and temporarily remove a router ospf global configuration command (for example, router ospf 1234 and no router ospf 1234).
  On a Cisco 7500 series router, many Virtual Private Networks (VPNs) configured in combination with a large number of channelized interfaces might result in a FIBDISABLE message.
  The FIBDISABLE message indicates that the Route Processor (RP) has not received a Forwarding Information Base (FIB) "keepalive" in the expected amount of time from the line card. When this situation occurs, the RP acts as if the interprocess communication (IPC) mechanism has malfunctioned and disables Cisco Express Forwarding (CEF) on that line card.
  Workaround: Disable distributed switching.
  A Protocol Independent Multicast route processor (PIM RP) might discard the data-header register from the first-hop Cisco routers and display a LOG_NOTICE level (5) syslog message. It is rate-limited to no more than once every 5 seconds.
  The syslog exhibits the following message:
  %PIM-5-REG_ENCAP_INVALID: Bad register from 172.16.2.2 for (172.16.2.5, 224.1.1.1). Trace = 14 0 0 -Process= "PIM Process", ipl= 0, pid= 48
  In addition to this syslog message, the RP ignores the data-header registers so the first-hop routers send a full register with encapsulated user IP packet about once every 3 minutes if the source stays active. This situation might result in a transient packet falling out of sequence because the register packet with encapsulated data is handled at the process level. This situation might also cause the RP that is configured with Multicast Source Discovery Protocol (MSDP) to send encapsulated data to the service adapter (SA) with excess frequency.
  Workaround: Upgrade to Cisco IOS Release 12.0(9.6)S2 or a later release.
  Due to new RFC 2328, the calculation of summary route costs has changed. This situation might create suboptimal routing if all of the area border routers (ABRs) are not upgraded to the new code at the same time.
  Workaround: Upgrade all of the ABRs to the new code.
  When prefixes learned from Customer Edge (CE) devices (either through Border Gateway Protocol (BGP) or Interior Gateway Protocol (IGP) and redistributed into BGP) get into the Virtual Private Network (VPN) routing and forwarding tables (VRFs), "route-target export" will always be applied if configured.
  If export-map is also configured and if it has the set extcommunity RT1 RT2 command, then all the previous RTs are replaced with RT1 RT2.
  If however, the export-map has the set extcommunity RT1 RT2 additive command, then the previous RT list is retained and RT1 and RT2 are added to the set, resulting in a union. There is no workaround.

ISO CLNS

  If a Connectionless Network Service (CLNS) neighbor has multiple associated area addresses, then a Terminal Identifier Address Resolution Protocol addresses, then a Target Identifier Address Resolution Protocol resolve request will automatically choose the first entry. If the first entry is in the process of being replaced by the second area, then the Target Identifier Address Resolution Protocol request will fail.
  Workaround: Clear the CLNS cache.
  A Cisco 3660 router might experience a bus error at PC 0x6089E720, address 0x20340008 and reload after the show run command is entered twice consecutively or a show tech command is entered from the enable mode. This situation occurs with routers that have the following lines in their configuration:
  tarp run
  tarp route-static 39.840E.8003.3141.0000.0000.0000.00e0.c15c.0000.00 message type 1 2 4
  There is no workaround.

Miscellaneous

  A crypto Access Control List (ACL) that has a DENY ACE that specifies a TCP or User Datagram Protocol (UDP) port might cause fragments to be dropped or forwarded in the clear.
  Workaround: Arrange the crypto ACLs to have permits only when specifying ports. For example, instead of configuring:
  access-list 101 deny udp 200.200.20.0 0.0.0.255 200.200.30.0 0.0.0.255 eq 19 access-list 101 deny udp 200.200.20.0 0.0.0.255 eq 19 200.200.30.0 0.0.0.255 access-list 101 permit udp 200.200.20.0 0.0.0.255 200.200.30.0 0.0.0.255
  configure:
  access-list 101 permit udp 200.200.20.0 0.0.0.255 ne 19 200.200.30.0 0.0.0.255 ne 19 access-list 101 deny udp 200.200.20.0 0.0.0.255 200.200.30.0 0.0.0.255 eq 19 access-list 101 deny udp 200.200.20.0 0.0.0.255 eq 19 200.200.30.0 0.0.0.255
  The Dynamic Host Configuration Protocol (DHCP) proxy allows the same IP address to belong to two users on different ports that have the same user name.
  Workaround: Ensure that all users have unique user names.
  On a Cisco AS5800 series access server, you might see the following messages:
  %DIALn-3-MSG: %IPC-3-NOBUFF: The main IPC message header cache has emptied
  These messages are followed by a "-Traceback" line. The %DIALn-3-MSG is optional and the "n" indicates which dial shelf slot number the subsequent message originates from. Usually, the latter message is followed by:
  %DSIP-3-IPC_PORT: IPC subsystem API error (nip invoke ipc send), failed to get ipc_message
  This situation usually occurs when you use the autoselect ppp line configuration command, or the autoselect slip line configuration command, or the autoselect arap line configuration command. Under a high call rate, this situation might cause calls to not be established successfully. There is no workaround.
  A Catalyst 5000 series router might continually reboot and write crashdumps to boot flash memory that subsequently use up available space in boot flash memory when you netboot in the presence of traffic. When boot flash memory is out of space, you cannot boot from boot flash memory. In order to bring up the router, you need to use the download command from the Supervisor. Once the router is booted, you cannot display boot flash memory or delete files on boot flash memory.
  Workaround: Boot from Flash with the configuration register set to 0x101.
  Alternate Workaround: Disable the crashdump collection.
  A Cisco 7200 series router with an 8-port channelized E1 card running Cisco IOS Release 12.0(4)T might experience a memory leak in the small buffer pool.
  Workaround: Monitor the input/output (I/O) memory and reload the router when it is too low.
  The interface delay metric is set incorrectly for port channel interfaces where one or more Gigabit Ethernet interfaces are grouped into a channel. The delay for a single Gigabit Ethernet interface is 10 microseconds. The delay for a port channel made up of one or more Gigabit Ethernets is 100 microseconds. The incorrect setting might seriously impact routing protocols that use interface delay as part of the metric—for example, Enhanced Interior Gateway Routing Protocol (EIGRP), and might cause the routing protocol to take a route through a single interface over a route through a port channel.
  Workaround: Manually configure an appropriate delay under the port channel interface by entering the delay tens of microseconds interface configuration command.
  Writing to an AT Attachment (ATA) device might cause the device to become unusable and result in the following error message:
  ATA_Status time out waiting for 1
  There is no workaround.
  A Cisco 1720 router that is running Cisco IOS Release 12.0(5)T might reload with the following error message if you shut down a serial interface while packets are being routed to a dialer interface:
  System returned to ROM by error - a SegV exception, PC 0x802Ac04C
  There is no workaround.
  In a Channel Associated Signalling (CAS) environment, a DS0 might become unavailable if the line is seized for an outgoing call and there is no response from the remote end. In this situation, the Cisco AS5300 access server is configured to establish the maximum number of outgoing calls possible for the number of DS0s in the system. Each unavailable DS0 will result in the following message:
  no ds0 available for modem
  There is no workaround.
  A Cisco 7200 series router configured with 250 ATM Address Resolution Protocol (ARP) clients might reload during a stress testing. There is no workaround.
  A Cisco 1700 series router and a Cisco 2600 series router that are running Cisco IOS Release 12.0(5)T cannot start ISAKMPA SA using the rsa-encr authentication method. The debug output reports that "Hash payload is incorrect!" There is no workaround.
  A Cisco router might reload in a Context-Based Access Control (CBAC) configuration. There is no workaround.
  When two routers are connected back to back using PA-2FE cards, when you use the shut command on one interface of a 2FE, the line protocol on the other end does not go down. However, physically removing the cable at one end causes the line protocol on the other end to go down. There is no workaround.
  If an output rate limit is configured on a non-Versatile Interface Processor (VIP) interface (for example, AIP or FIP) on a Cisco 7500 series Route Switch Processor (RSP) with Cisco Express Forwarding (CEF) enabled, packets cannot be switched out of that interface.
  Workaround: Disable CEF.
  Forwarding of bootp/dhcp address request UDP packets fail because of encapsulation failure.
  Workaround: Use Cisco IOS Release 12.0(5)T.
  A Versatile Interface Processor (VIP) might not forward packets to the Route Switch Processor (RSP). All receive (RX) buffers are in use but the output interfaces do not hold these buffers. As traffic stops reaching the Route Switch Processor (RSP), the routing protocol session goes down and the Tag Distribution Protocol (TDP) session is lost. There is no workaround.
  A node route processor (NRP) ATM interface stops sending when there are multiple particles with data-length 0 at the last particle.
  The only way to exit this situation is to use the shut command followed by the no shut command. There is no workaround.
  Sensitive customer information (such as telephone numbers) might be available at privilege level 1, which is the default for a nonenabled user.
  Workaround: Configure the default privilege level to be 0 by entering the privilege level 0 line configuration command.
  If Internet Key Exchange (IKE) is overloaded with many tunnels and keepalives, IKE can no longer keep up with the traffic and tunnels time out.
  Workaround: Either reduce the number of tunnels or increase the keepalive time. Adapt keepalive time to the overall number of tunnels so that the more tunnels you have the more you increase the keepalive time.
  Pings on a Cisco 7500 series tag-switching router that is running Cisco IOS Release 12.0 S, Cisco IOS Release 12.1, or Cisco IOS Release 12.0 ST might fail if a packet comes in as an IP packet on an ATM interface, the packet gets route-cache switched, the packet gets fragmented, and the fragments go out as tag packets through a serial interface.
  Workaround: Configure distributed Cisco Express Forwarding (dCEF) on a Cisco ATM port adapter.
  Voice packets are dropped and consequently voice quality is degraded with Real Time Protocol (RTP) compression over Frame Relay.
  If one end of the Frame Relay link is running Cisco IOS Release 12.0(7)T and the other end is running a previous release, a mismatch occurs in the number of compression context slots (256 for Cisco IOS Release 12.0(7)T and either 16 or 32 for the previous release). This situation results in invalid context numbers being received by the older release and causes data loss.
  Workaround: Run Cisco IOS Release 12.07T on both ends.
  Alternate Workaround: upgrade one end to Cisco IOS Release 12.1(2) and configure the number of slots to match the other end.
  On Cisco routers that are running Cisco IOS Release 12.0(8), IP Security (IPSec) will send in the clear packets that need to be encrypted and fragmented. There is no workaround.
  Multiprotocol over ATM (MPOA) shortcuts might not check for the frame sizes being sent on them. If the frame size exceeds the maximum transmission unit (MTU) size of the egress Multiprotocol over ATM Client (MPC), egress MPC drops the packet, resulting in the loss of connectivity.
  Workaround: Configure matching MTUs on either side of shortcuts (for example, all LAN Emulation Clients (LECs) that are served by ingress and egress MPCs should have the same MTU). If MTUs cannot be altered, then do not configure MPOA on both sides.
  The ip rtp priority interface configuration command does not work in Cisco IOS Release 12.1(09). There is no workaround.
  When fast switching an IP frame that is fewer than 46 bytes in length to an ATM interface, the router always sets the length in the ATM adoption Layer 5 (AAL5) header to 54 bytes even though the length should be equal to the IP frame length plus the length of the AAL5 header, which is 8 bytes. There is no workaround.
  If the E1/T1 controller on which primary nfas_d is configured does not comes up after bootup of a Cisco access server, the access server will not accept any analog calls on other E1/T1 controllers. This situation occurs when RLM (Redundant Link Manager) is configured on the access server and does not occur in a normal Non-Facility Associated Signaling (NFAS) configuration. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)T will reload when H.323 proxy gateway is run under heavily stressed conditions due to some memory corruption problems. There is no workaround.
  If you copy a file through Simple Network Management Protocol (SNMP) using Trivial File Transfer Protocol (TFTP) to a flash file system when the space left in the Flash memory is less than the size of the file, you will receive a "ciscoFlashCopyStatus of copyOperationSuccess" message even though the file was not copied.
  Workaround: Perform the same function through the command-line interface (CLI).
  A Cisco router might reload if the show ip cef EXEC command is entered while the routing table is changing. There is no workaround.
  You cannot use SNMP to partition the flash.
  Workaround: Partition the flash through the command-line interface (CLI).
  A Cisco AS5800 access server that is acting as a gateway might fail CODEC negotiation on G.723 calls that match VoIP dialpeer 0.
  Workaround: Configure a matching VoIP dialpeer, and configure a CODEC class with the CODECs that are being used.
  A Gigabit Ethernet Interface Processor (GEIP) that is configured for Cisco Encryption Technology (CET) decrypts packets correctly but fails to encrypt packets that match the crypto policy and should be encrypted. In this situation, the GEIP forwards the packets unencrypted instead.
  Workaround: When network topology permits, use the VIP2-40 or VIP2-50 Versatile Interface Processor with one or two PA-FE port adapters.
  If a Cisco 2600 series router that is configured with an inverse multiplexing over ATM (IMA) module might experience IPFAST-2-PAKSTICK tracebacks whenever the ATM interface pings. There is no workaround.
  A Cisco 1600 series router might reload with a bus error if you choose Clickstart in the HTML management interface. Clickstart has subsequently been replaced by Windows Fast Step in Cisco IOS Release 12.0 and later releases.
  Workaround: Upgrade to Cisco IOS Release 12.0 or a later release.
  When a Route Switch Module (RSM) with a large compressed configuration is booted, the RSM might not be considered online by the Supervisor engine.
  Workaround: Boot the RSM without the configuration and use the copy tftp running-config command.
  In a configuration where a Customer Edge (CE) router is connected to multiple Provider Edge (PE) routers, the loss of one of the PE routers might completely remove routes to the CE router from the routing table on another PE router. This situation will only occur for routes exported to a Virtual Private Network (VPN) routing/forwarding instance that uses a different default Route Distinguisher (RD).
  Workaround: Enter the clear ip bgp command to clear the erroneous PE connections.
  A Cisco 7100 series router that is running IP Security (IPSec) Virtual Private Network (VPN) might fail to complete an Internet Key Exchange (IKE) exchange. The output of debugs indicate no packets are going out to the remote crypto peer for a Phase 1 exchange, causing the remote peer to continuously retransmit Phase 1. This situation persists even if you clear the Service Adapters (SAs) and reboot the router. There is no workaround.
  A Cisco AS5800 series access server might generate a spontaneous call setup to a random number. After this call setup is rejected by the switch, Layer 2 of the PRI starts flapping. No dial-in is possible until the controller or ISDN interface is reset. There is no workaround.
  Traffic shaping might cause a Cisco 3640 router to reload under the following conditions:
  There is no workaround.
  A Cisco 7200VXR series router might reload because of a memory corruption caused by an invalid size packets coming in from older cards (WS-C3900-2ISL with HW Rev 0.1.B) on a Cisco Catalyst 3900.
  Workaround: Replace the older version of the cards with Hw Rev 1.1.B or a later version.
  A Cisco Route Switch Processor (RSP) that is configured with a Multiprotocol over ATM (MPOA) Client might occasionally reload. There is no workaround.
  After a few hours of normal operation, a serial interface that is configured with the dialer dtr interface configuration command goes down before the idle timer expires or goes down with traffic on that line. There is no workaround.
  A Cisco Route Switch Processor (RSP) that is configured with a Multiprotocol over ATM (MPOA) client might have spurious memory access. This situation might degrade MPOA shortcut performance. There is no workaround.
  A Cisco router that is running a Multiprotocol over ATM (MPOA) server might reload if you shut down a neighboring interface that is running a MPOA client (MPC). There is no workaround.
  When a Customer Information Control System (CICS) term is forced to use ABEND on the host, the CICSB server will be disabled because the forced ABEND is interpreted as a host failure, and all connections to the server are brought down. Users that are connected to a TX Server might experience outage any time a transaction that is being run on the same server ABENDs. There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1 with an IPSec Service Adapter (ISA) board might reload if you enter the clear crypto sa global configuration command. There is no workaround.
  DTMF-RELAY H245 X does not work properly in Cisco IOS software on a Cisco 1750 router. There is no workaround.
  On a Multilink PPP connection between Cisco and some other vendor routers, IP Real Time Protocol (RTP) header compression (used for voice or video over IP) does not work. If RTP header compression is turned on, the PPP link does not negotiate at all. There is no workaround.
  After the ip audit notify command is configured to send alarms to the NetRanger Director, the router will stop forwarding alarms after an undetermined period of time. There is no workaround.
  A Cisco PA-A3 port adapter occasionally exceeds configured peak cell rate (PCR) limits for available bit rate (ABR) and unspecified bit rate (UBR) traffic classes. There is no workaround.
  A Common Object Module Transaction Interface (COMTI) session object might not check in the license if an exception error occurs. In this situation, the license can become exhausted, and clients cannot connect. There is no workaround.
  Security is not functioning properly with Common Object Module Transaction Interface (COMTI) servers. Transactions that require user IDs and passwords cannot be executed through the Cisco Transaction Connection TXConn subsystem. There is no workaround.
  On a cable modem (CM) registration, if the cable modem termination system (CMTS) already has an entry for the CM, it does not update the IP address field to reflect the real CM IP address.
  This situation only occurs if a Cisco IOS internal Dynamic Host Configuration Protocol (DHCP) server is used. There is no workaround.
  A Route Switch Processor (RSP)-based router with one or more Versatile Interface Processors (VIPs) that is running Cisco IOS Release 12.0 S (or any image with tag support) might experience a memory leak with Cisco Express Forwarding (CEF) and tag switching enabled and the no ip route-cache distributed command configured. This memory leak can be detected by repeatedly entering the show process memory | include OSPF command on the RSP console or vty.
  Workaround: Enable distributed CEF instead of CEF, or turn off tag switching.
  On a Cisco 7200 series router, the serial drivers may cause a memory leak when a reparented packet is sent. There is no workaround.
  If you enable distributed Cisco Express Forwarding (CEF) on a Cisco 7500 series router, and there are one or more prefixes in the IP routing table with the maximum six paths, the router might reload with the following error message:
  %SYS-2-CHUNKBADMAGIC: Bad magic number in chunk header...
  There is no workaround.
  When you remove modem cards on a Cisco AS5800 series access server without updating the startup configuration before the next reload, the access server might encounter a software exception when the access server receives the first modem call and will reload unexpectedly.
  Workaround: Use the write memory command to update the startup configuration before the next router reload so that the pool-range configuration command reflects the existing hardware configuration.
  The Local Management Interface (LMI) is not functioning properly on a Cisco router that is running Cisco IOS Release 12.0(7)T. This situation only occurs with Quad Integrated Communications Controller (QUICC) WAN interface cards (WICs) and not with Network Modules (NMs).
  This situation occurs not only with Frame Relay and LMI. It also occurs with point-to-point High-Level Data Link Control (HDLC) and PPP connections.
  This is a clocking-related issue and will affect certain channel service units (CSUs) or data service units (DSUs) and other devices attached to the following cards: WIC-2T, WIC-2A/S, and WIC-1T (on the Cisco 2600 series router, the Cisco 3620 series router, and the Cisco 3640 series router only). There is no workaround.
  On a Cisco 2600 series router or 3640 router that is running Cisco IOS Release 12.0(8.1) or later releases, a serial interface on a Cisco Network Management-8 access server card that is running Block Serial Tunneling (BSTUN) might not come up after a router reload.
  Workaround: Use the shut command followed by the no shut command, or use the clear interface EXEC command to clear the serial interface.
  On a Cisco router that is running Cisco IOS Release 12.1, Cisco Encryption Technology (CET) will refuse to make to establish encrypted connections.
  To diagnose this situation, enter the debug crypto sessmgmt command to see if you receive the following error message when you attempt to create a connection:
  CRYPTO: Syndrome gen status for conn_id 2000 slot 0:INVALID_INPUT
  There is no workaround.
  A Cisco router that is applying cable access router (CAR) rate-limit Cisco IOS commands to a Channelized E1 serial subinterface on a Virtual Private Network (VPN) Provider Edge (PE), the serial line might drop.
  Workaround: Reassign channels.
  If a dynamic crypto map has multiple entries, internet key exchange (IKE) negotiation might fail with the Tunnel Endpoint Discovery (TED) peer. There is no workaround.
  Spectrum Management task might use many CPU cycles and cause CPU utilization to increase. There is no workaround.
  When Cisco Express Forwarding (CEF) is configured as part of a large configuration (typically with access lists), following boot traffic that is directly addressed to the interfaces of a router might not be received. This condition can be observed on enabled interfaces where IP interfaces appear to be up, but the CEF interfaces are down.
  Workaround: Perform one of the following steps. Boot without CEF enabled. Disable and then re-enable CEF. Enter the no shutdown interface configuration command on each of the interfaces that are affected.
  A Cisco router might reload when configured in a Stack Group Bidding Protocol (SGBP) stack group. This situation might also result in memory corruption. The following SGBP error message occurs in the output of the debug sgbp error command:
  %SGBP-3-INVALID:MQ message with Cancel
  This rare situation might occur whenever PPP calls drop before the SGBP bidding process finishes. When calls drop, the likelihood that this problem occurs increases when either of the following commands is configured:
  multilink bundle-name both or multilink bundle-name endpoint.
  Workaround: To greatly reduce the likelihood of this situation, do not use the following commands:
  multilink bundle-name both or multilink bundle-name endpoint.
  DBConn TCP/IP passthru connections might cause a Cisco router to reload if the host disconnects prematurely. There is no workaround.
  In a situation where a Customer Edge (CE) router is connected to two Provider Edge (PE) routers, if a third PE changes its selected route to the CE router, it might fail to update the Virtual Private Network (VPN) label for the route, resulting in loss of connectivity to that CE.
  Workaround: Use the clear ip route command to clear the route.
  When Cisco Express Forwarding (CEF) is disabled, either through configuration or because of an internal error, any interface where you use the ip verify unicast reverse-path command will no longer forward any frames. This situation occurs on all platforms that support CEF and unicast Reverse Path Forwarding (RPF).
  Workaround: Reenable CEF, or disable the unicast RPF feature from the interfaces using the no ip verify unicast reverse-path command.
  Under rare circumstances, a permanent virtual circuit (PVC) on a Cisco 6400 series node route processor (NRP) might stop sending traffic.
  Workaround: Use the shut and no shut commands on the subinterface that is carrying the PVC.
  Alternate Workaround: Enable ATM traffic shaping.
  A Cisco 3600 series router or a Cisco 2600 series router with an NM-8A/S or NM-4A/S network module interface that has the physical-layer async command configured on the serial interface might see extraneous characters on the serial line during bootup. No other functionality is affected. There is no workaround.
  A Cisco AS5300 series access server that is running Cisco IOS Release 12.1(00.07)T might not be able to route incoming called party numbers using the called-number modem pool configuration command for Open Settlement Protocol (OSP) application.
  Workaround: Use the wildcard on the incoming called-number modem pool configuration command for OSP application.
  When a Cisco Virtual Private Network (VPN) client tries to secure multiple traffic flows to the same Cisco IOS router, and mode config is configured to give an IP address to the client, one of the traffic flows might not have Security Association (SA) set up, and might get the error CRYPTO-4-RECVD_PKT_INV_SPI on the router. There is no workaround.
  Under rare circumstances, a tunnel might have a drop adjacency on the line card while simultaneously having a valid adjacency on the Route Processor (RP).
  Workaround: Enter the clear cef linecard command to download the correct information to the line card.
  A Cisco router might reload with a bus error when the no shut command is entered on a second ATM interface on inverse multiplexing over ATM (IMA) card that is only being used as an ATM interface and not as part of an IMA group. There is no workaround.
  A Cisco 5000 series network access server (NAS) that is running Cisco IOS Release 12.0 or 12.1 might reload when it is running voice (SS7) traffic through the router. The following messages are displayed on the console:
   %ALIGN-3-SPURIOUS: Spurious memory access made
   %SYS-2-WATCHDOG: Process aborted on watchdog timeout, process = ISDN
  Workaround: Upgrade to Cisco IOS Release 12.1(2) or 12.1(2)T.
  When you enter the show sna privileged EXEC command, a Cisco router pauses for three to four minutes, and then displays all 0s in the terminating call IDs (TCIDs) for the intermediate Dependent Logical Unit Requester (DLUR) sessions. This situation might be accompanied by LU-LU sessions that also pause indefinitely. The following is the output of the show sna privileged EXEC command:
  Number of intermediate DLUR sessions 0
  SNA DLUR Assisted Intermediate Sessions PCID (hex) Primary LU Name Secondary LU Name Mode COS ---------------- ----------------- ----------------- -------- ------- 1> 0000000000000000 2> 0000000000000000 3> 0000000000000000 4> 0000000000000000 5> 0000000000000000 6> 0000000000000000 7> 0000000000000000 8> 0000000000000000 9> 0000000000000000 10> 0000000000000000
  There is no workaround.
  If Cisco Express Forwarding (CEF) is not enabled and Resource Reservation Protocol (RSVP) over ATM is used, excessive switched virtual circuits (SVCs) are created. There is no workaround.
  A Cisco Versatile Interface Processor (VIP) that is configured with Cisco Encryption Technology (CET) might not initiate a new crypto connection. The VIP will respond to a connection request from the remote end. However, if the remote end is also a VIP, the crypto connection might never be initiated, because both VIP routers will wait on a connection request. There is no workaround.
  A Cisco AS5300 series access server that is configured with Cisco Voice Manager (CVM) might reload if you use the show call active brief privileged EXEC command. There is no workaround.
  You cannot change the login password when connecting to DB2 through DBConn TCP passthru connections. There is no workaround.
  On a Cisco 7500 series router, if Cisco Express Forwarding (CEF) is enabled, Web Cache Control Protocol (WCCP) or one or more tunnel interface is configured, and packets that are redirected by WCCP or sent over a tunnel arrive over an Inter-Switch Link (ISL) interface, then packets terminating at the router that arrive over the same ISL interface might be dropped. For example, pinging to or from the router might show a 10 to 30 percent packet loss. There is no workaround.
  Cm_process_event_log. cm_process_event_log might cause a Cisco uBR900 universal broadband router to reload when an error message is called from the interrupt process level. This situation only occurs in firewall and IP Security (IPSec) ubr900 Cisco IOS software images.
  The router displays the following traceback messages:
  SYS-3-INVMEMINT: Invalid memory action (malloc) at interrupt level -Traceback= 802D048C 804C8DC8 8024CB3C 802D048C 804C8DC8 8024CB3C 802CEAC4 802D0 7BC 804C8DC8 8024CB3C 802D048C 804C8DC8 8024CB3C 802CEAC4 802D07BC 804C8DC8
  Workaround: Disable event_logging mib object on uBR900 firewall and IPSec images as soon as the router comes back online by setting docsDevEvControl.0 to 1.
  On a Cisco router with port channel interfaces with Inter-Switch Link/Token Ring Inter-Switch Link (ISL/TRISL) encapsulation, any changes on the encapsulation might cause a Cisco bus (cbus) restart. The system returns to normal operation after the cbus restart. There is no workaround.
  Heavy switched virtual circuit (SVC) setup or teardown loads on an ATM interface might cause a Cisco router to reload. There is no workaround.
  In a Voice over IP (VoIP) network with E1 R2 signalling combined with ISDN, the busy signal in R2 signalling might not be notified correctly to the far end of the ISDN side. There is no workaround.
  A Cisco uBR7200 series universal broadband router that is configured with a Cisco cable modem termination system (CMTS) is not protected against rogue cable modems that exhibit non-Data-over-Cable Service Interface Specifications (DOCSIS)-compliant behavior at initial ranging. There might be some modems on an upstream port that range with a negative timing offset. Such rogue modems can deceive the CMTS Map building code into using a large Map Advance when the dynamic Map Advance algorithm (the default) is active on each upstream port.
  The result of this situation is that all the other modems on that upstream port will be unable to send data upstream to the CMTS and will go offline. The router will generate a warning log message every time a modem with bad timing offset is detected. A sample warning message is follows:
  %UBR7200-5-BADTXOFFSET:Bad timing offset -10 detected for cable modem 0010.9500.0a6a
  Workaround: Configure static Map Advance on each cable interface of the router by using the cmts(config-if)# cable map-advance static command-line interface (CLI).
  When you run the interactive voice response (IVR) application on a Cisco 5300 series router, the router might pause indefinitely or reload if no IVR prompts are played in the tool command line (TCL) script when the call is connected. There is no workaround.
  When you run the interactive voice response (IVR) Debit Card application on a Cisco 5300 series router with rotary dial peer configured, the router might pause indefinitely with 20 to 30 simultaneous calls after 2 to 3 hours. There is no workaround.
  A Cisco RM7000 processor that is used by an NPE-300 network processing engine and a Cisco 7140 router might cause the router to execute instructions incorrectly or not at all. This situation might result in memory corruption or reload. There is no workaround.
  When more than three concurrent connections are open between a Cisco router and CheckPoint, subsequent connections take more than 90 seconds. The CheckPoint firewall repeats the last Quick Mode message whenever IP Security (IPSec) Service Adapter (SA) has successfully been negotiated. Retransmissions should only happen under error conditions, so after three retransmissions the router eliminates the SA. When Cisco IOS software interoperates with Checkpoint firewall, the SAs might take longer than usual to become established because of unnecessary retransmissions by the checkpoint firewall.There is no workaround.
  When you use the rate-limit interface configuration command on a Catalyst 6000 FlexWAN interface, the FlexWAN interface never reboots. This situation occurs when you reboot the Multilayer Switch Feature Card (MSFC) after an MSFC reset when the FlexWAN cards have not been reset.
  Workaround: Reset the FlexWAN cards and the MSFC after you use the rate-limit command.
  When a remote Cisco 2600 series router or a Cisco 3600 series router goes down, the BRI line protocol that is configured as "leased line 128K" might not go down properly. The router repeatedly displays the following message while the remote site is down:
  %LINK-3-UPDOWN: Interface BRI1/0, changed state to up" while the remote site has been down.
  This situation occurs when you use an NM-4BRI network module or an NM-8BRI network module and the router is configured with High-Level Data Link Control (HDLC) encapsulation. There is no workaround.
  A Cisco router might reload when reassembling in Session Connector. There is no workaround.
  If you run a TN3270 server on a channel port adapter or configure virtual telecommunications access method (VTAM) to perform connect-outs by Channel Interface Processor (CIP) Systems Network Architecture (SNA) on a channel port adapter, buffer leaks might occur that will affect service between Cisco IOS software and the channel port adapter. Both of these features use local explorers to establish SNA connections, and when these local explorers are received from the channel port adapter by Cisco IOS software, the receive buffer containing the local explorer is not freed properly, resulting in lost buffers. Eventually, Cisco IOS software will run out of channel port adapter receive buffers and stop receiving packets from the channel port adapter. Symptoms of this situation include ceased input packets on the channel interface. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(9)S or later releases might experience problems if you attempt to format, delete, or squeeze slot0: immediately after the show version command is entered or immediately after the router reloads. This is a flash timing-related issue, and subsequent commands that you enter will not be affected. There is no workaround.
  A Cisco AS5800 series access router that is running a Cisco IOS release prior to Release 12.0(1) and that is configured with E1 ports might result in a "ring no answer" message, even though a free channel and modem are available. There is no workaround.
  A spurious access might be logged on a Versatile Interface Processor (VIP) interface that has PPP encapsulation enabled and a VIP quality of service (QoS) feature that requires packet classification to be performed on the output VIP.
  Workaround: Disable the QoS feature.
  A Cisco router might reload with the following error messages:
  System returned to ROM by bus error at PC 0x6023A554, address 0x694B
  IOS (tm) 5300 Software (C5300-JS-M), Version 12.1(0.9), BETA TEST SOFTWARE Copyright (c) 1986-2000 by cisco Systems, Inc. Image text-base: 0x600088F8, data-base: 0x6101E000
  Stack trace from system failure: FP: 0x61F755C8, RA: 0x6023A554 FP: 0x61F75660, RA: 0x602394EC FP: 0x61F75688, RA: 0x6022C080 FP: 0x61F756C0, RA: 0x6022C3F8 FP: 0x61F75798, RA: 0x6022C694 FP: 0x61F757C0, RA: 0x60BE076C FP: 0x61F757F0, RA: 0x60BE0B98 FP: 0x61F75840, RA: 0x60BDCD04
  Workaround: Do not use the debug sgbp errors command.
  On Cisco 7500 series routers running Release 12.1, Encryption Service Adapter (ESA) specific commands, such as crypto card clear-latch 1 are not accepted. There is no workaround.
  On a Cisco router, 50 percent of pings might fail to receive replies when you use the ip cef global configuration command, so the router drops the majority of traffic directly addressed to an interface.
  Workaround: Enter the no shutdown command again on the affected interface.
  If maximum connection is matched or exceeded on an inbound Voice over IP (VoIP) dial peer, an incorrect cause code is returned to the originating router and the call is cleared with normal call clearing. This situation occurs if you attempt to reroute failed calls through the Public Switched Telephone Network (PSTN). There is no workaround.
  If the host abnormally disconnects a TCP DBConn connection during the connection setup phase, an orphaned connection might occur that must be manually reset.
  Workaround: Manually clear the orphan connections using the dbconn clear connection command.
  If a call is disenagaged from a gatekeeper by disengage requests (DRQs) sent to the gateway, the router might reload when the gateway attempts to access a freed memory space. Accessing freed memory is indicated by referencing 0xd0d0d0d (plus some offset). If the freed memory is now allocated for other purposes, this situation might not be obvious and the router might reload. There is no workaround.
  Two-phase commit over TCP passthru connections is not supported. There is no workaround.
  Hot Standby Router Protocol (HSRP) tracking does not work properly. When an interface is tracked, it goes into standby state. When it comes out of standby state, the information is not updated so the tracking interface gets stuck in standby state. There is no workaround.
  On Cisco IOS Release 12.1, dynamic encryption applications might not work correctly if two different remote routers attempt to establish encrypted connections to the same interface on behalf of the same end hosts. There is no workaround.
  A Cisco 7200 series router that uses an PA-A2 port adapter will allocate memory and not release it when you use the show controller command while the ATM interface is administratively shut down. Eventually, the system might run out of memory and need to be reloaded. There is no workaround.
  On a Cisco AS5800 series access server that is running Cisco IOS Release12.0(5)T or portware 2.7 or higher, once a modem is used for handling V110 calls, it cannot be reused to handle asynchronous calls. The asynchronous calls will fail and the user will be disconnected. There is no workaround.
  Systems Network Architecture Switching Services (SNASw) should be able to differentiate physical units (PUs) by Control Point name if the received idblk/idnum is identical. The first PU connects properly, but the second connection generates the following error on the problem determination log (PDLOG):
  **** 00000161 - EXCEPTION 256:2 (0) **** Invalid internal state detected FSM State: = 4 FSM Input: = 1 >From ../dcl/ndrpufsm.c 121
  The output of the show sna privileged EXEC command displays the following:
  grawn#sh snas pu Number of DLUR PUs 1
  SNA DLUR PUs PU Name PU ID State DLUS Name -------- -------- -------- ----------------- 1> M1PU 05D00000 Active NETA.CPDEREK
  After the second PU connects, the output of the show sna privileged EXEC command displays:
  grawn#sh snas pu Number of DLUR PUs 1
  SNA DLUR PUs PU Name PU ID State DLUS Name -------- -------- -------- ----------------- 1> M1PU 05D00000 Reset
  Workaround: Change idblk/idnum to 00000000.
  A Cisco 5300 series router that is running Voice over IP (VoIP) with T1 channel-associated signalling (CAS) might lose memory suddenly and severely (not a gradual memory leak). This situation occurs because variables on the memory stack are not properly initialized. The router might change the value based on what is left on the stack, which depends on the various code execution paths of the call state machine (CSM) manager for the T1 CAS handling.
  The router might reload when it runs out of memory in attempting to allocate 1552 bytes or more by the ossUserMalloc function. The router might also reload before the memory leak is detected. The router will need to eventually be reloaded to recover the memory loss. There is no workaround.
  On a Provider Edge (PE) router, theCisco Express Forwarding (CEF) table might lose the tags for a default route that belongs to certain Virtual Private Network (VPN) routing and forwarding tables (VRFs). This situation occurs every minute and the tag will come back after 2 to10 seconds. This situation occurs when the default is learned through IBGP from another PE router.
  Workaround: Configure a static default for that vrf pointing to the PE where you should learn it from.
  Under heavy traffic, a PA-A3 port adapter might experience a SAR0 reload. If this condition occurs on a Cisco 7200 series router, you must reload the router to recover normal operation. On a Cisco 7500 series RSP, this situation might result in commands from the RSP to the port adapter failing, but the port adapter should be able to recover without a router reload. There is no workaround.
  A Cisco 5200 series router, a Cisco 5300 series router, or a Cisco 5800 series router might experience a fatal error that causes a system to reload. The error only occurs if virtual private dial-up network (VPDN) groups or profiles are configured within resource-pooling profiles in the local Resource Pool Manager Server (RPMS) or the remote RPMS. There is no workaround.
  A systems Network Architecture (SNA) switch is correctly looking at MAXDATA for the segmentation of logical unit-logical unit (LU-LU) SESSION DATA to downstream physical units (PUs). However, the SNA switch is failing to segment system services control point (SSCP)-LU data to such PUs, including USS10 screens. There is no workaround.
  On a Cisco router that is running IP Security (IPSec), if an IPSec encrypted packet comes in through a process switched interface that does not have encryption enabled, you cannot decrypt the packet. In addition, the router might rise to 100 percent CPU utilization until the Security Association (SA) times out.
  To determine if this situation is the cause of 100 percent CPU utilization, compare the decrypt counters in the show crypto engine connections active privileged EXEC command with the counters in the show crypto ipsec sa EXEC command to learn if the counters are quite different.
  Workaround: Enable fast switching on the unencrypted interface.
  If an interface that is configured with IP Security (IPSec) and that uses the ip address negotiated interface configuration command and the crypto map interface configuration command is recycled (made active then inactive) more than once within the IPSec Service Adapter (SA) lifetime and a new IP address is assigned to the interface, IPSec will retain the previous IP address.
  Workaround: Delete the IPSec SA on the router that uses the ip address negotiated interface configuration command to clear the clear crypto sa global configuration command from the IPSec SA.
  Alternate Workaround: Use a static IP address instead of a dynamic IP address. Use the crypto map local-address global configuration command to identify a static IP address in IPSec.
  Multiprotocol Label Switching (MPLS) forwarding entries might not be built properly for routes learned through IP Border Gateway Protocol (BGP) so packets addressed to a BGP destination will not be forwarded along MPLS Label Switched Paths (LSPs) but will be forwarded as IP packets. There is no workaround.
  On a Cisco 7206 router or a Cisco 7206VXR router that is running Cisco IOS Release 12.1(1.1) and is configured with a Token Ring (TR) Inter-Switch Link (ISL) to a Cisco Catalyst 3900, Logical Link Control, (LLC2) type 2 packets from the Catalyst 3900 might remain in the input queue of the Fast Ethernet interface on the 7206 router and the 7206 router will be unable to accept any more packets on the interface.
  Workaround: Enter the hold-queue 4096 {in} interface configuration command to configure the queue in the largest size possible.
  In Cisco IOS Release 12.0 and Cisco IOS Release 12.0 T, the copy start tftp command uses a default save name of "startup-config" rather than the usual "routername-config."
  Workaround: Use the copy run tftp command.
  On a Cisco router that is configured with a VXC-2TE1 PA port adapter, the fragment option will not be available in the access-list command. The fragment option might be turned on even if it is not specified in the command. There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1(1.1) and that is configured with a VXC-2T1/E1 card might have digital signal processors (DSPs) that seize the trunk and not release it after a router reload. This situation will prevent any calls from occurring.
  Workaround: Enter the shut interface configuration command followed by the no shut interface configuration command on the affected DSP under the dspint DSPFarm. This is a temporary workaround that will restore the DSP until another reload of the router occurs.
  A Systems Network Architecture Switching Services (SNASw) router might reload during normal operation with watchdog_forced_here. There is no workaround.
  If you delete a Layer 3 port channel or subinterface when Hot Standby Router Protocol (HSRP) is configured with a secondary address, the system might restart. Layer 2 port channels are not affected.
  Workaround: Unconfigure the HSRP secondary addresses before deleting the subinterface or port channel.
  A missing H.323 ID configuration might cause the gateway to access (read) using a NULL pointer and might cause reloads on platforms where the service alignment global configuration command is a default, as it is on a Cisco 2600, 3600, 5300, and 7200 series router or 5800 series access server. A spurious read access to low memory will generate a syslog message instead of reloading the router. This situation can occur while the gateway is registering or unregistering from the gatekeeper while there is no H.323-ID configured.
  Workaround: Configure H.323-ID and ensure that the no gateway configuration command is entered first. Disabling the gateway prevents H.323 software from using a NULL pointer while the configuration process is not complete.
  A memory leak might occur when the interactive voice request (IVR) script calls for prompt/audio files that do not exist.
   Workaround: Ensure that all prompt/audio files are valid (and spelled correctly).
  A Cisco router might reload when interactive voice response (IVR) in a voice gateway requests authorization.
  Workaround: Enter the aaa authorization exec h323 group radius global configuration command when authorization is needed for voice gateway.
  Alternate Workaround: Upgrade to Cisco IOS Release 12.1(2) or Release 12.1(2)T.
  A Cisco router might reload on configuration. There is no workaround.
  A Cisco router might reload when the interactive voice response (IVR) in voice gateway sends an authentication/authorization request to the billing (RADIUS) server and the billing server is down.
  Workaround: Ensure that the billing server is running for the voice gateway.
  Alternate Workaround: Do not perform authentication/authorization in the IVR script.
  Alternate Workaround: Upgrade to Cisco IOS Release 12.1(2) or Release 12.1(2)T.

Novell IPX, XNS, and Apollo Domain

  IPX routes are not received on a Cisco 3660 router that is running integrated routing and bridging (IRB) with bridging on an ATM interface. There is no workaround.
  A cisco router might not make odd length process-switched packets even or odd length pings that originate from this router when the output interface is a Bridge Group Virtual Interface (BVI). There is no workaround.

TCP/IP Host-Mode Services

  A Cisco router might lose memory or reload during periods of heavy traffic while using a voice application or the gateway application. There is no workaround.
  When you enter the ip tcp selective-ack global configuration command on a Cisco router that is running Cisco IOS Release 12.1(1), a memory leak will occur. The leak is slow but will eventually reduce the available memory in the router and reload the router.
  Workaround: Enter the no ip tcp selective-ack global configuration command.

TN3270

  A TN3270 client on a Cisco access server might erroneously leave the keyboard in a locked state causing all keyboard input to be rejected until you press the master_reset keypress. This situation occurs when an application sends write-structured field commands with the keyboard-restore indicator on the associated write-control character. There is no workaround.

Wide-Area Networking

  When a group-asynchronous interface is a member of a dialer rotary-group, the encapsulation ppp command might disappear from the configuration after a reboot.
  Workaround: Configure encapsulation on the dialer interface instead of on the group-async interface.
  On a Cisco router that is running Cisco IOS Release 11.3 or Cisco IOS Release 12.0, if you use both the substitute-source option and the substitute-destination option in the same X.25 route command, the Source Address (SA) in the forwarded call is corrupted. There is no workaround.
  A member of a large-scale dialout stack group might continue to bid for outgoing calls even if it does not have the capability to actually make the outgoing call (T1/E1 cable is unplugged or PRI is not up, and so on).
  Workaround: Manually remove the member from the stack group until the network access server (NAS) is restored to full working condition.
  A Cisco router might reload because of illegal access to a low memory location when there are close to 4000 active X.25 calls. There is no workaround.
  A Cisco router might reload because of a bus error when accessing a freed call control block (CCB). There is no workaround.
  A Cisco 7200VXR router might reload with a bus error because the packet-by-packet compression code is being passed packets that contain particles. There is no workaround.
  A Cisco router might experience a bus error and reload if you enable the ATM bundle on the PA-A3-OC3MM ATM port adapter. There is no workaround.
  Multiprotocol Label Switching (MPLS) Virtual Private Network (VPN) traffic cannot be switched by LAN Emulation (LANE) interfaces. There is no workaround.
  If you try to use idle timeout with class in dialer maps for different locations, the individual timeout will not override local idle timeout configured under dialer interface. There is no workaround.
  A Cisco router might experience high CPU utilization when priority queueing is enabled because of X.25 background processes. There is no workaround.
  Timer data structures used by the dialer component might become corrupted. The corruption might occur because of operator-initiated actions that involve the removal of the dialer function on an interface. Examples of these actions would be configuring leased-line BRI, or entering the no dialer in-band interface configuration command. The symptom might not occur for days or weeks after the operator action that caused it. Due to this condition, these actions should be avoided. There is no workaround.
  A Cisco router will exhibit traceback messages when the router sends Inverse Address Resolution Protocol (ARP) messages for protocols like Novell, DECnet, and XNS. This situation will not affect router operation. There is no workaround.
  If a dialer profile (with multiple BRIs in pool) is configured with a very low load threshold value using the dialer load-threshold command and if PPP Multilink is also configured, then all the channels except the B channels on one BRI are not disconnected. This situation might cause the B channels in the pool to flap and could also result in the failure of subsequent pings to the same link.
  Workaround: Raise the load threshold to a higher value.
  The input queue on a Cisco 7500 series router might show 76/75, which can result on the line going down on a High-Speed Serial Interface (HSSI) port adapter even though the VIP console shows that the queue is empty and the line protocol is up. This condition only occurs when PPP encapsulation is enabled.
  Workaround: Move the card to a new slot, change to High-Level Data Link Control (HDLC), or reload the router.
  A Cisco universal access server might reload with a bus error when the router is used as a PPP callback server. There is no workaround.
  There is no way to specify that only one call per terminal endpoint identifier (TEI) should be allowed when a TEI for a second call is reused where there is a common local directory number (LDN) for all B-channels and the telco does not send an EndPoint Identifier (EID) information element (IE). This situation rarely occurs. There is no workaround.
  With X.25 over TCP (XOT), when there are many IP addresses in the X.25 route command and the first IP address is unreachable, a Cisco router will take the next IP address. The default X.25 parameters are then used instead of the ones configured on the X.25 interface, which causes the X.25 public switch to clear the call. There is no workaround.
  If you use X.25 over TCP (XOT) protocol translation and use the session-timeout line configuration command on the vty interfaces, the router might reload with a bus error. There is no workaround.
  When the IP connection for Redundant Link Manager/Signaling System 7 (SS7) fails, ISDN Layer 2 might not restart when the connection is restored. There is no workaround.
  Non-Facility Associated Signalling (NFAS) and Redundant Link Manager (RLM) interfaces do not have T309-enable activated by default. Loss of Layer 2 traffic might cause the loss of all calls.
  Workaround: Add T309-enable to the configuration on the NFAS and RLM interfaces.
  A Cisco router might fail to bring up Layer 2 channels for National ISDN and 5ESS Custom switch profiles where dual terminal endpoint identifiers (TEIs) are used. There is no workaround.
  A Cisco router will reload when you use the x25 routing global configuration command with the Domain Name System (DNS) feature. This situation occurs even if this command is not the first choice in the routing process. There is no workaround.
  When you use Frame Relay traffic shaping with the first-in, first-out (FIFO) queueing default, the queue count for the traffic shaping structure might be inconsistent with the queue count for the per-virtual-circuit (VC) queues. This condition occurs when the VC queue fills up and the system drops a packet to make room for critical traffic in the VC queue.
  Workaround: Remove and then configure Frame Relay traffic shaping to reinitialize the counter.
  In large scale dial-out (LSDO), once a Cisco router in a stack group gets a Stack Group Bidding Protocol (SGBP) discover message, it needs to find out how many free B channels it has and offer that value to the client so that the client can gather all the offers and use the router with most free B channels to dial out. If you configure sdn_get_num_free_bchan() to get the number of free B channels, a BRI sdn_get_num_free_bchan() should be isdn_get_num_free_bchan(), which is a function that cannot be configured. The function might not return the correct number of free B channels. There is no workaround.
  Dialer will not dial out if the Large Scale DialOut (LSDO) feature is configured and a Stack Group Bidding Protocol (SGBP) group is not configured.
  Workaround: Configure a random SGBP group by entering the following:
  dial-2600b#conf t Enter one configuration command per line and end with pressing Control Z dial-2600b(config)#sgbp gro dial-2600b(config)#sgbp group lsd dial-2600b(config)#sgbp dial-bids dial-2600b(config)#end
  A Cisco 7500 series router that is running Cisco IOS Releases 11.1(31.1)CC, 12.0(9.1), 12.0(8.6)S1, 12.0(9)S, 12.1(0.8), and 12.1(0.8)T with a PA-A3 port adapter will not send Operation, Administration, and Maintenance (OAM) cells. There is no workaround.
  A Cisco router that is connected to ISDN BRI lines that deactivate Layer 1 (typically done by European telcos to save power on BRI lines when no ISDN calls are active) and that use dialer interfaces (either legacy DDR or dialer profiles) will not be able to dial out.
  Workaround: Remove the dialer interfaces and configure the BRI interfaces instead.
  A gateway-to-gateway call that does not use the interactive voice response (IVR) application produces two authentication, authorization, and accounting (AAA) records. One record has cause 0x0 and the other has cause 0x10. The same two records appear on both the originating gateway and the terminating gateway.
  A gateway-to-gateway call that does use IVR produces two sets of AAA records. One record on the originating gateway has cause 0x10 and the second record has cause 0x29. Both records on the terminating gateway have cause 0x29. There is no workaround.
  A Cisco AS5800 series access server that is configured with Digital Number Identification Service (DNIS) screening and authentication, authorization, and accounting (AAA) pre authorization disconnects even though the "isdn guard-timer < > on-expiry" is configured to "accept." There is no workaround.
  A memory leak in a Cisco router might occur if you use the show isdn active command on a tty other than the console port while calls are in process. There is no workaround.
  A Cisco router might reload because of a bus error in fair_dequeue_inline. There is no workaround.
  If isdn no_status_enq is not configured, calls will be dropped during link failure and recovery. There is no workaround.
  Certain configurations might cause spurious memory accesses, failure at start up, or incomplete configuration data processing. There is no workaround.
  A Cisco 4000 series router might reload while running as a Layer 2 Tunnel Protocol (L2TP) Network Server (LNS) for two Cisco 5300 series L2TP Access Concentrators (LACs). This situation is caused by IP cache entry and might occur when some virtual private dial up network (VPDN) clients are shut down to troubleshoot L2TP session problems on another router that is running a voice image when more VPDN clients are added.
  When the L2TP subsystem attempts to fast switch a packet, it calls ip_fastswitch() which in turn calls ipcache_lookup_fast(destination) to look up the cache entry. The return cache entry points to a virtual access interface that should have been an Ethernet interface. This situation causes the packet to enter the L2TP subsystem again, resulting in another L2TP+UDP+IP header encapsulation. This looping continues until the stack runs out when the L2TP subsystem encapsulates so many times that memory locations above the packet get overwritten. There is no workaround.
  When Frame Relay traffic shaping is configured and either custom queueing or priority queueing is enabled at the virtual circuit (VC) level, an alignment error might occur. The error occurs when the priority/custom queue list checks User Datagram Protocol (UDP) packets and the current packet is a UDP packet that is queued because of traffic shaping. There is no workaround.
  A Cisco network access server (NAS) might process a RLM SERVER_SWITCHED message like a LINK_UP message and drop calls even if the Redundant Link Manager (RLM) link never failed.
  Workaround: Upgrade to Cisco IOS Release 12.1(2).
  A Cisco router will reboot if you delete a dialer map by using the no dialer map command when a PPP callback server is configured and there is a callback pending expiration of the callback timer that uses that dialer map.
  Workaround: Shut down the interface, wait for an interval corresponding to the callback timer, remove the dialer map, then enter the no shut command on the interface. The callback timer has the same value as the enable timer and is set by the dialer enable-timeout command.
  On a Cisco router that is running Cisco IOS Release 12.1T, IP Control Protocol (IPCP) on multilink group interfaces might fail to negotiate.
  Workaround: Move the configuration for the multilink group to a virtual template.
  If you enable conditional debugging on a Cisco router or use the debug condition EXEC command, the router might reload. There is no workaround.
  The B-channels on the "primary-d" NFAS-RLM group will erroneously become active when the D channel becomes active. There is no workaround.
  A Cisco 7500 series router that is running Cisco IOS Release 12.0(7)T with Gigabit Ethernet Interface Processor (GEIP) and ATM interfaces might reload with the following message:
  System restarted by bus error at PC 0x600705F8, address 0x8
  There is no workaround.
  With conditional debugging enabled on a Cisco router, in the output of a debug log you can see PPP negotiation occurring in an analog call but not in an ISDN call. There is no workaround.
  Incoming calls on a NET3 ISDN line might fail because of a missing information element (IE) in the router response to the incoming SETUP message. There is no workaround.
  After 48 hours of connecting and disconnecting data calls, a Cisco router will exhibit the following messages:
  2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911 2d14h: %ISDN-6-INVALID_TIMER: LIF_RemoveTimer: Invalid Timer Handle, caller 0x6004BECC handle -1 2d14h: %ISDN-6-NO_TIMER: No Free Timer Entry, caller 0x6004F210, timers used 30911
  There is no workaround.
   If you use the dialer hold-queue interface configuration command, a system might unexpectedly restart if two dial-on-demand calls are initiated on an ISDN interface using an NTT switch type, when sending broadcast packets to two separate destinations where both calls fail to connect.
   Workaround: Remove the dialer hold queue command.
  A Cisco AS5300 series access server that is running Cisco IOS Release 12.1 T might exhibit a "spurious memory access" message in the output of the show command. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(7)T might reload when Domain Name System (DNS)-based X.25 routing is used. There is no workaround.
  On a Cisco 800 series router that is running Cisco IOS Release 12.0(7)T, plain old telephone service (POTS) flash functionality does not work for NI switchtype, although it does work for 5ESS switchtype. There is no workaround.
  After a failover, ISDN Layer 2 might not recover.
  Workaround: Enter the shutdown interface configuration command followed by the no shutdown interface configuration command on the serial interface to restore the digital subscriber lines (DSLs).
  If you use the shutdown interface configuration command on a Non-Facility Associated Signaling (NFAS) PRI group followed by the no shutdown interface configuration command, a Cisco router will leave the "nfas_d none" digital subscriber lines (DSLs) in the ADMINISTRATIVELY DOWN state. if the D channel is up (and the ISDN Layer 2 is up), calls come in and get connected but PPP fails.There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1(1.1) or Cisco IOS Release 12.1(1)T might experience problems with X.25 address insertion and removal when the router forwards a call. The router might display "%SYS-4-REGEXP" error messages and tracebacks. There is no workaround.

Resolved Caveats—Release 12.1(1)

All the caveats listed in this section are resolved in Release 12.1(1). This section describes only severity 1 and 2 caveats.

Access Server

  If an NM-AM modem is set to mark, space, or even parity, the modem autoconfigure type line configuration command will not function properly.
  Workaround: Use the script startup line configuration command and the script reset line configuration command.

Basic System Services

  A Cisco router might stop responding to Simple Network Management Protocol (SNMP) queries. This situation occurs when the number of SNMP request packets to be processed at the same time is high. There is no workaround.
  Under rare circumstances, a Cisco router might reload with timer corruption. There is no workaround.
  Under a sustained heavy load of calls, authentication, authorization, and accounting (AAA) requests backup, using up memory until memory runs out and the network access server reloads. There is no workaround.
  If a serial interface is frequently flapping, the router might pause indefinitely with a stack trace indicating that it is in usecdelay() as a result of cbus_mci_serial_reset() being called while at interrupt level. This situation rarely occurs.
  Workaround: Enter the shutdown interface configuration command on the serial interface that is flapping.
  When ISDN calls are brought up on a Cisco router that has ISDN traps configured, memory leaks will occur in the Simple Network Management Protocol (SNMP) trap process. In addition, some ISDN traps will be lost because of the invalid variable bindings (varBinds) that cause the leaks.
  Workaround: Remove all snmp-server enable traps [isdn] global configuration commands from the router.
  A Cisco router that is running Cisco IOS Release 12.0(7)T cannot execute boot configuration commands from Flash, and exhibits the following error message:
  %Error opening nvram:/startup-config (File system is in an inconsistent state)
  When this message is displayed, no configuration is loaded. If you enter the copy startup-config running-config command and then enter the no shutdown interface configuration command, the router will come back on line. There is no workaround.
  A call on a Cisco MC3810 multiservice access concentrator might be routed to a port that is busy and waste a digital signal processor (DSP) resource. This problem can be detected when you enter the show voice dsp privileged EXEC command and see DSPs that are in a BUSY state but not referencing a specific port, and when you enter the show voice call {summary} privileged EXEC command and see the ports stay continuously in a ST_DISCONN_PENDING state.
  Workaround: Shut down or remove any POTS dial peers that reference ports that are busied out from the private branch exchange (PBX) so that calls will not be routed to those ports.
  If you enable IP route-caching (IP fast-switching) on a PPP serial interface that is part of a multilink bundle, traffic that is destined for that bundle might stop.
  Workaround: Shut down the serial interfaces before entering the ip route-cache EXEC command, and then enable the interfaces.
  A Cisco router or access server might reload when a reference is made to a teletype (TTY)/Telnet session that has been externally ended. There is no workaround.
  A Cisco 7200 series router that is running PPP stack hardware compression with a compression service adapter (CSA) might reload because of a memory leak. This situation occurs when a PA-MC-T1 port adapter is used. It can also occur when a multichannel E3 PA port adapter is used.
  Workaround: If the CPU can handle the load, enter the compress {predictor | stac [csa slot | software]} interface configuration command.
  Outbound Challenge Handshake Authentication Protocol (CHAP) does not function properly on Cisco routers.
  Workaround: Disable the AAA access control model by entering the no aaa new-model global configuration command.
  Cisco 7200 series routers with an NPE-300 network processing engine installed will not boot up when certain Cisco IOS Release 12.0(5)XE3 subset images are installed. The router will pause indefinitely in a very early stage of booting up, and the only way to recover the router is to perform a power cycle. For systems that are set for auto boot, you will need to enter the break command to abort the boot process and return to the ROM monitor before the Cisco IOS Release 12.0(5)XE3 image is launched for execution. You will then need to either modify the software configuration register to revert to a manual boot of some other known good image, or switch the PCMCIA flash card with a known good image in case the system is set for default image boot from the slot0: PCMCIA card. There is no workaround.
  You might not be able to change the size of priority queues in priority queueing with virtual template interfaces on a Cisco 3640 router that is running Cisco IOS Release 12.0(8) or Cisco IOS Release 12.0(7). There is no workaround.
  Traceback messages might appear when a Cisco 3620 router or a Cisco 3640 router is reloaded. The tracebacks occur because the router is unable to lock an uninitialized semaphore. Other functions are affected. There is no workaround.
  When a communication error occurs after TACACS+ TCP handshake, the router slips to the next method instead of taking the next server in the same authentication, authorization, and accounting (AAA) server group. There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.0(7) might reload because of a bus error or pause indefinitely when FRF.9 payload compression is enabled with a hardware compression adapter and Frame Relay Traffic Shaping has already been configured on the serial interface.
  Workaround: Use FRF.9 software compression.
  This caveat adds support for a new revision of a hardware component that fixes a previous error. For the benefit of users that have not upgraded to the new hardware, it will also exhibit a warning error message that indicates the old hardware revision.
  Cisco 7200 series routers with NPE-175 or NPE-225 network processing engines must upgrade to Cisco IOS releases that incorporate this change (for example, Cisco IOS Release 12.0(9) and later releases or Cisco IOS Release 12.0(9)S and later releases). Use of older Cisco IOS releases might result in unpredictable malfunctions. Please see the following document for further information:
  http://www.cisco.com/warp/customer/770/fn8611.shtml
  A Cisco router might reload when hardware compression is configured in conjunction with Generic Traffic Shaping and Fancy Queueing on a compression service adapter (CSA) at the main interface. This situation only occurs when all 3 features are configured.
  Workaround: Use software compression.
  A Cisco MC3810 multiservice access concentrator might reload after a power cycle.
  Workaround: Remove the T1 cable while the router is booting.
  A Cisco router might reload when Resource Reservation Protocol (RSVP) and class-based weighted fair queueing (CBWFQ) are enabled on the same interface. There is no workaround.
  A Cisco router might reload because of memory corruption or stack overflow if you use the show ip bgp regexp pattern privileged EXEC command.
  Workaround: Simplify the regexp pattern.
  A voice gateway might reload when a voice call disconnects if you use the gw-accounting [h323 | syslog] global configuration command and you do not properly configure the authentication, authorization, and accounting (AAA) accounting method list for voice.
  Workaround: Use the aaa accounting connection h323 {start-stop} [radius | tacacs+] global configuration command.

DECnet

  If you configure a Cisco router using the decnet routing-timer seconds interface configuration command, the router might create a DECnet in-routing filter access list number after you reload the router. The access list number will be the same as the time in seconds specified in the decnet routing-timer seconds interface configuration command, and might be invalid. There is no workaround.
  If a Cisco router is running a release earlier than Cisco IOS Release 11.3 and is using the DECnet Phase IV routing protocol, the DECnet network might experience loops or the incorrect selection of routes for periods of time up the value entered with the decnet routing-timer interface configuration command. When a network link goes up or down, one of the routers attached to that link might start sending DECnet traffic down the wrong route. You can enter the show decnet traffic EXEC command to show the number of messages that have been discarded with the "too many visits" count. If you enter the show decnet route configuration command, you can check the routes selected by the router.
  Workaround: Reduce the time in the decnet routing-timer seconds interface configuration command to reduce the time taken for the network to converge.
  A Cisco router might reload if it pings a DECnet address across an ISDN BRI connection. Pinging IP continues to work, as does DECnet routing. There is no workaround.

IBM Connectivity

  A Cisco router that is running Cisco IOS Release 12.0(5)XN might experience alignment errors and exhibit the following error message:
  %ALIGN-3-SPURIOUS at dlsw_work (0x60E9E53C) %ALIGN-3-SPURIOUS:Spurious memory access made at 0x60E9E53C reading 0x1
  There is no workaround.
  A Cisco router that is running Block Serial Tunneling (BSTUN) might experience an input queue wedge. Reloading the router will temporarily fix the situation. There is no workaround.
  A Cisco 7500 series router that is running a Cisco IOS release higher than the Cisco IOS Release 12.0(3c) might experience a situation where attached Network Basic Input/Output System (NetBIOS) PCs are not reaching the Ethernet-attached NetBIOS server. When this situation occurs, the router sends an incorrect LAN Emulation Address Resolution Protocol (LE_ARP) request, and is unable to resolve the route descriptor for the local LANE cloud and the Token Ring behind the Token Ring switch. A Logical Link Control, type 2 (llc2) frame out of the data-link switching (DLSw) cloud directed to the PC will trigger a LANE LE_ARP lookup on the router for the wrong route descriptor. As a result, the LANE data direct virtual channel connection (VCC) is only used in one direction from the Token Ring switch to NetBIOS PCs, and the NetBIOS connection pauses indefinitely. There is no workaround.
  With new hardware, frames that were previously forwarded by configuring the source-bridge old-sna interface configuration command will no longer be forwarded. To reduce the overhead in processing frames not destined for a particular MAC address or frames not destined to be bridged over the bridge immediately connected to the MAC address, the hardware will not receive the packet for processing. As a result, the source-bridge old-sna interface configuration command no longer has any functional purpose. There is no workaround.

Interfaces and Bridging

  A High-Speed Serial Interface (HSSI) port adapter is disabled with following error message if the load reaches approximately 40 Mbps:
  %MUESLIX-1-STOPFAIL: Mx serial, Hssi2/0 Stop Failed at disable port
  %MUESLIX-1-STOPFAIL: Mx serial, Hssi2/0 Stop Failed at disable port
  There is no workaround.
  A PA-4R-DTR port adapter or a Cisco 2600 series router sometimes inserts at the wrong ring speed. The interface will recognize the incorrect ring speed and remove itself from the ring. If the router is connected to a Smart controlled access unit (CAU), the Smart CAU might disable the port because of the incorrect ring speed. In this situation, the router will try to re-insert into the ring, but it will not be able to. This condition is rare, and will not cause any physical problems with the ring. Unless the router is connected to a Smart CAU, which will wrap the port automatically, this condition is difficult to detect. There is no workaround.
  A Cisco 7206VXR router that is running Cisco IOS Release 12.0(6.5)T2 and has a network processing engine (NPE)-300 network processing engine might reload with a bus error. There is no workaround.
  The address that is assigned by Dynamic Host Configuration Protocol (DHCP) might not be accepted if the remote client that is terminating a PPP session into your virtual template has tried to use a different address. There is no workaround.
  A Cisco 7200 series router with a PA-4R port adapter might accept a packet with an invalid frame check sequence (FCS) and pass it on to remote source-route bridging (RSRB). This situation does not occur in the PA-4R-DTR port adapter. There is no workaround.
  A Cisco AS5800 universal access server with E1 PR1 lines that is running Cisco IOS Release 12.0(4)XJ2 or Cisco IOS Release 12.0(5)T1 might drop Serial Line Internet Protocol (SLIP) connections after Cisco IOS installs a route for a remote user. PPP users (both asynchronous and ISDN) can connect, but the SLIP users cannot connect with MICA modems that are running 2620 portware. There is no workaround.
  A Cisco PA-A1 port adapter does not detect the Loss of Cell Alignment (LOCA) error. There is no workaround.
  When performing an online insertion and removal (OIR) on a working PA-A3 port adapter with an interface configured for operation, administration, and maintenance (OAM), the permanent virtual connection (PVC) and the interface might not come up. If a second OIR is performed, both the PVC and the interface come up and traffic is restored. This situation will only occur when per-VC queueing is enabled on the interface. There is no workaround.
  A Cisco 7500 series router with a VIP2-based Fast Ethernet port adapter might exhibit the following message:
  %CBUS-3-CCBPTIMEOUT: CCB handover timed out, CCB 0x5800FFB0, slot 9 -Traceback= 60338B60 603362B8 603368D8 60310404 60335C28 6023418C 602B6AFC 602B6 AE8
  Workaround: Enter the no cdp run global configuration command to disable Cisco Discovery Protocol (CDP).

IP Routing Protocols

  A Cisco router that has tag switching enabled and is running Cisco IOS Release 12.0(5)T might reload if a tag advertisement appears in a certain time window when a related routing update takes place. An ATM interface transition might cause this condition. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0 T with conditional advertisement does not advertise the prefixes specified in advertise-map. There is no workaround.
  If you have an Open Shortest Path First (OSPF) external route and a floating static route (as backup for the external route) with the nexthop interface OSPF enabled, and if there is a flap in external route, floating static will be installed in the routing table. The floating static route will not get replaced by the external route, even when the external route comes back up.
  Workaround: Enter the clear ip route {network [mask]|*} EXEC command.
  Open Shortest Path First (OSPF) Router and Summary link-state advertisement (LSA) might not install the network into the routing table.
  Workaround: Clear the routing table.
  Alternate Workaround: Restart the OSPF process.
  Network Address Translation (NAT) might stop creating new entries after running for an indeterminate amount of time.
  Workaround: Clear the NAT translation table.
  The dataless header register does not work properly in Cisco IOS Release 12.0(7)S. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(5)T and later releases, Cisco IOS Release 12.1, or Cisco IOS Release 12.0 ST might reload if the clear ip bgp {*} EXEC command is entered on a peer provider edge (PE) router. The following stack trace is exhibited:
  bgp_fwdentry_info
  bgp_v4class_update_fwdtable_walker
  rn_walktree_version
  bgp_update_fwdtable bgp_router
  The reload occurs on the local PE when a PE Internal Border Gateway Protocol (IBGP) session is cleared on the remote box. If the PE is importing routes from other PE devices, clearing BGP session on the remote PE will cause the local PE to reload. There is no workaround.
  A Cisco router might reload if certain IP packets addressed to a multicast destination with an invalid IP header are received. On a Route Switch Processor (RSP), this situation might cause a bus error reload with RSP-specific code and ipmulticast_les_fastswitch on a stack trace at:
   address=0x58200000.
  This situation might also cause a generic memory corruption reload on the RSP or other nonparticle-based platforms. This bug does not exist in Cisco IOS Release 11.1CC or Cisco  IOS Release 11.2.
  Workaround: Disable multicast fast switching on all interfaces by entering the no ip mroute-cache interface configuration command.
  If a Cisco router receives a Resource Reservation Protocol (RSVP) RESV message to refresh link-state packet (LSP) for which it is the source, and if the RESV message contains Multiprotocol Label Switching (MPLS) label other than the one previously received, the router will attempt to perform a label change operation. If the label change operation fails, the router might reload while executing the appropriate error handling procedures. This situation rarely occurs. There is no workaround.
  Protocol Independent Multicast (PIM)v2 Candidate Rendezvous Point (RP) advertisement packets are discarded by a router if the packets were sent by that router. This breaks the Candidate-RP and bootstrap router (BSR) mechanism in PIMv2.
  Workaround: Configure Candidate-RP in routers that are not BSR.
  A Cisco 827 router might reload or experience spurious access in iprouting_destination_valid. The spurious access is caused by reading an invalid location, so memory is not corrupted. There is no workaround.
  Sources of multicast packets that send at a low frequency might have their (S,G) expire in the first-hop designated router (DR). The corresponding (S,G) might also expire in the rendezvous point (RP).
  If the (S,G) state is later created from a Protocol Independent Multicast (PIM) Join packet from downstream, then the register flag is not set for this (S,G) in this router. If a packet is then received from the source, the register packet will not be sent to the RP so the A-flag for the (S,G) will not be set in the RP, which would prevent the subject source from being advertised to the Multicast Source Discovery Protocol (MSDP) peers. There is no workaround.
  A Cisco 7200 series or Cisco 7500 series router might restart with a bus error when you are entering the clear ip bgp dampening EXEC command. This situation will occur only when some routes are in the dampened state when the command is issued. There is no workaround.

ISO CLNS

  When two events occur within a short period of time that cause a Cisco router that is running Intermediate System-to-Intermediate System (IS-IS) to regenerate a new link-state packet (LSP), a race condition might occur that can cause the router to skip the second LSP generation. This situation results in the router not advertising its current state to the rest of the network. (For example, an adjacency might be missing from its LSP, or an old adjacency that does not exist anymore is still advertised.) All the routers in the network will then compute invalid IP routes.
  This condition occurs only in Cisco IOS Release 12.0(5)T and later releases, and can only occur when IS-IS is configured for IP routing. If a router runs IS-IS in Connectionless Network Service (CLNS) only, this condition will not occur. If a router has only a few IP routes in the IP routing table, this situation is less likely to occur than if there are many IP routes in the IP routing table.
  Workaround: Enter the shutdown command interface configuration command followed by the no shutdown interface configuration command on any interface that is configured for IS-IS.

Miscellaneous

  When printing is performed over asynchronous lines by using software flow control, large numbers of overruns occur. There is no workaround.
  On a Cisco 5200 series, 5300 series, or 5800 series router with NFAS configured, modems might not get released upon call completion or termination. There is no workaround.
  On a Cisco AS5800 series access server, you might see the following messages:
  %DIALn-3-MSG: %IPC-3-NOBUFF: The main IPC message header cache has emptied
  These messages are followed by a "-Traceback" line. The %DIALn-3-MSG is optional and the "n" indicates which dial shelf slot number the subsequent message originates from. Usually, the latter message is followed by:
  %DSIP-3-IPC_PORT: IPC subsystem API error(nip invoke ipc send), failed to get ipc_message
  This situation usually occurs when you use the autoselect ppp line configuration command, or the autoselect slip line configuration command, or the autoselect arap line configuration command. Under a high call rate, this situation might cause calls to not be established successfully.
   Workaround: Upgrade to a Cisco IOS release that contains the fix for this caveat.
  Formatting boot flash memory on a Cisco 7100 series router might disable access to boot flash memory. This situation occurs with crypto images in Cisco IOS Release 12.0(5)XE, Cisco IOS Release 12.0(5)XE1, and Cisco IOS Release 12.0(5)XE2, and the situation might occur with some noncrypto images.
  Workaround: Format Flash with a working image, or do not format Flash.
  A Cisco 2600 series router running Cisco IOS Release 11.3T might reload due to a SegV exception when a LAN Extender (LEX) interface is enabled.
  There is no known workaround.
  A Cisco router might reload because of a memory corruption. There is no workaround.
  Cisco Encryption Technology (CET) might not function correctly when configured on a Cisco router with generic routing encapsulation (GRE) tunnel. There is no workaround.
  In Cisco IOS releases prior to Cisco IOS Release 12.0(7)T, IP Security (IPSec) images for the Cisco 805 router do not support the virtual private dial-up network (VPDN) feature. There is no workaround.
  When you enable bridging on a Fast EtherChannel (FEC), the host functionality might not work properly, but bridging functionality is not affected. There is no workaround.
  A Cisco 4500 series router might slow down or pause indefinitely if you configure a mac-forward filter over transparent bridging on a Token Ring interface. There is no workaround.
  When Network Address Translation (NAT) and Cisco Express Forwarding (CEF) switching is enabled on Cisco 7200 series routers and Cisco 7500 series routers, all incoming packets take the fast-switching path instead of the CEF path. There is no workaround.
  Authentication, authorization, and accounting (AAA) for H.323 calls might show incorrect time stamps in comparison to the time synchronized through Network Time Protocol (NTP) as shown on the router. There is no workaround.
  The second port on a Cisco router with a PA-2T3 port adapter and a PA-2E3 port adapter will not come up if the first port is in a shutdown state.
  Workaround: Enter the no shutdown interface configuration command on the first port.
  A Cisco AS5300 series access server that is running Cisco IOS Release 12.0(07)T12.0(07)T or Cisco IOS Release 12.0(04)XH01 might send a LOOP_CLOSURE message for a busy call on a channel-associated signalling receive and transmit T1 trunk. This situation occurs with voice calls. There is no workaround.
  If you are entering the virtual-profile if-needed global configuration command on a Cisco AS5200, Cisco AS5300, or Cisco AS5800 access server that is running Cisco IOS Release 12.0(7.3)T, pings and IP packets fail because of reported encapsulation failures on the dialer or serial interface. This situation does not occur if virtual-profile if-needed is not configured, but the router performance is affected if virtual-profile if-needed is not configured. There is no workaround.
  A Cisco router that has been configured for point-to-point or static generic routing encapsulation (GRE) tunnel copies the type of service (ToS) field from the payload IP packet header to the outgoing delivery IP packet header. But this feature will not work for multipoint or dynamic GRE tunnels. There is no workaround.
  A Cisco router that is running with a high load and using IP Security (IPSec) might reload after an indeterminate amount of time (5 minutes to 5 hours or more). There is no workaround.
  When calls are redirected multiple times in the telephone network before arriving at a Cisco AS5300 access server by a 5ESS PRI, the redirecting number that is tunneled by H.323 will be the Original Called Number (OCN) (the first party to forward the call) rather than the Redirecting Number (RDN) (the last party that forwarded the call directly to the Cisco AS5300). But services that access the RDN will need the final RDN rather than the OCN. There is no workaround.
  If an EXEC session is running label switching commands such as show tag forwarding, then the router might reload when the no tag ip command is configured by another EXEC session. There is no workaround.
  When a gatekeeper does not receive responses to its Location Requests (LRQs) from a peer gatekeeper, the first gatekeeper might become unresponsive to requests from its own clients, and no new calls are admitted. If you enable debug ras on the gatekeeper, you will receive the following repeated messages:
  00:16:59: RASLib::RASSendLRQ:LRQ (seq# 8) sent to 1.14.93.83
  00:17:02: RASlib::ras_sendto:msg length 58 from 1.14.93.92:9679 to 1.14.93.83:1719
  00:17:02: RASLib::RASSendLRQ:LRQ (seq# 8) sent to 1.14.93.83
  00:17:05: RASlib::ras_sendto:msg length 58 from 1.14.93.92:9679 to 1.14.93.83:1719
  00:17:05: RASLib::RASSendLRQ:LRQ (seq# 8) sent to 1.14.93.83
  00:17:08: RASlib::ras_sendto:msg length 58 from 1.14.93.92:9679 to 1.14.93.83:1719
  No corresponding messages indicate that an LCF or LRJ response with the same sequence number (seq# 8 in the example) was received. Generally, when the communication path to the problem peer recovers, the situation should be resolved. However, under certain circumstances, you may need to enter the shutdown command followed by the no shutdown command on the gatekeeper. There is no workaround.
  If you enter the show users command, you might see that the Virtual Exec process is running on the modem even when no users are connected. This situation does not affect the allocation of the modem for subsequent sessions. The next user that logs in to the modem will clear the condition. There is no workaround.
  A Cisco router might not be able to properly decapsulate packets that are Authentication Header (AH)-encapsulated when flow switching is enabled. This situation is true for both flow switching and Cisco Express Forwarding (CEF) flow switching.
  Encapsulation Security Protocol (ESP)-encapsulated packets work correctly. Because ESP already provides authentication services along with the encryption, it is not necessary to use AH.
  Workaround: Do not specify AH in any transform proposals.
  Alternate Workaround: If AH-encapsulation is necessary, do not specify flow switching on the crypto interface.
  On a Cisco 7500 series router with Cisco Express Forwarding (CEF) enabled, if an output rate limit with qos-group is configured on a non-Versatile Interface Processor (VIP)-based interface, for example, ATM Interface Processor (AIP), and FDDI Interface Processor (FIP), then packets cannot be switched out of that interface.
  Workaround: Disable CEF.
  Lock and Key idle-timers won't get reset when there are packets match the dynamic ACLs created by Lock and Key. There is no workaround.
  When IOS Firewall is running on a Cisco router with Fast Ethernet subinterfaces using 802.1Q encapsulation, Context-based Access Control (CBAC) appears to run in fast-switching mode, but CBAC stops inspecting traffic after an indeterminate period of time. This situation does not occur in process switching mode.
  Workaround: Configure CBAC for process switching.
  Forwarding of bootp/dhcp address request UDP packets fail because of encapsulation failure.
   Workaround: Use Cisco IOS Release 12.0(5)T.
  A Cisco 7500 series router with Multiprotocol Label Switching (MPLS) configured might corrupt large packets that arrive on a channelized serial interface (CT1/CE1) and then switch to a tag MPLS-enabled interface. This situation occurs during the fragmentation process of the packet. Incoming tag MPLS packets are not affected.
  This situation only occurs for Route Switch Processor (RSP) switched packets. Versatile Interface Processor (VIP) distributed switching is not affected. There is no workaround.
  The txconn subsystem has functionality that disables TCP servers whenever a session to an associated host becomes unavailable. When this condition occurs, none of the active TCP client connections to the disabled server are reset, which leaves a client connected to a disabled server.
  In configurations where txconn is front-ended with Cisco DistributedDirector, this situation can prevent Cisco DistributedDirector from detecting the disabled server in a timely manner. There is no workaround.
  Txconn might not set the security type to NAP_SAME when it receives the already verified indicator in the FMH-5 security indicator field from TXSeries client connections. TXConn assumes the wrong security type program when the security subfields contains a userid. This situation only affects TXSeries client connections.
  When TXSeries is connected to txconn as a client, any attempt to connect specifying "already verified" might fail, depending on how security is configured on the Customer Information Control System (CICS) Server TELNET Application Segment (APPL).
  Workaround: When using TXSeries as a client, configure multiple virtual storage (MVS) CICS connection definitions for the router specifying no security. This workaround will allow TXSeries clients to connect.
  If you are using SNA Switching Services in an environment where sessions traverse High Performance Routing Broadband Network (HPR-BN) connections in the CS/390 hosts, HPR session activation might fail.
  Workaround: Use Intermediate Session Routing (ISR) over the BN connection instead of HPR.
  When compression packets are process-switched on a High-Speed Serial Interface (HSSI), and the ip rtp header-compression EXEC command, the ip route-cache interface configuration command, or the ip cef global configuration command is configured on the HSSI, packets will take the process path instead of the fast path. There is no workaround.
  When a first label switch router (LSR) is transmitting Multiprotocol Label Switching (MPLS) encapsulated IP frames to a second LSR that is removing the last label and transmitting the resultant IP frame onto an Inter-Switch Link (ISL), then IP packets less than 44 bytes will be received as cyclic redundancy check (CRC) errors. There is no workaround.
  On Cisco 7500 series routers that are running Cisco IOS Release 12.0 T, deleting a crypto map from an interface on a Versatile Interface Processor (VIP) will disable all crypto maps on all interfaces on that VIP. There is no workaround.
  Compressed packets cannot be fast-switched on a PA-MC-8E1 port adapter. If fast switching is enabled on a PA-MC-8E1 port adapter, the packets will only be compressed in the process path. There is no workaround
  When socket errors occur, the H.323 process that uses the socket event might cause the application programming interface (API) to spin with 98 percent to 100 percent CPU. There is no workaround.
  If there is a cable modem connected to another modem card and MIB walk docsIfCmtsObjects containing docsIfCmtsCmStatusTable, the Cisco uBR7246 Universal Broadband Router might appear to pause indefinitely.
  Workaround: Reload the router after performing an online insertion and removal (OIR).
  A Route Switch Module (RSM) that is running Cisco  IOS Release 12.0 T and is configured for IP routing on a Token Ring VLAN with the IP directed broadcast option enabled might send out subnet broadcasts that it receives on its own subnet back to this same subnet.
  Example: The RSM receives the subnet broadcast 10.10.10.255, and the RSM IP address is 10.10.10.1/24. The RSM will send back out on its own subnet a broadcast to the destination IP address 255.255.255.255, which is incorrect. On Layer 2, this frame goes to the destination MAC address ffff.ffff.ffff and is sent as an explorer on the Token Ring. The RSM should not do anything with the IP subnet broadcast as long as it is received on the interface that is directly connected to the router.
  Workaround: Create an access list and disallow subnet broadcasts of the router subnet to enter the RSM.
  Example: 
     interface vlan xxx type trbrf

     ip address 10.10.10.1 255.255.255.0

     ip access-group 101 in

     access-list 101 deny ip any host 10.10.10.255

     access-list 101 permit ip any any
  If you enter the show cable flap-list EXEC command while baseline privacy is active, the router will reload. There is no workaround.
  If a Dependent Logical Unit (LU) Requester/Dependent LU Server (DLUR/DLUS) connection fails, downstream LU-LU sessions might be terminated on downstream devices that are configured for ANS=CONTINUE, which should not normally be affected. There is no workaround.
  The cable source-verify dhcp cable interface configuration command does not function properly when baseline privacy interface (BPI) is active. There is no workaround.
  A Cisco 7200VXR router might experience a situation where switched virtual circuits (SVCs) are disconnected intermittently and then recovered after 7 to 20 hours. There is no workaround.
  If you configure a subinterface on a Cisco 2600 series router for operation, administration, and maintenance (OAM), and the permanent virtual circuit (PVC) bounces, the PVC and the interface will not come back up unless you enter the shutdown interface configuration command followed by the no shutdown interface configuration command on the interface. This situation does not occur if the PVC is configured without OAM. There is no workaround.
  After you reload a Cisco router, the PPP virtual template command protocol disappears under the virtual circuit (VC) class ATM. This condition causes the PPP sessions to lose their binding with the virtual template. This situation occurs only if you have configured 'encapsulation aal5snap' in the ATM VC class, and does not occur if the VC class is configured as aal5mux. There is no workaround.
  A Route Switch Module (RSM) might corrupt the Routing Information Field (RIF) on a Logical Link Control (LLC1) frame when running data-link switching (DLSw). There is no workaround.
  A Route Switch Processor (RSP) that is configured with Multiprotocol over ATM (MPOA) clients might reload. There is no workaround.
  The show interface display EXEC command might show that a Cisco 2-port High-Speed Serial Interface (HSSI) or a Cisco PA-2T3 port adapter is experiencing abnormal transmit underruns on some Versatile Interface Processors (VIPs) with fast protocol control information (PCI) transactions. There is no workaround.
  MC16E cards might experience difficulties when you configure downstream modulations. Symptoms of the problem include ranging bursts that do not arrive upstream.
  Workaround: Use the shut command and the no shut command on the interface.
  This workaround adds extra physical layer (PHY) reconfiguration when you perform a hardware reset of the MAC, but this situation only occurs when a downstream PHY parameter is modified.
  A Cisco router that is configured for Systems Network Architecture Switching Services (SNASw) might reload repeatedly until SNASw functionality is stopped. There is no workaround.
  A memory leak might occur when the interface that is trying to send traffic by an ATM interface supports only packets (not particles) and either the cable is pulled out of ATM interface or the virtual circuit (VC) set up to transfer is in a down state. This situation does not occur during normal operation. There is no workaround.
  Pings on a Cisco 7500 series tag-switching router that is running Cisco IOS Release 12.0 S, Cisco IOS Release 12.1, or Cisco IOS Release 12.0 ST might fail if a packet comes in as an IP packet on an ATM interface, the packet gets route-cache switched, the packet gets fragmented, and the fragments go out as tag packets through a serial interface.
  Workaround: Configure distributed Cisco Express Forwarding (dCEF) on a Cisco ATM port adapter.
  Swapping a cable line card that is not supported in a Cisco uBR7200 chassis with a supported line card several times might cause the router to reload in the routine to add a subinterface entry in the ifTable. There is no workaround.
  Configuring a downstream channel ID on a cable interface might cause cable modems to hang up while waiting for valid channel Manufacturing Automation Protocol (MAP) messages. In this state, the cable modem termination system (CMTS) sends the MAP messages, but a stale parameter in the MAP messages causes the cable access routers to ignore these MAP messages.
  Workaround: Enter the clear interface EXEC command on the CMTS after configuring the downstream channel ID. This workaround will cause the MAP messages to refresh and pick up the latest UCD change count parameter.
  On Cisco routers that are running Cisco IOS Release 12.0(8), IP Security (IPSec) will send in the clear packets that need to be encrypted and fragmented. There is no workaround.
  A Cisco PA-2T3 port adapter might show increasing overruns in the show interfaces EXEC output when one of the two ports is in DOWN state.
  Workaround: Put the DOWN port into ADMIN SHUT state.
  If the signal level for the macro instruction (MF) digits is too low, it will disappear before reaching the PBX. Any gain or attenuation on the voice port only affects the audio stream. The cable length parameter only adds attenuation.
  Workaround: Add a configuration parameter to change the level of the signal for the MF digits.
  Multiprotocol over ATM (MPOA) shortcuts might not check for the frame sizes being sent on them. If the frame size exceeds the maximum transmission unit (MTU) size of the egress Multiprotocol over ATM Client (MPC), egress MPC drops the packet, resulting in the loss of connectivity.
  Workaround: Configure matching MTUs on either side of shortcuts (for example, all LAN Emulation Clients (LECs) that are served by ingress and egress MPCs should have the same MTU). If MTUs cannot be altered, then do not configure MPOA on both sides.
  A resign message is not sent when the standby IP address of a Hot Standby Router Protocol (HSRP) group is deconfigured. This situation will delay the standby router taking over as the active router until the active timer expires in the standby router. This time is determined by the standby hold down time, which is 10 seconds by default. During this time, packets that are forwarded to the standby virtual address will not reach their destination.
  Workaround: Lower the priority of the group to below the priority of the standby router before deconfiguring the standby IP address.
  After an initial program load of a Dependent Logical Unit Server (DLUS) system, SNA switch DLUR routers might exhibit a "Pending Inactive" state for the DLUS as seen in the output of the show snasw dlus EXEC command.
  Workaround: Stop and start the SNA switch subsystem on the router.
  The ip rtp priority interface configuration command does not work in Cisco IOS Release 12.1(09). There is no workaround.
  A Cisco SC2200 series signalling controller might forward the call setup signalling to a Cisco AS5800 series access server. When a call comes into the AS5800, the AS5800 access server reloads. There is no workaround.
  A Cisco router might reload because of memory corruption when it receives a large LOCATE message from an end station. The router might reload repeatedly until the end station is stopped.
  Workaround: Reduce the number of transmission group (TG) vectors reported by the end station on the LOCATE message.
  A Cisco Catalyst 5000 might treat the link between the physical interface and the port channel interface as a multidrop link when the MAC address of members of a port channel is different from the MAC address of the port channel itself.
  Workaround: Delete the members of the port channel and then add then add them back in.
  IKE Tunnel Endpoint Discovery (TED) might fail with the following error message:
  %CRYPTO-4-IKMP_BAD_MESSAGE
  There is no workaround.
  A route switch processor (RSP) with an interface configured with IP Security (IPSec) crypto map and the switching mode is Cisco Express Forwarding (CEF), the RSP might reload trying to decrypt an IPSec packet. This situation is only seen when CEF switching is on. There is no workaround.
  After a Cisco router reloads, it might remove the "service-policy output" configuration line from the serial interface. In this situation, the max-reserved-bandwidth interface configuration command is entered on this interface after the service-policy global command. Because the service-policy global command is read by the router first, the router functions as if only 75 percent of the serial interface bandwidth is available, which is the default for the max-interface-bandwidth interface configuration command. When the router is functioning at this level, the router recognizes there is not enough bandwidth for class-based weighted fair queueing (CBWFQ), and the router removes the "service-policy output" line from the configuration. If the max-interface-bandwidth interface configuration command was read by the router first, this situation would not occur because the CBWFQ configuration would be valid.
  Workaround: Create a dummy subinterface off the main interface. No configuration of this subinterface is necessary and a permanent virtual connection (PVC) does not need to be assigned. However, when the router reloads, the service-policy global command under the serial interface will not be removed.
  Received packets that have been padded by the previous hop are corrupted by the Multiprotocol Label Switching (MPLS) distributed Cisco Express Forwarding (dCEF) label imposition code and will result in IP checksum errors at their final destination or at an intermediate hop, depending on the network configuration.
  Workaround: Disable dCEF globally or on a per-VIP interface basis.
  If the service compress config is enabled, entering the show startup-config configuration command or the show config configuration command will result in memory leak. There is no workaround.
  A Cisco 3660 router might display a duplicate name for the second BRI WAN interface card (WIC) module if it is installed in a slot that already has one BRI module. There is no workaround.
  A Cisco router might reload while running SNA Switching (SNASw). This situation is caused by a rare condition in Dependent Logical Unit Requester (DLUR) when the links to downstream physical unit (DSPUs) are going up and down very rapidly. Under these circumstances, it is possible for DLUR to receive a DSPU_INACTIVE signal from configuration services, fail to correlate it to the appropriate physical unit (PU) CB, and then release it. The next signal received from convergence sublayer (CS) for this PU is a DSPU_ACTIVE, which DLUR is not expecting. This situation causes an invalid state error in DLUR, which leads to a reload shortly afterward. There is no workaround.
  When configured to provide access control, the Kerberos client on Cisco products will fail all authentications when the expiration of the credential falls between January and February of a leap year.
  Workaround: Choose an alternate form of authentication such as TACACS+ or RADIUS.
  A Cisco router might reload if you enter the show cable flap-list EXEC command. The probability of the router reloading can be reduced significantly if the flap list aging time is kept at its default or set to a very high value such as 86400. Also, the probability of the router reloading is greatly increased if you leave the command output paused because of automore processing (for example, where --More-- is displayed). There is no workaround.
  The default file system is not recognized by the delete EXEC command if only the filename is specified. The entire path of the file to be deleted (for example, delete filesystem:filename) must be included in the command line. There is no workaround.
  A Cisco 7200 series router that is running ATM LAN emulation (LANE) with a PA-A3 port adapter might reload if a packet with invalid encapsulation is sent to the port adapter. There is no workaround.
  Shutting down an recEive and transMit (E&M) port on a Cisco 2600 series router or a Cisco 3600 series router might result in the following error message:
  %VTSP-4-FSM_BAD_EVENT: Invalid FSM Input on channel UNKNOWN: state=-1 event=8
  There is no workaround.
  A Cisco router that receives a recent upstream throughput enhancement might reload under certain specific upstream configurations or timing state conditions. This situation occurs when an upstream port is configured with nondefault upstream parameters that cause a single request opportunity to span more than one minislot. A sample nondefault upstream setting that can cause this situation are as follows:
  If the upstream channel is using a nondefault setting like the one described, the following timing conditions can trigger a reload:
  Even if there is no console activity, the router might reload, although this situation is rare.
  Workaround: Maintain all upstream parameters at default values.
  Pinging from a Customer Edge (CE) router to its own address or using Telnet to reach the router might cause the attached Provider Edge (PE) router to reload. There is no workaround.
  A Cisco router might reload if the show ip cef EXEC command is entered while the routing table is changing. There is no workaround.
  On a Cisco router that is running Cisco IOS Release 12.0, a Cisco Encryption Technology (CET) encrypted packet coming in while the router is restarting might prevent encryption from working on that interface for 15 minutes.
  When this situation occurs, the show crypto map privileged EXEC command will show the Connection ID as BAD, and the connection will not be listed by either the show crypto cisco connections privileged EXEC command or the show crypto engine connections active privileged EXEC command. There is no workaround.
  A Cisco AS5800 access server that is acting as a gateway might fail CODEC negotiation on G.723 calls that match VoIP dialpeer 0.
  Workaround: Configure a matching VoIP dialpeer, and configure a CODEC class with the CODECs that are being used.
  Packets that have been padded by the previous hop and then received by a Versatile Interface Processor (VIP) Ethernet, Fast Ethernet, or Gigabit Ethernet might be dropped if they are supposed to be processed by the Route Processor (RP) instead of the VIP.
  Workaround: Disable distributed Cisco Express Forwarding (dCEF) on the ingress interface.
  SNA switch currently requires an UNBIND response from the secondary logical unit (SLU) after sending it a request. Some devices do not return this response if the request is immediately followed by a DACTLU. There is no workaround.
  Under rare circumstances, cable access server entries might exist on multiple interfaces if the modem is moved from one interface to another, and certain timing conditions are met. When cable bundling is used, this situation might result in loss of connectivity for the modem.
  Workaround: Disconnect the cable to the modem, and then enter the clear cable modem {mac-addr} reset privileged EXEC configuration command before moving the modem to another interface.
  Alternate Workaround: Disconnect the cable to the modem first, then clear the modem status with the clear cable modem {mac-addr} reset privileged EXEC configuration command before moving it to another interface.
   A Cisco AS5800 series network access server (NAS) might experience dial-out failures, (Layer 2 goes down). The D-channel interface needs to be cleared to recover.
  Workaround: Upgrade to Cisco IOS Release 12.1(1), 12.1(2), 12.1(1)T, or 12.1(2)T.
  Traffic shaping might cause a Cisco 3640 router to reload under the following conditions:
  There is no workaround.
  If there is a device connected to a noncable interface on a Cisco uBR7200 series universal broadband router (uBR), and if the uBR acts as the Dynamic Host Configuration Protocol (DHCP) relay agent or as the DHCP server for the device, the system might reload when a DHCP relay is sent to it.
  Workaround: Do not use a Cisco uBR7200 as a DHCP relay agent or as DHCP server for devices on noncable interfaces.
  When a Customer Information Control System (CICS) term is forced to use ABEND on the host, the CICSB server will be disabled because the forced ABEND is interpreted as a host failure, and all connections to the server are brought down. Users that are connected to a TX Server might experience outage any time a transaction that is being run on the same server ABENDs. There is no workaround.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1 with an IPSec Service Adapter (ISA) board might reload if you enter the clear crypto sa global configuration command. There is no workaround.
  A Common Object Module Transaction Interface (COMTI) session object might not check in the license if an exception error occurs. In this situation, the license can become exhausted, and clients cannot connect. There is no workaround.
  Security is not functioning properly with Common Object Module Transaction Interface (COMTI) servers. Transactions that require user IDs and passwords cannot be executed through the Cisco Transaction Connection TXConn subsystem. There is no workaround.
  Serial drivers might cause a memory leak when a reparented packet is sent. There is no workaround.
  If you enable distributed Cisco Express Forwarding (CEF) on a Cisco 7500 series router, and there are one or more prefixes in the IP routing table with the maximum six paths, the router might reload with the following error message:
  %SYS-2-CHUNKBADMAGIC: Bad magic number in chunk header...
  There is no workaround.
  When you remove modem cards on a Cisco AS5800 series access server without updating the startup configuration before the next reload, the access server might encounter a software exception when the access server receives the first modem call and will reload unexpectedly.
  Workaround: Use the write memory command to update the startup configuration before the next router reload so that the pool-range configuration command reflects the existing hardware configuration.
  If a dynamic crypto map has multiple entries, internet key exchange (IKE) negotiation might fail with the Tunnel Endpoint Discovery (TED) peer. There is no workaround.
  Spectrum Management task might use many CPU cycles and cause CPU utilization to increase. There is no workaround.
  In Cisco IOS Release 12.0(7)T, there have been rare occurances of network access server (NAS) restarts when Stack Group Bidding Protocol (SGBP) is configured. The following SGBP error message occurs in the output of the debug sgbp error command:
  %SGBP-3-INVALID:MQ message with Cancel
  This situation occurs if PPP calls drop before the SGBP bidding process finishes. With either the multilink bundle-name both or the multilink bundle-name endpoint commands configured, the chance of occurance increases.
  Workaround: Upgrade to Cisco IOS Release 12.1(2)T or a later release.
  DBConn TCP/IP passthru connections might cause a Cisco router to reload if the host disconnects prematurely. There is no workaround.
  In a situation where a Customer Edge (CE) router is connected to two Provider Edge (PE) routers, if a third PE changes its selected route to the CE router, it might fail to update the Virtual Private Network (VPN) label for the route, resulting in loss of connectivity to that CE.
  Workaround: Use the clear ip route command to clear the route.
  If Cisco Express Forwarding (CEF) is not enabled and Resource Reservation Protocol (RSVP) over ATM is used, excessive switched virtual circuits (SVCs) are created. There is no workaround.
  A Cisco uBR7200 series universal broadband router that is configured with a Cisco cable modem termination system (CMTS) is not protected against rogue cable modems that exhibit non-Data-over-Cable Service Interface Specifications (DOCSIS)-compliant behavior at initial ranging. There might be some modems on an upstream port that range with a negative timing offset. Such rogue modems can deceive the CMTS Map building code into using a large Map Advance when the dynamic Map Advance algorithm (the default) is active on each upstream port.
  The result of this situation is that all the other modems on that upstream port will be unable to send data upstream to the CMTS and will go offline. The router will generate a warning log message every time a modem with bad timing offset is detected. A sample warning message is follows:
  %UBR7200-5-BADTXOFFSET:Bad timing offset -10 detected for cable modem 0010.9500.0a6a
  Workaround: Configure static Map Advance on each cable interface of the router by using the cmts(config-if)# cable map-advance static command-line interface (CLI).
  An R7K revision 1.X processor might ignore instructions on return from an exception. This situation does not occur in the R7K revision 2.1 processor.
  Workaround: Place a synchronization instruction immediately before any errata instructions.

Novell IPX, XNS, and Apollo Domain

  Tracebacks can occur on a Cisco Catalyst 5500 switch if the following actions are performed in quick succession:
  Workaround: Remove IPX from the interface before deleting the interface.
  A Cisco router might exhibit ALIGN-3-CORRECT and ALIGN-3-TRACE errors if it is upgraded to Cisco IOS Release 12.0(7.3)T. There is no workaround.

TCP/IP Host-Mode Services

  A Cisco router might lose memory or reload during periods of heavy traffic while using a voice application or the gateway application. There is no workaround.

Wide-Area Networking

  A Cisco router might experience call failure if the dialer is set to a maximum of 255 calls and the number of calls exceeds that number. There is no workaround.
  Deconfiguring a nonprimary span in an Non-Facility Associated Signaling (NFAS) group might cause the number of links that the dialer recognizes as available to the remaining rotary group to drop by too much (possibly to zero).
  Workaround: Deconfigure and reconfigure the whole NFAS group.
  Ping packets for protocols other than IP might be dropped if you configure all BRI interfaces with the isdn fast-rollover-delay 1 interface configuration command, and you are running Cisco IOS Release 12.0(6.2)T. The first several hundred ping packets pass, but then the later packets are dropped. In this situation, pings will fail even if you remove the isdn fast-rollover-delay from all BRI interfaces. You will need to reload the router. This situation occurs with IPX, AppleTalk, CLNS, DECnet, VINES, and XNS in legacy DDR, but does not occur in IP and bridging. There is no workaround.
  A Cisco 7200VXR router might reload with a bus error because the packet-by-packet compression code is being passed packets that contain particles. There is no workaround.
  Under rare conditions, a Multiprotocol over ATM client/Multiprotocol over ATM server (MPC/MPS) will receive a route change from the Next Hop Resolution Protocol (NHRP) side of the cloud but will not purge the affected cache entry. There is no workaround.
  A Cisco router might experience a software-forced reload because of watchdog timeout if thousands of switched virtual circuits (SVCs) or LAN Emulation (LANE) clients are configured on a single ATM interface.
  Workaround: Reduce the number of SVCs or LANE clients that are configured on the ATM interface.
  Permanent virtual circuit (PVC) bumping does not function properly if you configure a VC that is down to be protected and then make it not protected again.
  Workaround: Enter the shutdown command followed by the no shutdown command on the subinterface.
  The dialer idle-timeout does not reset for the inbound interesting packets, for all the serial interfaces except the first interface each of which is configured with dialer idle-timeout either interface configuration command for the case when fast switching is allowed. As a result all but the first serial interface drop the call.
  Workaround: Use process switching instead of fast switching in which case all serial interfaces reset idle timer as expected.
  A Cisco router that is running Cisco IOS Release 12.0(6.5)T and later releases might experience traceback errors on interfaces where Multilink PPP encapsulation was removed before weighted fair queueing (WFQ) was added.
  Workaround: Shut down the interface before issuing the no ppp multilink interface configuration command or the fair-queue interface configuration command.
  A Cisco 7200 series router that is running Cisco IOS Release 12.0(7) might reload with a bus error at PC 0x60584578, address 0xD0D0D51. There is no workaround.
  The Integrated Local Management Interface (ILMI) on a Cisco router might respond with the invalid value 0 instead of noSuchName when the interface is queried by the peer switch ILMI on ILMI MIB objects atmfAtmLayerIlmiVersion and atmfAtmLayerNniSigVersion. There is no workaround.
  Removal of the RLM GROUP from configuration will cause the serial interface to be removed. There is no workaround.
  The frame-relay de-group 1 local-dlci interface configuration command might disappear from the running configuration and stop working for the referenced data-link connection identifier (DLCI) if you configure the frame-relay inverse-arp interface configuration command and the corresponding permanent virtual connection (PVC) status changes to Inactive or Deleted.
  Workaround: Use the static frame-relay map interface configuration command and reference the same DLCI that is used for the frame-relay de-group 1 local-dlci interface configuration command.
  A Cisco router that is running Cisco IOS Release 12.0(7)S and supports a large number of ISDN Primary Rate Interfaces (PRI)s might reload due to low memory. There is no workaround.
  Set asynchronous balanced mode extended (SABME) commands might be ignored by a Cisco AS5300 universal access server for extended periods of time after a link failure. When the debug isdn event or debug isdn q931 commands are enabled, a "Syncing discards" message will be displayed for each SABME. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0 or later releases might reload because of a bus error at the symbol dialer_clear_map. There is no workaround.
  A Cisco 7206 router that is running Cisco IOS Release 12.0(8.0.1)T might experience a software forced reload. There is no workaround.
  A Cisco AS5300 access server with no service align detect configured will reload if you enter the no pri-group command on a controller that has no pri-group configuration.
  Workaround: Enable service align detect in configuration mode. This will prevent the reload, but might result in an error message for accessing invalid memory.
  If you are running SNA switching with a Fast Ethernet connection to the host, the connection will function normally until an UNBIND is sent at the end of the user session. The UNBIND rsp is sent to the non-canonical (Token Ring format) MAC address; it is not recognized by the CIP, and the Logical Link Control (LLC) session pauses indefinitely.
  Workaround: Use a MAC address that bit-swaps back to itself (such as 0000.6666.6666).
  On a Cisco 804 router that is running Cisco IOS Release 12.0(5)T or Cisco IOS Release 12.0(6.4)T, the IP image might reload with a memory protection violation if you try to load a web page through ISDN. In this situation, the router displays the following trace message:
  WARNING! Illegal read access
  Memory Protection Violation: epc =0x001A4098, location of fault eva =0x00000???, read fault address error=0x00000082, illegal address pti =0x00000000 pte =0xFFFFF100
  Memory Protection Violation: epc =0x001A417C, location of fault eva =0x00000???, write fault address error=0x00000080, illegal address pti =0x00000000 pte =0x02102100
  There is no workaround.
  A Cisco router might reload when initiating a mnemonic connection that is defined by an X.25 hostname address and optional call user data (CUD).
  Workaround: Precede the name with a pad EXEC command before initiating the outgoing mnemonic connection. For example, if an X.25 host name "amble" is defined by the x25 host amble 12345 cud go global command, use the pad amble EXEC command instead of the one-word mnemonic name "amble".
  If the Layer 2 link is lost while using X.25 over D channel, a Cisco router will attempt to reestablish the link by using the terminal endpoint identifier (TEI) value of B channel. The router will need to be reloaded to reestablish the X.25 connection. There is no workaround.
  When X.25 over TCP (XOT) connects X.25 links that are configured for modulo 128, the packet assembler/disassembler (PAD) session will pause indefinitely. In this situation, entering the show x25 vc EXEC command will return a "Window is closed" message. There is no workaround.
  A Cisco 7500 series router with a VIP2-50 Versatile Interface Processor and a PA-A3 port adapter does not react to available bit rate (ABR) explicit rate (ER) congestion marking. The output rate of an ABR connection does not decrease after receiving a resource management (RM) cell with the ER field value lower than the commitment, concurrency, and recovery (CCR) value. There is no workaround.
  In a large-scale dial-out environment, the convergence of the routing protocol might not redistribute the host routes quickly enough, which results in both an incoming and an outgoing connection on different platforms to the same destination. There is no workaround.
  Timer data structures used by the dialer component might become corrupted. The corruption might occur because of operator-initiated actions that involve the removal of the dialer function on an interface. Examples of these actions would be configuring leased-line BRI, or entering the no dialer in-band interface configuration command. The symptom might not occur for days or weeks after the operator action that caused it. Due to this condition, these actions should be avoided. There is no workaround.
  On a Cisco router with dialer profiles, dialer load-threshold, and no multilink, an overload call might not be done even though the load threshold is exceeded. This situation occurs when an incoming call is received with a particular caller ID but there is no dialer string configured with the same phone number.
  Workaround: Initiate first and overload calls from the same router.
  A Cisco router that is running Cisco IOS Release 12.0(8) and is configured with a LAN Extender (LEX) interface might display any of the following error messages:
  %LINK-2-INTVULN: In critical region with interrupt level=0, intfc=Lex0
  -Process= "RIP Send", ipl= 0, pid= 73
  -Traceback= 1A2006 66872 1A697A 22F1E4 22E924 22EAC0 380A36
  %LINK-2-INTVULN:In critical region with interrupt level=0, intfc=Lex0
  -Process= "IP Input", ipl= 0, pid= 12
  -Traceback= 1A2006 66872 1A697A 22F1E4 224EAA 2257E0 224902 224A5C 224B80
  %LINK-2-INTVULN:In critical region with interrupt level=0, intfc=Lex0
  -Process= "ARP Input", ipl= 0, pid= 7
  -Traceback= 1A2006 66872 1A697A 1FE4AC 2398C4 239E80 1FE6D
  There is no workaround.
  A NAS might send incorrect RADIUS accounting packets to the RADIUS server when a virtual private dialup network (VPDN) is enabled. In this situation, the acct-input-octets value is always -1. There is no workaround.
  If a dialer profile (with multiple BRIs in pool) is configured with a very low load threshold value using the dialer load-threshold command and if PPP Multilink is also configured, then all the channels except the B channels on one BRI are not disconnected. This situation might cause the B channels in the pool to flap and could also result in the failure of subsequent pings to the same link.
  Workaround: Raise the load threshold to a higher value.
  When using dialer profiles, the dialer state could get stuck in the call pending state. In this situation, outgoing ISDN calls can no longer be initiated.
  Workaround: Reload the router.
  Under certain conditions, the frame-relay inverse-arp command does not function properly for IPX, Novell, DECnet, and VINES for multipoint interfaces. This situation is more likely to occur when IP is running on the interface.
  Workaround: Create a static map for IPX, DECnet, and other protocols by entering the frame-relay map interface configuration command, or change the interface from multipoint to point-to-point.
  When you run Redundant Link Manager (RLM) and Non-Facility Associated Signaling (NFAS) on a Cisco 5000 series network access server (NAS), you can see Customer Information Control System (CICS) in a state of LOCMAN because of the order that the NFAS group was configured on the controllers.
   Workaround: Configure NFAS groups consecutively, starting with the first controller.
  A Cisco 1750 router or 3660 router that is configured for voice might reload occasionally with a SegV exception or a bus error pointing to the function dequeue when the ISDN is up. There is no workaround.
  Multilink PPP max links values might not be cleared on virtual access after disconnection. This condition will result in the value being applied to the next user of that virtual access. This situation also occurs with Multilink PPP min links and Multilink PPP load-threshold values. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.0(5)T or later and is performing FRF.12 fragmentation on the permanent virtual connection (PVC) might alter the payload of resent frames that are originated by the router. This condition will not affect traffic that is being routed by the router, but it might affect features like data-link switching (DLSw), serial tunnel (STUN), Block Serial Tunneling (BSTUN), and Border Gateway Protocol (BGP), and traffic that consists of resent TCP packets sourced by the router.
  Workaround: Disable fragmentation in the PVC.
  When you change the maximum transmission unit (MTU) size while configuring a BRI as leased line, the IP MTU size does not automatically adapt to the new size as will normally happen on other interfaces. The IP MTU size can be configured manually, but after a read-write community string (wr) and a reload, the IP MTU values will get reset to 1500 again. There is no workaround.
  A Cisco router might become unresponsive if you use protocol translation to X.25. The console continuously displays the following error message:
  SYS-2-INSCHED: event dismiss within scheduler
  There is no workaround.
  Termination of Multilink PPP sessions over Layer 2 Forwarding Protocol (L2FP) or Layer 2 Tunneling Protocol (L2TP) on a Cisco 7100 series router might cause the input queue to go to 76/75 on the virtual access interfaces. Traffic will then fail to pass and the output of debug vpdn {l2x-error} will display the following message:
  L2X: Coalesce of packet failed.
  Workaround: Disable Multilink PPP.
  A Cisco router might fail to bring up Layer 2 channels for National ISDN and 5ESS Custom switch profiles where dual terminal endpoint identifiers (TEIs) are used. There is no workaround.
  A Cisco router that is connected to ISDN BRI lines that deactivate Layer 1 (typically done by European telcos to save power on BRI lines when no ISDN calls are active) and that use dialer interfaces (either legacy DDR or dialer profiles) will not be able to dial out.
  Workaround: Remove the dialer interfaces and configure the BRI interfaces instead.

Obtaining Documentation

World Wide Web

You can access the most current Cisco documentation on the World Wide Web at http://www.cisco.com, http://www-china.cisco.com, or http://www-europe.cisco.com.

Documentation CD-ROM

Cisco documentation and additional literature are available in a CD-ROM package, which ships with your product. The Documentation CD-ROM is updated monthly. Therefore, it is probably more current than printed documentation. The CD-ROM package is available as a single unit or as an annual subscription.

Ordering Documentation

Registered CCO users can order the Documentation CD-ROM and other Cisco Product documentation through our online Subscription Services at http://www.cisco.com/cgi-bin/subcat/kaojump.cgi.

Nonregistered CCO users can order documentation through a local account representative by calling Cisco's corporate headquarters (California, USA) at 408 526-4000 or, in North America, call 800 553-NETS (6387).

Obtaining Technical Assistance

Cisco provides Cisco Connection Online (CCO) as a starting point for all technical assistance. Warranty or maintenance contract customers can use the Technical Assistance Center. All customers can submit technical feedback on Cisco documentation using the web, e-mail, a self-addressed stamped response card included in many printed docs, or by sending mail to Cisco.

Cisco Connection Online

Cisco continues to revolutionize how business is done on the Internet. Cisco Connection Online is the foundation of a suite of interactive, networked services that provides immediate, open access to Cisco information and resources at anytime, from anywhere in the world. This highly integrated Internet application is a powerful, easy-to-use tool for doing business with Cisco.

CCO's broad range of features and services helps customers and partners to streamline business processes and improve productivity. Through CCO, you will find information about Cisco and our networking solutions, services, and programs. In addition, you can resolve technical issues with online support services, download and test software packages, and order Cisco learning materials and merchandise. Valuable online skill assessment, training, and certification programs are also available.

Customers and partners can self-register on CCO to obtain additional personalized information and services. Registered users may order products, check on the status of an order and view benefits specific to their relationships with Cisco.

You can access CCO in the following ways:

You can e-mail questions about using CCO to cco-team@cisco.com.

Technical Assistance Center

The Cisco Technical Assistance Center (TAC) is available to warranty or maintenance contract customers who need technical assistance with a Cisco product that is under warranty or covered by a maintenance contract.

To display the TAC web site that includes links to technical support information and software upgrades and for requesting TAC support, use www.cisco.com/techsupport.

To contact by e-mail, use one of the following:

Language

E-mail Address

 English


 tac@cisco.com


 Hanzi (Chinese)


 chinese-tac@cisco.com


 Kanji (Japanese)


 japan-tac@cisco.com


 Hangul (Korean)


 korea-tac@cisco.com


 Spanish


 tac@cisco.com


 Thai


 thai-tac@cisco.com

In North America, TAC can be reached at 800 553-2447 or 408 526-7209. For other telephone numbers and TAC e-mail addresses worldwide, consult the following web site: http://www.cisco.com/warp/public/687/Directory/DirTAC.shtml.

Documentation Feedback

If you are reading Cisco product documentation on the World Wide Web, you can submit technical comments electronically. Click Feedback in the toolbar and choose Documentation. After you complete the form, click Submit to send it to Cisco.

You can e-mail your comments to bug-doc@cisco.com.

To submit your comments by mail, for your convenience many documents contain a response card behind the front cover. Otherwise, you can mail your comments to the following address:

Cisco Systems, Inc.
Document Resource Connection
170 West Tasman Drive
San Jose, CA 95134-9883

We appreciate and value your comments.





hometocprevnextglossaryfeedbacksearchhelp
Posted: Thu Sep 28 09:30:31 PDT 2000
Copyright 1989-2000©Cisco Systems Inc.