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

Table of Contents

Caveats for Cisco IOS Release 12.1 T

Caveats for Cisco IOS Release 12.1 T

September 17, 2000

Text Part Number 78-10764-03 Rev. B7

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 hardcopy documents were printed.

This document lists severity 1 and 2 caveats for Cisco IOS Release 12.1 T, up to and including Release 12.1(3)T. 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 improve this document, we would appreciate 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:

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 Cisco Connection Online (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 T, 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(3)T


 July 24, 2000


 Release Notes for the Cisco 1000 Series  Routers for Cisco IOS Release 12.1 T


 Release12.1(3)T


 July 24, 2000


 Release Notes for the Cisco 1400 Series  Routers for Cisco IOS Release 12.1 T


 Release12.1(3)T


 July 24, 2000


 Release Notes for the Cisco 1600 Series  Routers for Cisco IOS Release 12.1 T


 Release12.1(3)T


 July 24, 2000


 Release Notes for the Cisco 1700 Series  Routers for Cisco IOS Release 12.1 T


 Release12.1(3)T


 July 24, 2000


 Release Notes for the Cisco 2500 Series  Routers for Cisco IOS Release 12.1 T


 Release12.1(3)T


 July 24, 2000


 Release Notes for Cisco 2600 Series for  CiscoIOS Release12.1


 Release12.1(3)T


 July 24, 2000


 Release Notes for Cisco 3600 Series for  CiscoIOS Release12.1


 Release12.1(3)T


 July 24, 2000


 Release Notes for Cisco 4500 Series for  CiscoIOS Release12.1


 Release12.1(3)T


 July 24, 2000


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


 Release12.1(3)T


 July 24, 2000


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


 Release12.1(3)T


 July 24, 2000


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


 Release12.1(3)T


 July 24, 2000


 Release Notes for the Cisco MC3810 for  CiscoIOS Release12.1


 Release12.1(3)T


 July 24, 2000

Open Caveats—12.1(3a)T4

Cisco IOS Release 12.1(3a)T4 is a rebuild release for Cisco IOS Release 12.1(3)T.

  The Router Auto Configuration functionality on cable modem interfaces causes problems with the Data Over Cable Service Interface Specification (DOCSIS) registration process and causes a double Dynamic Host Configuration Protocol (DHCP) lease to occur. DOCSIS Acceptance Test Plan (ATP) tests have failed and IP Security (IPSec) configurations are broken.
  Workaround: Do not configure static IP addresses in the DHCP server for the given modem's MAC address.
  
  The Router Auto Configuration functionality has been added to support cable modem interfaces. The behavior of this functionality causes problems with the DOCSIS registration process and causes a double DHCP lease to occur. DOCSIS ATP tests have failed and IPSec configs are broken. The only workaround is to not configure static IP addresses in the DHCP server for the given modem's MAC address.

Resolved Caveats—12.1(3a)T3

Cisco IOS Release 12.1(3a)T3 is a rebuild release for Cisco IOS Release 12.1(3)T. The caveats in this section are resolved in Cisco IOS Release 12.1(3a)T3 but might be open in previous Cisco IOS releases.

  The PA-MC-8T1 port adapter reports that Firmware paused indefinitely and then reloads the router. There is no workaround.
  The service-policy {output} global configuration command on a PA-A3 port adapter is deleted after a link status change. This situation causes Class-Based, Weighted Fair Queueing (CBWFQ) to fail. There is no workaround.

Resolved Caveats—12.1(3a)T2

Cisco IOS Release 12.1(3a)T2 is a rebuild release for Cisco IOS Release 12.1(3)T. The caveats in this section are resolved in Cisco IOS Release 12.1(3a)T2 but might be open in previous Cisco IOS releases.

  A Cisco 7200 series router that is running Cisco IOS Release 12.1(3)T and that is using Multiprotocol Label Switching (MPLS) on a Virtual Private Network (VPN) might not be able to forward IP packets received on a VPN Routing/Forwarding (VRF) interface to the MPLS core. The router reloads because of a bad pointer when the router forwards packets on the IP to MPLS path.
  Workaround: Upgrade to a Cisco IOS release that contains the fix.

Open Caveats—12.1(3)T

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

Access Server

  If unbundled Modem ISDN channel aggregation (MICA) firmware has been configured on a Cisco AS5800 series access server, the firmware is downloaded serially to each Service Processing Element (SPE) across the AS5800. Each SPE takes about 20 seconds, so a fully populated AS5800 using MICA might take up to 30 minutes before all SPEs are available for use. There is no workaround.

Basic System Services

  When certain vendor Point-to-Point Protocol (PPP) compression is configured, Cisco RADIUS misinterprets the compression as a STAC compression. There is no workaround.
  A Cisco router might reload with a Read Bus Error exception in cdp_cache_ager when there are several 10 to 15 minute long modem calls (enough for 16 T1 PRI interfaces). 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.
  When you run Voice over ATM (VoATM) on an E1 line, the CPU usage might be high. There is no workaround.
  Hardware compression (SA-COMP/1) on a Cisco 7200 series router or a Cisco 3640 router does not work properly when used with Multilink PPP. The show compression command might show software compression and all fields zero. All data will not be affected, but there is no way to know if compression is happening or not. There is no workaround.

IBM Connectivity

  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.

Interfaces and Bridging

  A Cisco 7500 series router might experience several SONET alarm messages in the console when clock source is not present. This situation might occur when you connect two Packet-over-SONET (POS) line cards with the same clock source line.
  Workaround: Use the clock source internal command on one of the line cards before connecting the two line cards or before entering the no shut command on the router.
  An extended ping from a remote site might cause a Cisco router to reload. There is no workaround.
  A Cisco 7206VXR router that is running Cisco IOS Release 12.1(2.3)T2 might reload with the following messages:
  %OSPF-5-ADJCHG: Process 1, Nbr 212.152.191.106 on FastEthernet1/0 from LOADING to FULL, Loading Done Queued messages: -Traceback= 60EF3E54 600E7A4C 600EB154
  *** System received a Bus Error exception *** signal= 0xa, code= 0x40c, context= 0x61f25ad0 PC = 0x6055b87c, Cause = 0x420, Status Reg = 0x34008002
  There is no workaround.

IP Routing Protocols

  A Cisco router that is using Multiprotocol Label Switching (MPLS) traffic engineering might reload in rsvp_send_static_path. There is no workaround.
  On a Cisco AS5800 series access server that is running Cisco IOS Release 12.1(3a)T1, a memory leak might occur if the ip multicast-routing global configuration command is enabled. This situation might eventually lead to the system reloading. You can detect the memory leak by examining the output of the show processes memory EXEC command, where you will see the "EXEC" process continuously increasing its memory usage. There is no workaround.
  In Cisco IOS software that is running Multiprotocol Label Switching (MPLS) traffic engineering, a tunnel with nonzero bandwidth might be up but have no bandwidth allocated. There is no workaround.
  When the distribute-list router configuration command and the set community route-map configuration command are used with the same peer router, and a Border Gateway Protocol (BGP) route is denied by a distribute-list, then community might not be set on any of the subsequent routes permitted by the distribute-list.
  Workaround: Add a "permit any" access-list to the route-map.
  When a Cisco router that is running Cisco IOS Release 12.1(3)T and Resource Reservation Protocol (RSVP) receives a RSVP Reserve (RESV) message that increases an existing reservation and the corresponding weighted fair queuing (WFQ) or ATM resources are not available, then the router will reload.
  Workaround: Configure endpoints to terminate their reservation, and then to ask for a new, larger reservation if they need to increase the size of their reservation.

LAT

  A Cisco router or access server might reload when you remove a local-area transport (LAT) configuration while a LAT call is up.
  Workaround: Change the configuration after clearing the call.

Miscellaneous

  A Cisco AS5800 access server running Cisco IOS Release 12.1(2a)XH or 12.1(3)T might drop packets for ports on a 324 Universal Port card. This situation occurs during modem retrains and only when there are high output packet rates. There is no workaround.
  Packets traversing the Multiprotocol Label Switching (MPLS) to MPLS and the MPLS to IP feature switching paths might end up being double-counted by the router as indicated by the counters visible in the output of the show interfaces rate-limit EXEC command. There is no workaround.
  A Cisco AS5800 access server running Cisco IOS Release 12.1(2a)XH or 12.1(3)T that has a configuration containing a reference to 324 Universal Port cards might reload when loading a release earlier than Cisco IOS Release12.1(1.1).
  Workaround: Ensure that the feature board in the highest numbered slot of the dial shelf chassis is not a 324 Universal Port card.
  With foreign exchange station ground start (FXS GS) running on a Cisco 5300 series access server, incoming calls are not answered when the access server communicates with a Netmeeting client. There is no impact to ISDN, channel-associated signalling (CAS), or R2 customers. There is no workaround.
  A Cisco AS5300 series access server that is using the AS5300 voice gateway Real Time Streaming Protocol (RTSP) client for dynamic interactive voice response (IVR) prompting might reload.
  Workaround: Do not use RTSP for dynamic prompting.
  Weighted Random Early Detection (WRED) is not supported on a Cisco Route Processor Module (RPM) MGX8850 series wide-area edge switch. There is no workaround.
  If you connect a device to Fast Ethernet on a Cisco 2600 series router that has full-duplex set and auto negotiate capability disabled, you must set the router to match the speed and duplex setting of the other device. There is no workaround.
  The console of a Cisco 2600 and 3600 series router might display memory allocation errors when the network module slot is populated with one of the following network modules (and when no I/O memory size is specified in the configuration):
  NM-1CT1, NM-2CT1: 1 port/2 port channelized T1/ISDN PRI NM-1CT1-CSU, NM-2CT1-CSU: 1 port/2 port channelized T1/ISDN PRI with CSU NM-1CE1U, NM-1CE2U: 1-port/2-port channelized E1/ISDN PRI Unbalanced NM-1CE1B, NM-2CE2B: 1-port/2-port channelized E1/ISDN PRI Balanced
  Workaround: Specify the I/O memory size with the memory-size iomem global configuration command set as follows:
  memory-size iomem 20
  After the I/O memory size has been set, reload the router.
  When you use a Real Time Streaming Protocol (RTSP) server to play prompts on a Cisco IP phone, only the first prompt can be heard. Subsequent prompts result in silence. There is no workaround.
  When a Route Processor Module (RPM) card is stressed with high loads, the RPM might reset itself. There is no workaround.
  A Cisco AS5300 series voice gateway that is running an interactive voice response (IVR) script might reload when both of the following conditions are met:
  Workaround: Use only the call application voice global configuration command when there are no active calls in the gateway.
  An ATM Voice Processing Deck or Multicast Address Resolution Server (MARS) card (NM-1A-ATMOC3-MM-1V) with a single dual T1/E1 voice/WAN interface for MARS card (VWIC-1DSU-T1 or VWIC-1DSU-E1) inside it shows "N/A" in CiscoView for the sonetMediumTable. There is no workaround.
  A Cisco router might experience cable modem reloads if you configure access list number 100 or 101.
  Workaround: Use other access list numbers.
  Q Signalling (QSIG) does not correctly transport call diversion on a no-reply call. There is no workaround.
  On a Cisco 3620 or 3640 router, the hardware compression does not work on the interfaces of NM-1FE2W, NM-2FE2W, NM-1FE1R2W, or NM-2W Network Modules, even when a NM-COMPR Compression Network Module is available.
  On a Cisco 2600 series router, the hardware compression does not work on the interfaces of NM-2W Network Modules, even when an AIM-COMPR2 2 E1 data compression Advanced Integration Module is available.
  On a Cisco 3660 router, the hardware compression does not work on the interfaces of NM-1FE2W, NM-2FE2W, NM-1FE1R2W, or NM-2W Network Modules, even when an AIM-COMPR4 4 E1 data compression Advanced Integration Module is available. There is no workaround.
  Extended Authentication (Xauth) to Cisco IOS software fails after the first rekey. There is no workaround.
  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.
  Alternative workaround: Enter the memory cache-policy io uncached command. However, entering this command might sacrifice packet switching performance.
  A Simple Gateway Control Protocol (SGCP)-controlled FAX call does not work. There is no workaround.
  A Cisco 2600 or 3600 series router that is running Cisco IOS Release 12.0(7)XK1 on a Voice over IP (VoIP) network with g729ar8 configured under the dial-peer codec might select codec g729r8 pre-ietf when you use the show call active voice command. This situation results in the call dropping or proceeding with degraded voice quality.
  Workaround: Add the following configuration to the destination gateway:
  dial-peer voice xxx voip incoming called-number, codec g729abr8
   or
  dial-peer voice xxx voip incoming called-number, voice-class codec x where voice class codec is defined as voice class codec x codec preference 1 g729r8, codec preference 2 g729br8, codec preference 3 g729abr8.
  A Cisco AS5800 series access server running Cisco IOS Release 12.1(2a)XH or 12.1(3)T and the cvIfCfgTable of the CISCO-VOICE-IF-MIB Simple Network Management Protocol (SNMP) MIB does not contain entries for voice ports, even if voice-capable ports do exist. There is no workaround.
  On a Cisco 4500 series router with ATM Route Bridge Encapsulation (ATM RBE) used on a permanent virtual circuit or connection (PVC), the router uses process switching instead of fast switching on incoming 1483 bridge-encapsulated IP packets. The router uses fast switching on IP packets destined for an RBE PVC. There is no workaround.
  On a Cisco AS5800 access server running Cisco IOS Release 12.1(2a)XH or 12.1(3)T with a Network Processing Engine-200 (NPE-200) router shelf and AS58-324UPC-CC feature boards, a memory leak might occur in the Pool Manager process for particular system configurations, which after several days, can deplete the amount of free processor memory to the extent that a system reload occurs. This situation does not affect NPE-300 VXR router shelves, nor does it affect NPE-200 router shelves if AS58-324UPC-CC feature boards are not used. There is no workaround.
  A Cisco 7206 VXR router that is running Cisco IOS Release 12.1(1a)T1 with a Network Processing Engine-300 (NPE-300) might reload several times because of memory allocation failures in the I/O memory pool and memory corruption (bad block). Even if the router is configured for core dumping, it will fail to write out the core dump because of an ALIGN-1-FATAL error. There is no workaround.
  When the cell bus clock is configured for 42 MHz with only one Route Processor Module (RPM) or single cell bus, the ATMizer might pump optical carrier (OC-6) because there is no back pressure.
  Workaround: Configure the cell bus clock to 21 MHz and place only one RPM in each adjacent slot.
  The Queue Engine-0 (QE-0) bandwidth (OC-6) is shared between slots 1-6, 9-14 and Processor Switching Module (PXM) ports. Back pressure from the QE to the Route Processor Modules (RPMs) will result if the RPMs try to send at higher rate than the QE can handle. The back pressure will cause each card to send cell/idle-cells at a rate slower than the OC-3 rate. There is no workaround.
  Open Settlement Protocol (OSP) does not work properly because of Octet3/3a in ISDN SETUP. OSP requires a full E1 64 calling number in the call SETUP message, and that calling number does not get propagated through the voice telephony service provider (VTSP).
  Workaround: Upgrade to Cisco IOS Release 12.1(3)T. In this release, if the incoming setup message does not include the calling number, call control applications programming interface (CCAPI) will include the answer number configured in matching plain old telephone service (POTS) dial-peer routers or will include the called number if the answer number is not configured. The calling Octet3 will be marked calling line ID restriction (CLIR).
  A Cisco voice gateway that uses a custom interactive voice response (IVR) script might experience memory corruption and reload. 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.
  Connection ID does not increment for Open Settlement Protocol (OSP) calls if multiple terminating gateways are available and a terminating gateway is unable to correctly process the call. There is no workaround.
  A Cisco AS5300 series access server might reload with the following error message:
  System returned to ROM by bus error at PC 0x60C1B050, address 0xD0D0D19 System restarted System image file is "flash:c5300-is-mz.121-1a.T1.bin"
  There is no workaround.
  A Cisco AS5300 series access server might experience a software-forced reload in RTCP_CONSTRUCT_RR_BLOCK. There is no workaround.
  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.
  The CCH323_CT process leaks memory in a gateway that has H.245 Terminal Capability Set message formatting. The CPU load of the CCH323_CT is high, indicating that the CCH323_CT is in a loop. There is no workaround.
  An interactive voice response (IVR) script might fail to collect digits when the script connects the call and begins to collect digits without first playing a prompt to the user.
  Workaround: Play a prompt to the user before collecting digits.
  Digit collection on the IP leg of a Media Gateway Control Procotol (MGCP) call will fail. There is no workaround.
  When a Cisco router runs more than six calls simultaneously, voice quality starts to degrade. The output of the show rtpspi statistics Service Provider Interface (SPI) command shows that packets are lost when this threshold is reached. This situation is caused by ATM traffic shaper queueing.
  Workaround: Specify the full T1 average rate on the variable bit rate parameters.
  When you configure Circuit Emulation Service (CES) over switched virtual circuits (SVCs) with the CES network module (NM), the connection might pause indefinitely in the COMING UP state.
  Workaround: Use the shut command followed by the no shut command on the T1 controller that is marked as down.
  When you change a Circuit Emulation Service (CES) connection over E1 switched virtual circuits (SVCs), it might take 20 seconds or more before the CES connection begins to pass CES traffic from the time when the Cisco Connection Manager indicated that the connection was in the UP state. There is no workaround.
  A distributed committed access rate (DCAR) will not match on any Multiprotocol Label Switching (MPLS) packets when you use an MPLS experimental bit rate limit access list configured within the CAR on an output interface. There is no workaround.
  The modular quality of service (QoS) set mpls experimental command does not work in Cisco IOS Release 12.1(4)T. There is no workaround.
  With the rate-limit interface configuration command action set-mpls-exp-transmit 0-7 or set-mpls-exp-continue 0-7 configured, the keyword set-mpls-exp-transmit or set-mpls-exp-continue will be missing from the output if you use the show running interface interface command. The show running command causes the router to reload. There is no workaround.
  All PC-to-phone calls result in spurious memory access and might cause a Cisco AS5300 series access server to reload. This situation occurs only with voice calls and not with modem calls. There is no workaround.
  A Cisco 7500 series router that is running Cisco IOS Release 12.1(2.3)T1 or a later release might display an excessively high number of packet drops when a distributed committed access rate (DCAR) is configured on the input interface. A significantly large percentage of the packets are sent to the Route Switch Processor (RSP) and switched there.
  Workaround: Perform Multiprotocol Label Switching (MPLS) in a nondistributed mode.
  The service-policy output policy-map-name interface configuration command does not function on any Multiprotocol Label Switching (MPLS) packets in the IP-to-MPLS path on a Cisco 7200 series router. There is no workaround.
  Weighted Random Early Detection (WRED) will classify all Multiprotocol Label Switching (MPLS) packets as precedence 0 in the MPLS-to-MPLS and MPLS-to-IP paths, regardless of their actual MPLS experimental field value. There is no workaround.
  A Cisco 7500 series router that is running Cisco IOS Release 12.1(2.3)T2 might experience inconsistencies between the Label Forwarding Information Base (LFIB) table on the Route Switch Processor (RSP) and the line cards. The inconsistency occurs under the following conditions:
  Because of the inconsistency, the line cards are missing all LFIB entries whose outgoing interface is a non-VIP interface configured with the service policy output interface configuration command. There is no workaround.
  On a Cisco router that is running Cisco IOS Release 12.1(2)T, ISDN Data over Voice (DOV) calls fail to connect. There is no workaround.
  Outgoing fax calls do not work on a Cisco AS5800 series access server with an E1 controller. There is no workaround.
  When a Cisco router calls from ISDN (Net5) to a Public Switched Telephone Network (PSTN), and a PROGRESS message is received at the terminating gateway, the PROGRESS message will not be forwarded to the originating point, and inbound ringing will be heard.
  Workaround: Configure a Q Signalling (QSIG) switchtype or upgrade to Cisco IOS Release 12.1(3)XI.
  If you use a translation rule beginning with ".%" and test it with the test translation-rule global configuration command, a Cisco router reloads with the following error message:
  translation-rule 40 Rule 1 .%527.% 7
  cpe_gw#test translation-rule 40 8888527999
  === Flushing messages (13:10:17 MET Fri Jun 30 2000) === Keyword "CRASH {ROUTER} Queued messages *** System received a SegV exception *** signal= 0xb, code= 0x1200, context= 0x819f80f8 PC = 0x80dc4e40, Vector = 0x1200, SP = 0x81e9a240
  There is no workaround.
  ClearChannel Codec cannot be entered in the codec preference list of a voice class. It must be entered in the dial-peer router. There is no workaround.
  A Cisco AS5300 series access server might exhibit the following error on a T1 port:
  Restore frozen DS0
  Rx ABCD mismatch first seen on DS0
  Rx ABCD mismatch first seen on DS0
  Restore frozen DS0
  Restore frozen DS0
  Rx ABCD mismatch first seen on DS0
  There is no workaround.
  Rotary functionality does not work on a Cisco 5300 series access server that is running Cisco IOS image 12.1(02.03)T02. This situation does not occur with earlier releases.
  Workaround: Use dial-peer entries whose session target is not a DNS entry.
  A memory leak occurs each time a Cisco router exhibits the following error message:
  %IDTATM25-3-TXHANG: IDB= ATM0, TX hang, Interface reset
  If the TXHANG error condition persists over several hours, the router will eventually run out of memory and display the following error message:
  %SYS-2-MALLOCFAIL: Memory allocation of 560 bytes failed from 0x223E660, pool Processor, alignment 0
  The router must be reloaded at this point. There is no workaround.
  Compressed Real Time Protocol (CRTP) fails to resynchronize after a packet is lost.
  The output of the show ip rtp header-compression serial x/y dlci zzz EXEC command shows a high error rate. The number of errors increases continuously after you clear the counters.
  The output of the debug ip rtp header-compression EXEC command shows the following "bad sequence context" and "RESYNC" messages:
  RHC Serial2/1: Bad sequence context115: expeced sequence 14, received sequence 0 RHC(fast) Serial2/1: RESYNC:context115: expectedsequence 2, received sequence 1 - reordering ?
  The output of the sh ip rtp header-compression serial x/y dlci zzz detail EXEC command provides the following information:
  Connection 254: IP source: 10.1.0.1, IP destination: 10.5.3.2 UDP source: 16452, UDP destination: 16940, RTP ssrc: 212823558 Last packet sent is compressed-rtp and last sequence sent is 14 Resynchronization feedback has not been received
  There is no workaround.
  Voice over IP (VoIP) Disconnect cause is incorrect for T1 channel-associated signalling (CAS) signalling types after Cisco IOS Release 12.1(2.1)T. There is no workaround.
  When a gateway receives an incorrectly encoded H.245 protocol data unit (PDU) with an extra byte of data, the gateway keeps attempting to decode the extra byte, causing an unconditional printing of an error message to the console.
  Workaround: Reload the gateway.
  The Re-direct Info, Display Info, Endpoint Info, Inclusion of calling number, Octet3/Octet3a for calling and called number, and Calling Sub Address information elements are not passed properly between ISDN connections and H.323 Voice over IP (VoIP) connections on a Cisco AS5300 series access server that is running Cisco IOS Release 12.1(2.3)T1 or a later release. There is no workaround.
  Unconfiguring PRI from the controller does not remove the voice port configuration. There is no workaround.
  An interactive voice response (IVR) script with ivr-out in authorization/authentication requests might cause memory corruption.
  Workaround: Write a script that does not use this functionality.
  After you upgrade to Cisco IOS Release 12.1(2.3)T1, the connection with the RADIUS server fails after a reload.
  Workaround: Change the RADIUS server in the configuration. You will need to reapply this workaround if you reboot the router.
  Incoming Q Signalling (QSIG) calls sent to a voice gateway reload the gateway. When facility_IE is present in an incoming message, the gateway invokes rose functions to decode facility_IE. When a decode error occurs, the gateway calls free routines to free bad memory and the gateway reloads. There is no workaround.
  When calls in a Signalling System 7 (SS7) interconnect with an H.323 solution, the voice path is cut through in only one direction when a Continuity Test (COT) request is made by the SC2200 Signalling Controller on the egress side of the solution (egress being AS5300-to-NI2+-to-SC2200-to-SS7). This situation does not affect COT requests coming in from the remote switch to the solution.
  Workaround: Do not set a COT percentage on any SC2200 that will be servicing 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.
  Fax calls might fail when Frame Relay IP Real-Time Transport Protocol (RTP) priority is configured on Voice over IP over Frame Relay (VoIPoFR). There is no workaround.
  On a Cisco AS5400 access server, a number of the port statistics in the CISCO-MODEM-MGMT-MIB might return misleading results because of the way that Service Processing Element (SPE) statistics are distributed across the ports of an SPE. There is no workaround.
  On a Cisco AS5800 series access server that is running Cisco IOS Release 12.1(3a)T1, Simple Network Management Protocol (SNMP) queries for variables in the CISCO-MODEM-MGMT-MIB will fail for ports numbered 72 and above on Double-Density Modem (DMM) feature boards. SNMP queries for variables in this Management Information Base (MIB) will also fail for all ports on all types of feature boards that are in a slot with a higher number than the first DMM card. There is no workaround.

Novell IPX, XNS, and Apollo Domain

  Internetwork Packet Exchange Control Protocol (IPXCP) for virtual-template interfaces does not come up in "listen" mode if you use the no ipx ppp-client interface configuration command to dissociate the interface from the master. There is no workaround.

TCP/IP Host-Mode Services

  On a Cisco AS5800 series access server that is running Cisco IOS 12.1(3a)T1, a memory leak might occur if the ip tcp header-compression interface configuration command or the ip tcp header-compression passive interface configuration command is enabled on an interface. This situation might eventually lead to the system reloading. You can detect the memory leak by examining the output of the show processes memory EXEC command, where you will see the "PPP Auth" process continuously increasing its memory usage.
  Workaround: Disable IP TCP header compression.

Wide-Area Networking

  The isdnBearerCallConnectTime variable is not properly updated for incoming calls through the BRI interface. However, it is updated properly for outgoing calls. This situation affects Simple Network Management Protocol (SNMP) users and does not compromise router functionality. There is no workaround.
  Pings fail for Multilink PPP with certain vendor Point-to-Point Compression MPPC. Even though compression appears to be functioning correctly, the link is torn down before resynchronization can occur. Cisco does not recommend using PPP payload compression (MMPC only) with a virtual private dial-up network (VPDN). There is no workaround.
  A Cisco 1-port ATM-enhanced OC3/STM1 multimode port adapter (PA-A3-OC3MM ATM) used with a Cisco series 7200 or 7500 series router might pause indefinitely with the following message:
  %ATMPA-3-SARCRASH: ATMx/x: SAR1 Chip Crashdump:
  There is no workaround.
  If the dialer pool-member interface configuration command is removed while active calls are within that dialer profile pool, the router may pause indefinitely. To recover in this state, power cycle the router to reload.
  Workaround: Ensure that the dial-on-demand routing (DDR) interface is shut down before making any changes.
  A Cisco 2600 or 3600 series router that is using a Frame Relay (FR) ISDN switch cannot handle packets terminating at the switch.There is no workaround.
  A Cisco 2600 series router reloads while pinging a remote router.
  Workaround: Do not use a test script to configure a Frame Relay (FR) switched virtual circuit (SVC).
  When you add or delete PPP configurations in a test script on a Cisco 4500 series router, the following kernel level stack trace might occur:
  Initial SP = 0x60f7bc90, Initial PC = 0x603c7544, RA = 0x603b3e9c
  Workaround: Perform a shut on the interface before deleting the PPP encapsulation.
  Failure to use the correct syntax for the vpdn:tunnel-Type attribute in a RADIUS profile might cause the requesting router to reload. There is no workaround.
  Where you try to use a Cisco Route Switch Processor (RSP) as a switch between an ISDN cloud and a Frame Relay (FR) cloud, the RSP does not send anything back when it receives the ping packets from the other side. Instead the log shows the following message:
  1d02h: %CBUS-3-BADVC: MIP 1/0 got bad VC packet 30810800 45000064 01290000 FF01906B 14010101 14010102 000093D3 02781E6C dt219#
  The same test with FR clouds on both ends of the switch does not experience the problem. There is no workaround.
  If you remove at least two T1 lines or clear all the active ports in a T1 channel-associated signalling (T1 CAS) environment that is configured with virtual templates and that has active calls, a Cisco router will reload with the following traceback message:
  0x6033788C:process_handle_watchdog(0x60337858)+0x34 0x6033B340:signal_receive(0x6033b290)+0xb0 0x60331ACC:watchdog_forced_here(0x60331a08)+0xc4 0x604A9284:ppp_manager(0x604a896c)+0x918 0x6030A4BC:r4k_process_dispatch(0x6030a4a8)+0x14 0x6030A4A8:r4k_process_dispatch(0x6030a4a8)+0x0
  Workaround: Use the shut command on the interface prior to pulling the card while traffic is active.
  A Cisco 800 series router might reload when a VPN connection is attempted. There is no workaround.
  On a Non-Facility Associated Signalling (NFAS) configuration, the ISDN subsystem might hold a large amount of processor memory that causes the system to run out of memory.
  Workaround: Reload the system.
  Some vendor PBX systems connected to a Cisco 3600 series router through ISDN might not work properly. Calls from the PBX to the router work correctly. Calls from the 3600 to the PBX fail. There is no workaround.
  After you upgrade to Cisco IOS Release 12.1(1a)T1, a PPP link with compression enabled no longer works between a Cisco 3600 series router and a 760 router.
  Workaround: Disable the compression.
  Non-Facility Associated Signalling (NFAS) and Signalling System 7 (SS7) for voice calls using the SC2200 signalling controller do not work on a Cisco AS5300 series access server that is running Cisco IOS Release 12.1(2a)XH or Release 12.1T. There is no workaround.
  A Cisco 4000 series router that acts as a home gateway in a virtual private dial-up network (VPDN) application is not stable and might reload after several hours of operation. There is no workaround.
  Frame Relay fragmentation (FRF.12) cannot be configured on a Cisco 2500 series router. There is no workaround.
  A Cisco AS5300 series access server experiences several PPP problems during slow dial-in connections, such as the V.110 global system for mobile (GSM) calls.
  With Cisco IOS Release 12.0(7)T, the AS5300 access server closes the PPP connection with the "Missed 5 keepalives" message. Traces taken at the line prove that the link control protocol (LCP) keepalives are sent by the PPP client.
  With Cisco IOS Release 12.1(2a)T1, the AS5300 access server repeatedly sends IP Control Protocol (IPCP) CONFREQ messages, and the PPP clients fail to connect. There is no workaround.
  In any release of Cisco IOS software, when the High-Speed Serial Interface (HSSI) is used with Multilink PPP and the interface maximum transmission unit (MTU) is not set to 1500, the links will fail to negotiate.
  Workaround: Set the MTU to 1500.
  A Cisco 7500 series router that is running Cisco IOS Release 12.1(2)T might reload by watchdog timeout. Before reloading, the router displays the following messages:
  %SYS-3-CPUHOG: Task ran for 11584 msec (84/31), process = Exec, PC = 6030B148. -Traceback= 6030B150 60279E48 6027A72C 6027A13C 60094C80 602D4D80 602D1CA8 602D4A60 6029401C 6028C218 6029A880 6028C334 6029401C 602A031C 602F39F4 602F39E0CMD: 'en' 22:02:32 MET-DST
  %SYS-2-WATCHDOG: Process aborted on watchdog timeout, process = IP SNMP. -Traceback= 6031FFBC 60323A70 60319AA4 603F18A0
  There is no workaround.

Resolved Caveats—Release 12.1(3)T

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

Access Server

  The default bundled portware for a Cisco AS5300 or an AS5800 series access server needs to be 2.7.2.0. The current default firmware is 2.7.1.1, which does not support inbound messaging.
  Workaround: Manually download 2.7.2.0 from CCO and configure the access server to download this firmware to the modems using the spe global configuration command.
  If you configure Service Processing Element (SPE) firmware on a Modem ISDN channel aggregation (MICA) for a Cisco 5800 series router shelf, an SPE with calls will result in the reloading of the router shelf when the calls drop. This situation occurs only when new firmware is configured. There is no workaround.

Basic System Services

  A Cisco router that is running a Source-Active (SA) Agent jitter operation might experience a memory leak. There is no workaround.
  When you start a Service Assurance (SA) agent operation of any type on a Cisco 1600 series router, a CPU exception error message occurs. There is no workaround.

Interfaces and Bridging

  802.1Q bridging does not work on a Cisco Route Switch Processor (RSP) that is running Cisco IOS Release 12.1(2.2)PI or a later release. There is no workaround.

IP Routing Protocols

  A Cisco router that is using Multiprotocol Label Switching (MPLS) traffic engineering might reload in or near rsvp_rrr_path_query. Sometimes this reload is accompanied by a SYS-6-STACKLOW message. There is no workaround.
  A Cisco router that is running Cisco IOS Release 12.1 T and that is using Resource Reservation Protocol (RSVP) might reload when handling static path reservations. This situation occurs when the router runs Routing with Resource Reservation (RRR) tunnels or Common Open Policy Service (COPS) applications. There is no workaround.
  When a Cisco router that is running Cisco IOS Release 12.1(2)T and Resource Reservation Protocol (RSVP) receives a reservation refresh from a downstream router or application with a modified flow specification, the router fails to issue an updated RSVP Reservation (RESV) message to the upstream router and will stop sending RESV refresh for the flow. As a result, the reservation will be torn down. The implication for the flow is that it does not have the reservation installed on all the routers along the path and is not guaranteed quality of service (QoS). There is no workaround.
  A Cisco router might reload if the generic routing encapsulation (GRE) Network Address Translation (NAT) entries were deleted prior to deleting the Point to Point Tunneling Protocol (PPTP) control session NAT entry, when there was an address change on the interface on which it is overloaded. There is no workaround.

Miscellaneous

  Maximum-length Message Signaling Units (MSUs) are discarded when you use the Cisco Signalling System 7 (SS7) Signalling Link Terminal (SLT). There is no workaround.
  A voice call on a Cisco 1750 router that is running Cisco IOS Release 12.1(1.1)T might fail without generating a ring tone. On a Cisco 1750 router that is running Cisco IOS Release 12.1(1) and that has the same configuration, the call succeeds. There is no workaround.
  If you perform Xauth and mode configuration on a Cisco 2500 series router that is running Cisco IOS Release 12.1(1)T, IP Security (IPSec) negotiation occurs but no IP is assigned. The router exhibits the following message:
  %SYS-3-HARIKARI: Process Crypto IKMP top-level routine exited
  Crypto will not work until the router is reloaded. There is no workaround.
  A Cisco 1750 series router might experience echo problems while running Cisco IOS Release 12.1(1a)T1. There is no workaround.
  On a Cisco 5800 series or 5400 series router using DFC108s with firmware 6.76, a "0x7002 Session Abort" message might occur because of extra data being sent to the CSMV6 after it has entered the terminate state.
  Workaround: This "Session Abort" message causes the Service Processing Element (SPE) to be marked for "Recovery," and the SPE is recovered during the maintenance time. This process returns the SPE to a working state.
  In Cisco IOS Release 12.1(1.x)T, if two (or more) dial-peers are configured with destination patterns terminated by T, then calls with a dial number completed by # key are routed to the first matching dial-peer. The router does not check any other dial-peers, which can have longer match or better preference configured. There is no workaround.
  The PVC with voice encapsulation will not work on 7500 platform. There is no workaround.
  On a Cisco AS5800 access server running Cisco IOS Release or 12.1(3)T, outgoing foreign exchange station (FXS) calls might not connect reliably. Calls are especially unlikely to connect if a large number of digits are being dialed or detection of dial tone is slow. There is no workaround.
  On a Cisco 7200 series router, traffic is not classified correctly in the mpls->ip path for traffic from provider-edge routers (PE) to customer-edge routers (CE) for "untagged destinations." This situation results in packets getting incorrect weight and therefore, incorrect service. There is no workaround.
  The H.323 protocol stack on Cisco IOS software uses a third-party Abstract Syntax Notation One (ASN.1) library, and Cisco does not have control over the source code. Under some situations, when there are encoding errors and no free memory is available, a Cisco router reloads inside the third-party library code. In the third-party library code, the _oss_freeGlobals() function is called by the ASN encoder, to free all allocated memory whenever an error is detected. However, the encoder does not free the memory gracefully and causes the router to reload.
  Workaround: Upgrade to Cisco IOS Release 12.1 T, which uses a Cisco-developed ASN1 library code.
  Cot Loopback fails in a Cisco router that is running Cisco IOS Release 12.1 PI or 12.1 T. There is no workaround.
  On a Cisco AS5400 or AS5800 access server running Service Processing Element (SPE) image 6.82 or earlier, a "Session Abort 0x70030BAD" message might occur on a heavily stressed DFC108 module. The condition is intermittent. There is no workaround.
  A Cisco AS5300 series access server that uses new controller cards (either having 4 ports or 8 ports T1/E1) with four serial interfaces might randomly reload. There is no workaround.
  When you reload a Cisco router that it is running Cisco IOS Release 12.1(2)T, any configured secondary IP addresses on Ethernet and Bridge Group Virtual Interface (BVI) interfaces are removed from the configuration.
  During bootup, the router displays the following error message:
  Secondary addresses not allowed with negotiated addresses
  Workaround: Reconfigure the secondary IP.
  Using Extended Authentication (Xauth) in Cisco IOS software will result in the following message:
  %SYS-3-HARIKARI:Process Crypto IKMP top-level routine exited
  There is no workaround.
  There is a missing incoming_guid copy in the Cisco IOS Release 12.1 PI that gets synchronized into Cisco IOS Release 12.1(2.3)T.
  Workaround: Insert the copy line.
  When Channel-Associated Signalling (CAS) preauthentication is enabled after Dial-In over ISDN/automatic number identification (DNIS/ANI) collection for E1 R2 Signalling, Cisco IOS software will not send an answer signal until it gets a response from the RADIUS server. If the call is accepted, Cisco IOS software will send a register answer signal and a connect-through-the-line answer. If the call is rejected, Cisco IOS software will send a busy signal and send idle to disconnect the call. There is no workaround.
  The bootflash file system might be corrupted after a system reload. This situation occurs only after a crash file is written to bootflash. There is no workaround.
  Voice traffic is dropped when you use compression and a call cannot be resumed. If an error appears at the decompressor when you use Voice over IP over Frame Relay (VoIPoFR) and fast switching or Cisco Express Forwarding (CEF) switching, all subsequent packets in the flow are dropped and voice connection is never regained.
  Workaround: Use process switching or do not use compression.
  On a Cisco 5300 series access server that is running Cisco IOS Release12.1(2)T and Release 12.1(3)T, the debug cas command does not work correctly. There is no workaround.
  The Open Settlement Protocol (OSP) feature does not work because of the missing roaming capability on the new tool command line (TCL) interface infrastructure. There is no workaround.
  The gatekeeper expects to receive the destination IP address tagged with the OUTBOUND_GTW_IP string. An OUTBOUND_GW_IP tag causes the destination IP address fail to be processed by the gatekeeper. There is no workaround.
  Frames with Ethernet frame check sequence (FCS) that are queued for process-level handling are dropped incorrectly with a format error. 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.
  FAX Non Standard Facility (NSF) cannot be configured for Voice over Frame Relay (VoFR) or Voice over ATM (VoATM) on a dial-peer router that is running Cisco IOS Release 12.1(2)T. There is no workaround.
  Tool command line (TCL) interface script authorization fails while sending Mind Server with correct account or password information. This situation affects all debit card TCL scripts. There is no workaround.
  Calls from PBX2 to PBX 1 are connected, but no ring tone is heard. Debugging shows that a "T" digit is matched although there is no "T" digit configured in the dial-peer destination pattern. This situation causes the router to miss the ring tone from the called party.
  Workaround: Turn off H.323 fast-start functionality on all VoIP dial-peer routers by performing a req-qos controlled-load.
  On a Cisco AS5800 series access server that is running Cisco IOS Release 12.1(2a)XH or Cisco IOS Release 12.1(3)T, the show modem operational-status and the show modem configuration EXEC commands display nothing for Modem ISDN channel aggregation (MICA) Hex Modem Module/Double-Density Modem (HMM/DMM) modem ports.
  Workaround: Use the show modem [slot/port] or the show modem log [slot/port] EXEC commands to view modem history statistics.
  If you use the ip address-pool [dhcp-proxy-client] command to allocate a Cisco IOS software internal address using Dynamic Host Configuration Protocol (DHCP), the interface on which the DHCP DISCOVER packet was sent might acquire the DHCP allocated address. 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 dsl 0 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 alternative workaround (below). If you reload the router, E1 will be inaccessible again.
  Alternative workaround: Use the shut command followed by the no shut command on the controller. Though doing so clears the problem, the initial situation recurs after reloading.
  When Multilink PPP and fast switching or Cisco Express Forwarding (CEF) switching are configured, Real Time Protocol (RTP) packets are not compressed.
  When you use Multilink PPP for link fragmentation and interleaving (LFI) in a voice and data environment, the ip rtp header-compression command has no effect on most platforms and can cause a Cisco 7200 series router to reload.
  Workaround: Turn off fast switching or CEF switching by using the no ip cef and the no ip route-cache global configuration commands.
  A Cisco 7200 series router will reload when ClearChannel Codec is configured on the dial-peer router because the memory allocated for Terminal Capability Set elements is one less than the total number of available codecs. The last codec entry is ClearChannel Codec, so the router reloads.
  Workaround: Do not use ClearChannel Codec.
  On a Cisco router that is running Cisco IOS Release 12.1(2)T with the Airline Product Set (ALPS) set and encapsulated Airline Control (ALC), ALPS circuit processing queues always queue data regardless of the state of the peer or the circuit. Reliance on failure processing to discard the packets is a solicited error because the router sends service message after the loss of a packet. The circuit failure notification takes 15 to 30 seconds. If the packet lifetime timer has been exceeded, the router generates a service message. The time to generate a service message is random because it depends on when the message is queued. Because the timers are hardcoded, the CRT is locked until a 30 second timer expires at the agent-set control unit (ASCU) or the service messages is received.
  Workaround: Upgrade to a Cisco IOS release that has the fix for this bug.
  Cisco gateways that are registered to some vendor gatekeepers and that send an Admissions Confirm (ACF) message with ClearTokens consisting of unknown object identifiers in the nonstandard field of the ClearToken structure might experience a memory leak in the CCH323_CT process.
  Workaround: Set the gatekeeper not to send ClearTokens and disable the gateway security.
  The tool command line (TCL) interface 1.0 interactive voice response (IVR) infrastructure does not properly handle ISDN PROGRESS indication messages. This situation can result in incomplete calls, inaudible ringback, and busy tones when you use certain ISDN switchtypes.
  TCL session application ignores the CC_EV_CALL_PROGRESS event from the call control applications programming interface (CCAPI), so this event is not passed to the Voice over IP (VoIP) Service Provider Interfaces (SPIs). This condition breaks the alerting function for calls terminating on channel-associated signalling (CAS) trunks. There is no workaround.
  When Cisco voice gateways operate with alternate endpoints as directed by the gatekeeper, the free event queue buffers might get depleted.
  Workaround: Disable the alternate endpoints.
  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.
  After a user runs Simple Gateway Control Protocol (SGCP) calls for several minutes, a Cisco router experiences 100 percent CPU usage and cannot process any calls. The router will not reload automatically (which only takes about 1 minute). The CPU high utilization is at VOIP_RTCP process. There is no workaround.
  When voice and data are both running in Frame Relay Low Latency Queuing (FR LLQ) configuration, some of the data packets are being classified as voice. This situation will result in police and consequent packet drops in the priority queue, which will cause bad voice quality.
  Workaround: Turn on process-switching on the incoming interfaces for voice and data.

Wide-Area Networking

  Under rare circumstances, a Cisco router might experience a software-forced reload. The router displays the following queued messages:
  %SYS-3-LOGGER_FLUSHING: System pausing to ensure console debugging output. validblock_diagnose, code = 11 current memory block, bp = 0x61BA5080, memory pool type is Processor data check, ptr = 0x61BA50A8 next memory block, bp = 0x61E02608, memory pool type is Processor data check, ptr = 0x61E02630 previous memory block, bp = 0x61E02660, memory pool type is Processor data check, ptr = 0x61E02688
  %SYS-3-BADMAGIC: Corrupt block at 61BA5080 (magic 61E0058C)
  -Traceback= 6036EBD0 603705D0 602FAEDC 602E09A8 60302C54 6030EEA4 60361794 60361780
  %SYS-6-MTRACE: mallocfree: addr, pc 61BA50A8,602FAED4 61BA4DCC,602F5274 61BA4DCC,40000020 61CE6A20,602F5164 61CE6A20,40000016 61CE5E64,60369FDC 61BA5484,60369FDC 61CD53CC,60369FDC
  The following combination of actions causes this situation to occur:
  The service password-encryption global configuration command is already enabled or is enabled before the third action.
  The username password global configuration command is entered with unencrypted passwords for more than one user.
  A command causes the running configuration to be displayed. (Examples: the write terminal command, the copy running-config startup-config command, or answering "yes" at the "System configuration has been modified. or Save?" prompt after using the reload EXEC command.)
  During the conversion from unencrypted to encrypted passwords, all passwords except the final encrypted password become corrupted. For example, if username one password one, username two password two, username three password three, is entered, the encrypted passwords represented by "one" and "two" will become corrupted. The passwords will become one of two things:
  Random characters that are invalid for a password.
  The same as the final password. In the example above, all three passwords would be the same as the password for "three".
  These passwords are not corrupted until after the running configuration has been displayed or saved, so the first commands, such as write terminal and copy running-config startup-config, will show a correct configuration, but the corresponding subsequent commands will show the corrupted passwords.
  If the password is corrupted, that username will not be able to log in. If the password is changed to the last entry, the earlier usernames might be able to log in using the password of the last user. If users are deleted, the router might reload. If the passwords are already encrypted, this problem does not occur so there will be no problem with existing configurations.
  Workaround: Configure and display passwords one at a time. To add the passwords in the example above, the following command sequence could be used: conf t, service password-encryption, username one password one, end, show running, conf t, username two password two, end, show running, conf t, username three password three, end, show running, conf t.
  Alternate Workaround: Reload immediately after configuration. For example, the following command sequence will work: conf t, service password-encryption, username one password one, username two password two, username three password three, end, copy running-config startup-config, reload.
  Overlap receiving for BRI might not work for NET3. Overlap is currently supported for NET3 and Q Signalling (QSIG) switchtypes only. There is no workaround.
  PPP over Ethernet (PPPoE) sessions on a Cisco 6400 series node route processor (NRP) disconnect at random because of dropped keepalive packets. There is no workaround.
  If Calling Party number information element (IE) is received without any digits, the calling_oct3a field is set to 0. Instead, it should fill in the Octet 3a of this IE, so that upper layer applications get it. There is no workaround.
  On a Cisco router that is running Cisco IOS Release 12.1(0.9)T or an earlier release, a ping cannot go through a BRI interface until all of the following conditions are met:
  There is no workaround.
  ISDN does not send the Called Party "Number Type" information received from the call switching module (CSM) in the outgoing SETUP message. There is no workaround.
  On a Cisco AS5800 access server that is running Cisco IOS Release 12.1(2a)XH or 12.1(3)T, if the ppp callback {accept} command is enabled on an asynchronous or group-asynchronous interface and if a call is made from a client into the access server, then the call will fail (regardless of whether callback is enabled or disabled for that particular user). There is no workaround.
  When a network access server (NAS) is reloaded, it sends the wrong Non-Facility Associated Signalling (NFAS) interface number to SC2200 in a Group Service Message that causes Circuit Identification Codes (CICs) to be in the LOCMAN and REMMAN state on SC2200. There is no workaround.
  Non-Facility Associated Signalling (NFAS) voice calls and Signalling System 7 (SS7) voice calls using an SC2200 signalling controller fail. There is no workaround.
  When a voice application on an ISDN Basic Rate Interface (BRI) port that is running Cisco IOS Release 12.1(2)T has both B-channels of the BRI port in use, and a third call is tried from the Voice over IP (VoIP) call leg to this BRI port, the new call fails as it should. However, the disconnect procedure sometimes will cause one of the established calls where the VoIP call leg cannot hear from the BRI side.
  Workaround: Limit the number of calls on the dial-peer to two.
  When a Non-Facility Associated Signalling (NFAS) primary D interface is shut down, all of the calls are dropped. There is no workaround.
  A system that is configured with Signalling System 7 (SS7) and ISDN interconnect application with Voice over IP (VoIP) could reload after running for several hours. There is no workaround.
  A traceback message might be produced because of dereferencing an invalid idb pointer. There is no workaround.
  All virtual private dialup network (VPDN) functionality is defective. There is no workaround.

Resolved Caveats—Release 12.1(2)T

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

Basic System Services

  A Cisco MC3810 series multiservice access concentrator does not include the option to configure a G.729 "pre-ietf" codec when running Voice over IP (VoIP). This situation will affect the operability of MC3810 with other products running VoIP when those products are using Cisco IOS releases prior to Cisco IOS Release 12.0(4)T.
  Workaround: Upgrade to Release 12.0(5)T (or a later release) for those products operating with MC3810 and running VoIP.
  The ds0-group for a Cisco MC3810 multiservice access concentrator needs to be reconfigured every time it is rebooted. There is no workaround.

EXEC and Configuration Parser

  Rotary groups under virtual vtys might not work. Attempts to make connections to that rotary group will show a message indicating that the connection is being queued even though the vty is free. There is no workaround.

Interfaces and Bridging

  A Cisco Packet OC-3 Interface Processor (POSIP) might reload with a bus error. There is no workaround.

IP Routing Protocols

  A Cisco router that is using Multiprotocol Label Switching (MPLS) traffic engineering might reload in or near rsvp_rrr_path_query. Sometimes this reload is accompanied by a SYS-6-STACKLOW message. There is no workaround.
  Due to new RFC 2328, the calculation of summary route costs have changed. This situation might create suboptimal routing, if all of the are border routers (ABRs) are not upgraded to the new code at the same time.
  Workaround: Upgrade all of the ABRs to the new code.
  A Cisco router might reload if the generic routing encapsulation (GRE) nat-entries were deleted prior to deleting the Point to Point Tunneling Protocol (PPTP) control session nat-entry, when there was an address change on the interface on which it is overloaded. 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.
  Multicast packets are sent out with an "all routers" destination multicast MAC address of 01005e000002 that corresponds to the "all routers" multicast IP address of 224.0.0.2. However, the destination multicast IP address in the packet is the correct one.
  This method works on Windows 95 because Windows 95 accepts the packets with the "all routers" multicast address. With NT, the PC accepts the packets with the "all routers" multicast MAC address, but drops them when the multicast MAC address does not belong to the multicast IP address carried in the MAC frame.
  This situation occurs in the node route processor/Layer 2 Tunnel Protocol Access Concentrator (NRP/LAC) and not in the CPE. You can verify this circumstance by turning on the show ip mcache EXEC command at the NRP. There is no workaround.
  A Cisco 6400 series node route processor (NRP) that is running PPP over Ethernet (PPPoE) and Integrated Routing and Bridging (IRB) on the same NRP might occasionally experience some spurious memory accesses. There is no workaround.
  The alarm-forward feature on a Cisco 2611 router with a VWIC-2MFT-E1 Voice/WAN interface card does not work for E1 controllers on a Cisco router. This situation only affects the Cisco Signalling System 7 (SS7) Signalling Link Terminal (SLT) product. 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.
  A Cisco router that is configured with Dynamic Host Configuration Protocol (DHCP) might reload.
  Workaround: Remove DHCP from the router configuration.
  When a gatekeeper is operated with multiple Location Requests (LRQs) or multiple gatekeepers per zone, there is a possibility that multiple Location Confirmations (LCFs) from multiple LRQ requests to multiple destination gatekeepers might occur. Under this condition, the gatekeeper will create a remote endpoint based on the first LCF received. The subsequent calls might fail if the LCF received is not from the zones that are used to create the remote endpoint as done from the very first success of LCF. These subsequent calls will be erroneously rejected. This situation can be diagnosed using gatekeeper debug to identify the problem (debug gatekeeper main 5). With debug enabled, the error message that indicates failure to create remote endpoint will appear. There is no workaround.
  With Cisco IOS Release 12.0(7)XR, you should not try to explicitly destroy a static quality of service (QoS) profile created by a cable modem termination system (CMTS) and used for G.729 or G.711 upstream CBR service. These profiles occupy QoS profile index 3 and 4 in Cisco IOS Release 12.0(7)XR. The CMTS will not allow you to destroy the profiles. However, a failure in the qos profile delete command line interface (CLI) will cause the ubr7200 series router to reload when the first voice or fax call using the undeleted profile terminates.
  Workaround: Do not try to destroy CMTS-created static QoS profiles used for CBR voice or fax at index 3 and 4.
  A Cisco router will not check to learn if an interface data block (idb) needs to be initialized. There is no workaround.
  A Cisco gateway does not comply with the H.323 specification of the International Telecommunication Union (ITU) because the ITU recently changed the specification dictating that newly created calls for an H.450.2 Call Transfer or H.450.3 Call Deflection generate a unique callIdentifier. The previous specification dictated that new calls should reuse the existing callIdentifier from a previous call. This situation causes interoperability problems with third party Gatekeepers that expect a unique callIdentifier for forwarded or transferred calls, and the duplicate callIdentifier causes the gatekeeper to reject the new call. There is no workaround.
  The H.323 Implementor's guide published by ITU recently introduced language that dictates when creating a new call for an H.450.2 Call Transfer or H.450.3 Call Deflection, the new call SHALL generate a unique callIdentifier; previously the Implementor's guide specified that the new call should re-use the existing callIdentifier from a previous call.
  Since our gateway follows the old specification, we are now in violation of the H.323 Specification. Additionally, we are seeing interoperability problems with third-party Gatekeepers that expect a unique callIdentifier for forwarded/transferred calls; the duplicate callIdentifier causes the Gatekeeper to reject the new call.
  Also, the behavior will cause backward-compatibility problems between 12.1(1)T and future IOS releases of the Cisco Gateway. Since the post-12.1(1)T gateway must follow the new specification, it will be incompatible with 12.1(1)T.
  There is no workaround for this problem.
  A Session Initiation Protocol (SIP) subsystem will reload on Domain Name System (DNS) query when the DNS entries configured on the DNS server have a Time To Live (TTL) value set to zero.
  Workaround: Make sure that no DNS entry in the DNS server has TTL set to zero. This workaround does not impose any limitation.
  For calls where a nonzero Progress Indicator Octet 4 value other than 0x08 is received, you will hear a ringback tone for the original call. However, if the call is deflected or transferred, you will not hear a ringback tone after the call is deflected or transferred. There is no workaround.
  On a Cisco 2600 series router, a Cisco 3600 series router, or a Cisco Signalling System 7 Link Terminal (SS7 SLT), the firmware for the T1/E1 Multiflex Voice/WAN interface cards (VWIC-MFT-T1/E1) prematurely includes V.54 bit error rate testing features, which are not fully operational. There is no workaround.
  When H.323 security is enabled on a gateway, the ClearToken parameter is included in the H.323 Registration, Admission, and Status (RAS) messages sent by the gateway to the gatekeeper. While the RAS messages are sent, memory is allocated for ClearToken structure and its contents to build the ClearToken. After receiving an appropriate response from the gatekeeper, the ClearToken memory is not freed for registration request (RRQ) messages, which results in a memory leak. There is no workaround.
  Simultaneous traffic to the same cable modem (CM) with different IP Precedence rate limits configured using the Cisco vendor-specific IP Precedence rate limit results in erroneous rate limits being applied for nonprecedence data packets, which leads to packet loss and a fall in TCP throughput.
  Workaround: Disable the IP Precedence rate limit setting to prevent TCP throughput from suffering by removing the IP precedence rate limit setting in the CM configuration file. This will affect voice quality when voice and data traffic are sent simultaneously to a CM. If voice quality is more important to you than TCP throughput, then no change is required.
  The Secure Socket Layer Reference (SSLRef) Library is not properly integrated with the public key infrastructure (PKI) code base. Secure Socket Layer (SSL) verifies all certificates correctly, so it is important that the PKI configuration is correct. Prior to Cisco IOS Release 12.1(2)T, it was possible for SSL to appear to be verifying certificates even when the crypto ca identity global configuration command section of the configuration was incomplete.
  For example, you must either specify how to obtain a Certificate Revocation List (CRL) from the certification authority (CA) server, or specify crl optional (which is not a secure solution). It is necessary to make this specification before PKI can verify a certificate as being valid. Before Release 12.1(2)T, this specification would not have been necessary because the certificates were not being verified anyway. There is no workaround.
  In Signalling System 7 (SS7) applications, a Cisco router might not clear calls properly when the T1 or E1 physical layer fails. The show isdn {active} EXEC command shows the calls are still active. There is no workaround.
  The H.323 gateway memory leaks slowly when running with quality of service (QoS) in H.245. This situation occurs when a gateway is configured with Resource Reservation Protocol (RSVP) and QoS in the Voice over IP (VoIP) dial-peer for a value other than the best-effort default. The memory leak is approximately 64 to 100 bytes for each call. The router will eventually run out of memory.
  Workaround: Do not use RSVP and specific QoS other than the best-effort default. Once the leak occurs, there is no workaround except to reboot.
  In bundling, if the IP address on the master interface is configured with certain values, modems on all cable interfaces within the bundle will not be able to pass init(i).
  Workaround: Do not use an IP address pool that has a hex value of "e" in the upper nibble of the third byte (bits 12-15) in any of the IP addresses in the pool.
  A voice call on a Cisco 1750 series router that is running Cisco IOS Release 12.1(1.1)T might fail without generating a ring tone. On a Cisco 1750 router that is running Cisco IOS Release 12.1(1) and that has the same configuration, the call succeeds. There is no workaround.
  If a frequency hop is triggered on an upstream port serving CBR slots, a Cisco ubr7200 series router might reload.
  Workaround: Do not configure the frequency hopping feature on a U.S. port that will be serving CBR slots.
  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 2613 series router that is running Cisco IOS Release 12.1 T or Release 12.0(7)XK1 might reload while booting up.
  Workaround: Configure a memory-size iomem global configuration command in the startup-config file to read memory-size iomem 30.
  Alternative workaround: Upgrade to Cisco IOS Release 12.1.
  A Cisco router that is running Cisco IOS Release 12.1(1)T might experience the equivalent of a memory leak when session directory cache (SDR) duplicates entries.
  Workaround: Enable timeouts for SAP cache, using the ip sap cache-timeout command (for example, ip sap cache-timeout 15). In some instances, using this configuration improves memory consumption rates.
  If a controller on a Cisco 5300 series Signalling System 7 (SS7) configuration is down at boot time, the B-channels on that downed controller will show "in service" when you use the show isdn {services} global configuration command on the command-line interface (CLI). There is no workaround.
  This situation occurs when Session Initiation Protocol (SIP) is configured to use Domain Name System (DNS) services and the domain name servers fail to respond to the queries sent from the router when Time To Live (TTL) for the cached DNS entries expire. SIP Voice over IP (VoIP) calls using DNS services might cause the router to reload.
  Workaround: Use high TTL values in the domain name servers (7200 seconds or higher). Use high availability platforms for domain name servers.
  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.
  A Cisco 7200 series router that is running Cisco IOS Release 12.1(1.6) and that is configured with Multilink PPP (MLP) and Cisco Express Forwarding (CEF) switching might report the following alignment error messages on the console:
  %ALIGN-3-CORRECT: Alignment correction made at 0x60DD3E0C reading 0xF9A052B
  %ALIGN-3-TRACE: -Traceback= 60DD3E0C 603CC26C 6011C1F8 60361E18 60363A94 60363B58 00000000 00000000
  This situation only occurs when you use MLP and CEF switching simultaneously. The alignment error count increases rapidly. The following message appears in the output of the show align command:
  Router> align
  Alignment data for:
  7200 Software (C7200-JS-M), Version 12.1(1.6), MAINTENANCE INTERIM SOFTWARE
  Total Corrections 8095, Recorded 1, Reads 8095, Writes 0
   Initial Initial
   Address Count Access Type Traceback
   F9A052B 8095 16bit read 0x60DD3E0C 0x603CC26C 0x6011C1F8 0x60361E18
   0x60363A94 0x60363B58
  Workaround: Disable either MLP or CEF.
  Dial-plan enhancements with wildcards "%," "?," "+," "[]," "()" and other features such as digit stripping are missing for voice features on the following Cisco products: 1700, 3600, 2600, 7200, and 7500 series routers; MC3810 multiservice access concentrators; and the AS5300 and AS5800 access servers.
  Workaround: Upgrade to Cisco IOS Release 12.1(2)T.
  A Cisco AS5300 series access server that uses new controller cards (either 4 ports or 8 ports T1/E1) with four serial interfaces might randomly reload. There is no workaround.
  While running Continuity Test (COT) regressions, a NAS-1 network access server might reload when a call is placed. After the call is placed, the script is not able to get console access. Call control block (CCB) coder-decoder (codec) related fields are not initialized. There is no workaround.
  A Cisco AS5800 series access server with 4 CE1 PRI cards disconnects incoming ISDN calls with the following message:
  May 31 21:33:35.256 PDT: ISDN Se1/3/1:15: Ux_BadMsg(): Invalid Message for call state 11, call id 0xF6C, call ref 0x811C, event 0x13A0
  There is no workaround.
  When a Non-Facility Associated Signalling (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 remains in the proposed service state. B-channels in the proposed service state are not usable for outgoing calls.
   A change in the digital subscriber lines (DSLs) state through configuration (using the shutdown and no shutdown commands) does not restore the B-channel(s) to IN_SERVICE. A reload is necessary. 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.
  While a call is connected with PRI-NI switch type and without a prior DISCONNECT message received, a release request from layers above ISDN might force a release message to be sent out and result in memory corruption. There is no workaround.
  A manufacturer serial number is appended with 0x format when it should be displayed as an alphanumeric string. There is no workaround.

Wide-Area Networking

  ISDN Layer 3 timers might not expire predictably when there is no activity. There is no workaround.
  Modem calls over BRI do not work correctly. There is no workaround.
  A Cisco 3661 series router reports the following error message when placing an outbound ISDN call:
  VOICE ERROR 0x3A.
  However, incoming calls work properly on the same router.
  Workaround: Upgrade to Cisco IOS Release 12.1(00.09)PI, 12.0(07)XK, 12.1(01.03)T, or later.
  A Cisco router might reload in cdapi_appl_send_message() during an overlap receiving call with primary Q (point of the ISDN model) Signalling (QSIG) switchtype if partial digits have been received and T302 ISDN timer expires. There is no workaround.
  A Cisco router might exhibit a traceback message and switch to monitor mode when an ISDN interface is shut or when the interface is cleared. There is no workaround.
  An ISDN User Part (ISUP) call generator that is configured on a Signalling System 7 (SS7) interface might experience a memory leak. There is no workaround.
  The redirect number feature is not working for PRI Net5 switches. There is no workaround.
  In violation of Q.931 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.
  A router might reload if an attempt is made to remove the ppp pap sent-username interface configuration command. The ppp pap sent-username command will also not work correctly. There is no workaround.
  A Cisco 4500 series router might reload during periods of high PPP session activity (cells per second, not data rates) in PPP Manager because of a stack overflow. There is no workaround.
  Layer 2 Tunnel Protocol (L2TP) or Layer 2 Forwarding Protocol (L2F) might fail to forward virtual private dialup network (VPDN) calls when using Challenge Handshake Authentication Protocol (CHAP) or Microsoft CHAP (MSCHAP) authentication.
  This condition might occur when clients that are to be tunneled negotiate CHAP authentication with the L2TP Access Concentrator/network access server (LAC/NAS). The following error message will be displayed if you use the vpdn debug error command:
  VPDN: Fail to copy authentication info
  Workaround: There is no workaround if CHAP or MSCHAP authentication is required. If CHAP or MSCHAP is not required, use Password Authentication Protocol (PAP) authentication.
  Overlap receiving for BRI might not work for NET3. Overlap is currently supported for NET3 and Q Signalling (QSIG) switchtypes only. There is no workaround.
  If Calling Party number information element (IE) is received without any digits, the calling_oct3a field is set to 0. Instead, it should fill in the Octet 3a of this IE, so that upper layer applications get it. There is no workaround.
  ISDN does not send the Called Party "Number Type" information received from the call switching module (CSM) in the outgoing SETUP message. There is no workaround.

Resolved Caveats—Release 12.1(1)T

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

Miscellaneous

  A Cisco uBR7200 series universal broadband router that is running Cisco IOS Release 12.07XR might show a high CPU utilization despite very few modems and low data traffic. This situation will cause slow response times, and the router might be inaccessible by Telnet. There is no workaround.
  A Cisco 3640 that is running Cisco IOS Release12.1(0.5)T with channelized T1 and digital modems, might experience very low (25 percent) modem call success rate (CSR). Many of the calls that do train up might be at very slow modulations such as V.21 or V.22bis. Nonetheless, some calls do train up at fast (V.34/V.90/K56Flex) modulations and, once connected, work well. ISDN data callers are not affected.
  In calls from a handset into the Cisco 3640 router, about 70 percent of callers hear no modem answer back tone. Of the remaining 30 percent that do hear answer back tone, only the end of the tone is heard (V.22bis, V.21, etc.). About 10 percent of the time, the caller will hear the full answer back tone sequence starting with the normal V.8bis. There is no workaround.
  On a Cisco uBR7200 series universal broadband router that is running Cisco IOS Release 12.1, the setting of max-cpe is enforced. There is no workaround.
  A Cisco 5800 series router might reload when an asynchronous call is dialed. There is no workaround.

Cisco Connection Online

Cisco Connection Online (CCO) is Cisco Systems' primary, real-time support channel. Maintenance customers and partners can self-register on CCO to obtain additional information and services.

Available 24 hours a day, 7 days a week, CCO provides a wealth of standard and value-added services to Cisco's customers and business partners. CCO services include product information, product documentation, software updates, release notes, technical tips, the Bug Navigator, configuration notes, brochures, descriptions of service offerings, and download access to public and authorized files.

CCO serves a wide variety of users through two interfaces that are updated and enhanced simultaneously: a character-based version and a multimedia version that resides on the World Wide Web (WWW). The character-based CCO supports Zmodem, Kermit, Xmodem, FTP, and Internet e-mail, and it is excellent for quick access to information over lower bandwidths. The WWW version of CCO provides richly formatted documents with photographs, figures, graphics, and video, as well as hyperlinks to related information.

You can access CCO in the following ways:

For a copy of CCO's Frequently Asked Questions (FAQ), contact cco-help@cisco.com. For additional information, contact cco-team@cisco.com.

Note   If you are a network administrator and need personal technical assistance with a Cisco product that is under warranty or covered by a maintenance contract, contact Cisco's Technical Assistance Center (TAC) at 800 553-2447, 408 526-7209, or tac@cisco.com. To obtain general information about Cisco Systems, Cisco products, or upgrades, contact 800 553-6387, 408 526-7208, or cs-rep@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, a member of the Cisco Connection Family, is updated monthly. Therefore, it might be more current than printed documentation. To order additional copies of the Documentation CD-ROM, contact your local sales representative or call customer service. The CD-ROM package is available as a single package or as an annual subscription. You can also access Cisco documentation on the World Wide Web at http://www.cisco.com, http://www-china.cisco.com, or http://www-europe.cisco.com.

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





hometocprevnextglossaryfeedbacksearchhelp
Posted: Tue Oct 3 07:21:23 PDT 2000
Copyright 1989-2000©Cisco Systems Inc.