Table of Contents
Release Notes for Cisco Network Registrar 3.0R
These release notes describe the caveats and new features for Cisco Network Registrar 3.0R0.
 | Warning
The upgrade procedures for Solaris have changed from earlier revisions. Read "Upgrading From Previous Releases" section before proceeding. Do not run the Solaris pkgrm command before reviewing these instructions. The pkgrm command will delete your data. |
This document contains the following sections:
- Introduction 2
- System Requirements 2
- Software Compatibility 3
- User Interface and Server Interoperability 4
- What's New in This Release 4
- Installing Network Registrar on Solaris 5
- Upgrading From Previous Releases 5
- Upgrading From CNR Release 1.x 5
- Upgrading from CNR Release 2.x Solaris 5
- Upgrading from CNR Release 2.x Windows NT Server 6
- Using Network Registrar 6
- Cisco Connection Online 7
- Documentation CD-ROM 8
- Resolved Bugs 8
- Unresolved Bugs 12
Introduction
The release of Network Registrar 3.0 contains the following components:
- CD-ROM containing the software
- Network Registrar Concepts Guide
- Getting Started with Network Registrar
- Network Registrar CLI Reference Guide
- Network Registrar CLI User's Guide
- Network Registrar GUI User's Guide
- Network Registrar Web GUI Installation Instructions
- Release Notes for Network Registrar 3.0
Network Registrar includes Windows® NT servers, CLI, and GUI, and Solaris® servers, CLI, and a Motif® GUI.
You can run the servers and the associated user interfaces on the appropriate platform, or you can run the servers on one platform and the user interfaces on another.
Network Registrar 3.0 runs on Windows NT 4.0, Solaris 2.5.1, and Solaris 2.6. Network Registrar's 3.0 GUI also runs on Windows 95.
For all OS platforms, Cisco recommends that you install the current service pack (Windows) or recommended patch cluster (Solaris).
- Windows NT 4.0 must be run with service pack 3, or for Y2K compliance, on service pack 4.
- Solaris 2.5.1 has been tested with the recommended patch cluster that was current on February 11, 1998. In particular, it was tested with these patches installed:
Table 1: Solaris 2.5.1 Patches
103461-20
| 103558-11
| 103566-25
|
103582-15
| 103594-13
| 103603-06
|
103612-38
| 103622-09
| 103630-09
|
103640-17
| 103663-09
| 103680-01
|
103686-02
| 103690-05
| 103696-03
|
103738-05
| 103743-01
| 103817-02
|
103866-05
| 103879-04
| 103900-01
|
103901-08
| 103934-06
| 103959-06
|
104010-01
| 104166-03
| 104212-09
|
104246-05
| 104266-01
| 104283-03
|
104317-01
| 104331-04
| 104334-01
|
104338-02
| 104433-06
| 104516-03
|
104533-03
| 104613-01
| 104650-02
|
104654-03
| 104692-01
| 104708-09
|
104736-03
| 104776-02
| 104795-02
|
104893-01
| 104935-01
| 104956-04
|
104958-01
| 104960-01
| 104968-02
|
104976-03
| 105004-09
| 105050-01
|
105092-01
| 105251-01
| 105299-01
|
105344-01
| 105352-01
| 105520-01
|
- Solaris 2.6 has been tested with the recommended patch cluster that was current on February 11, 1998. In particular, Solaris 2.6 was tested with these patches installed:
Table 2: Solaris 2.6 Patches
105216-01
| 105393-01
| 105518-01
|
105615-02
| 105621-01
| 105665-01
|
105667-01
| 105379-01
| 105786-01
|
105669-02
| 105558-01
| 105375-03
|
105357-01
| 105356-01
| 105407-01
|
Network Registrar 3.0 is compatible with the following:
- DHCP clients---All RFC 1954, 2132, and 2133 compliant DHCP and BOOTP clients
- DNS servers---Any DNS server compliant with RFC 2136 Dynamic DNS Updates and RFC 1995 and 1996 (IXFR and NOTIFY)
- LDAP servers---Any LDAP v2 or v3 server, including servers from Netscape, Sun Microsystems, and Lucent Technologies
There is a known problem on Solaris that may cause the DHCP server to hang during initialization. The problem is described by Sun RFE 4071167. The deadlock in the Solaris name resolver client can be encountered by installing a linked set of patches from Sun, including 103663-08, and disabling host name caching in "nscd" (to improve name resolver client performance), which is recommended by Sun in RFE 1243174.
To avoid this problem, do one of the following:
- Do not install the linked patch set.
- Do not disable "nscd" host name caching.
- Install patch 103663-11 (Solaris 2.5.1) or 105755-01 (Solaris 2.6).
The following table describes the typical outcome when you run different versions of the user interface and the servers:
Table 3: Network Registrar Interoperability
| GUI and CLI
| Server
| Outcome
|
3.0
| 2.5
| Network Registrar runs successfully.
|
2.5
| 3.0
| Network Registrar UI displays the error message "Unsupported version number." The server is unaffected.
|
Cisco Network Registrar Release 3.0 offers many new features:
- DHCP Failover---Provides DHCP redundancy so that any one DHCP server is not a single point of failure. Failover allows main and backup DHCP servers to share responsibility for servicing any one DHCP scope. When both servers are operating, the main server handles all DHCP packets and informs the backup server when IP address leases are assigned, extended, or expired. If the main server fails, the backup server automatically assumes responsibility for servicing client DHCP packets.
- SNMP Notification---Can be used to warn of error conditions and possible problems with the DNS and DHCP servers. SNMP notifications can be sent for exhausting DHCP server scope addresses, exceeding a user-definable threshold of free addresses, starting or stopping a server, and other conditions.
- Data Export to ODBC---Able to export DNS and IP address information to a CSV text file or ODBC database. This allows administrators to export DNS and IP address information to any database. spreadsheet of other tool that supports CSV text file input of ODBC.
- DHCP Option Validation---Better validation of DHCP server options. DHCP options have prescribed formats and legal values for their option parameters. Network Registrar checks user-supplied option values against the legal values.
- Dynamic DNS Name Deletion---Able to remove all specific dynamic resource records. You can specify the resource records by name, or name and type.
- Faster GUI Start-up---Significantly reduces the amount of time it takes to start the Network Registrar GUI. The server manager displays a standard Windows tree control that contains a list of clusters at its top-level mode. Only when you click on a node, does Network Registrar read the contents of the node and thus eliminates the wait for unneeded information.
- Network Registrar Concepts Guide---Written for system evaluators and administrators, it provides conceptual information about Network Registrar features. It includes chapters on client-class, DHCP failover, data import and export, SNMP notification, LDAP, and other topics.
- Support for Other Operating Systems---This release of Cisco Network Registrar supports Windows NT, Solaris, HP-UX, and AIX.
If you are installing Network Registrar 3.0 for Solaris for the first time, follow the instructions in the Getting Started with Network Registrar guide, but instead of specifying pkgadd -d <CD-ROM>, where <CD-ROM> is the CD-ROM's mount point, specify pkgadd -d <CD-ROM>/solaris/see_readme_before_using.
For additional information, see the file README.solaris in the top-level directory of the CD-ROM.
When you upgrade from a previous release of Network Registrar, you can choose to save your database or overwrite it. If you overwrite your database you lose all information about your zones, scopes and leases.
To upgrade from any Network Registrar release numbered 1.x, you must first upgrade to release 2.0 before upgrading to release 3.0. Contact the Technical Assistance Center if you need additional information.
 | Warning
The upgrade procedures for Solaris have changed from earlier revisions. Do not run the Solaris pkgrm command before reviewing these instructions. The pkgrm command will delete your data. |
To upgrade from Network Registrar release 2.x, you must run the program upgrade_cnr in the solaris subdirectory of the CD-ROM. If you use the standard configuration of the Solaris automounter, this program will be located in the directory /cdrom/nr_3_0r0/solaris/upgrade_cnr.
The upgrade program does the following:
- Backups your existing data base
- Uninstalls your current version of Network Registrar
- Installs the latest version of Network Registrar at the location where the previous version was installed
Note You can not use the upgrade program to install a new version of Network Registrar in new location. You can not use the upgrade program to select options such as Complete, Server, and GUI. The upgrade program only upgrades the options that were previously installed.
To upgrade from Network Registrar 2.x to Network Registrar 3.0 on a Solaris server and to preserve your database, do the following:
Step 1 Login as root.
Step 2 Run the /cdrom/nr_3_0r0/solaris/upgrade_cnr program.
# /cdrom/nr_3_0r0/solaris/upgrade_cnr
If you mount the CD-ROM manually or have a non-standard automounter configuration, then replace /cdrom/nr_3_0r0 with the directory on which the CD-ROM is mounted.
Note You must run the script with an absolute path, as shown above. If you run it with a relative path (for example, by typing "cd /cdrom/nr_3_0r0/solaris" and then "./upgrade_cnr"), it will not work.
Step 3 Follow the instructions on the screen.
This completes the procedure for upgrading from Network Registrar 2.x to Network Registrar 3.0 on a Solaris server. When the upgrade is complete, Network Registrar's services will start automatically.
To upgrade from Network Registrar 2.x to Network Registrar 3.0 on a Windows NT server and to preserve your database, do the following:
Step 1 Run SETUP.EXE.
Step 2 Follow the instructions that appear on your screen.
Answer Yes to the question about upgrading your existing installation. This will preserves the data in your database.
To upgrade from Network Registrar 2.x to Network Registrar 3.0 on a Windows NT server without preserving your database, do the following:
Step 1 Run the uninstall program.
Step 2 Perform an initial installation as described in the "Installing Network Registrar on Windows" section on page 1-8 of the Getting Started Guide.
Note To non-interactively upgrade a previous release of Network Registrar, see the instructions in the text file \i386\UPGRADE2.ISS on the Network Registrar CD.
If you plan to use failover on a machine with multiple interfaces, you must explicitly configure the local server's name or IP address. Remember, only if the machine has a single interface, can you omit the name or IP address for the local server.
For instance, to configure server A as main and server B as backup in a situation in which each has only one interface, you would do the following:
Step 1 On server A specify the name of the backup server, by typing:
nrcmd> dhcp set failover-backup-server=serverb
Step 2 On server B specify the name of the main server, by typing:
nrcmd> dhcp set failover-main-server=servera
In this case, the lack of configuration for a main server on server A and the existence of the configuration for a backup server causes server A to assume it is a main server. Since server A has only one IP address and name, it uses that interface and IP address to communicate with server B.
If, however, server A has multiple interfaces and multiple IP addresses (and presumably multiple DNS names, one per IP address and interface), then the situation is more complicated.
For instance, if server A has two interfaces with the two DNS names servera and serverprime, and you were to configure server A as above, it would choose (unpredictably) a single interface from which to communicate with server B.
- If it chooses the same interface as configured on server B as the main server (in the example, servera), then failover will initialize correctly.
- If it chooses serverprime as the interface with which to communicate with server B then failover will not initialize correctly.
The correct way to configure a main server that has two interfaces is to explicitly name the interface on which you wish it to communicate with the backup server:
Step 1 On server A, specify the backup server, by typing:
nrcmd> dhcp set failover-backup-server=serverb
Step 2 On server A, specify the main server, by typing:
nrcmd> dhcp set failover-main-server=servera
This specifies which interface server A needs to use to communicate with server B, and ensures that failover will initialize correctly.
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.
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.
- In the graphical user interface, if you attempt to edit in a displayed tree a node name that is read-only, the graphical user interface will highlight the name, thus implying that it can be edited, although it cannot. If the name cannot be edited, it should not be highlighted when the user clicks on it. [CSCai00517]
- When a lease is activated or deactivated, the new state of the lease may not show up immediately in the "Lease Properties" tab of the graphical user interface, and the "deactivated" column may also be incorrect. Click the refresh button to cause the correct state to appear in the tab. [CSCai00705]
- When an administrator other than "admin" has locked a cluster, the "Cluster Locked" window does not display the name of the administrator. [CSCai00728]
- The DHCP standard specifies that each NAK packet returned by the DHCP server should have the same flags set as the request for which it is a response. However, Network Registrar may sometimes return a NAK packet with different flags from those in the request. [CSCai00803]
- When you add a new cluster in the graphical user interface, it collapses the subtrees for all other displayed clusters, rather than simply adding the new cluster to the tree in a collapsed state. [CSCai00967]
- If the Network Registrar server agent is not running, the graphical user interface will not start on Solaris. [CSCai01670]
- If you define a resource record with the same name as the zone, a dot is appended to the end of the name of the resource record. However, when you change the name of this resource record, the SOA record for the zone is deleted, and the NS records are changed to match the changed name of the resource record. Make sure to add the SOA record and restore the correct values for the NS records if this happens. [CSCai01763]
- If you create a subzone and type in capital letters, you may see a duplicate NS record in the resource tab, displayed in all lower case. To avoid this problem, use lower case letters to define your subzone names. [CSCai01785]
- You may use a maximum of 30 uniquely named scope selection tags for client-class configuration of all the scopes in a cluster. If the DHCP server encounters more than 30 tags when reading its configuration, then any scopes it encounters with "extra" tags will fail to configure and error messages will be logged for them. However, the server will start up and serve requests for any scopes processed before the extra tags were encountered. [CSCai02767]
- The force-lock command will return an error message if no lock currently exists. The error message is 407 Unable to release lock - 0xb unlock failed, err = 0xb [CSCai02914]
- The graphical user interface may take a long time to connect to clusters with large databases. [CSCai03039]
- If successful, the force-lock command returns the following output:
100 OK - 100 OK
- It should only print "100 OK" once. [CSCai03086]
- Although negative lease time values are not valid, the policy setLeaseTime method in nrcmd does not check that the value entered is non-negative. [CSCai03092]
- If you add or delete hosts using nrcmd, the changes will not be reflected in the Hosts tab of the DNS Server dialog in the graphical user interface. [CSCai03147]
- If a DHCP server attempts to delete a dynamic update to a DNS server that is no longer listening, subsequent DDNS updates to any server will hang in the queue and not be processed. Contact Technical Support if you think you are having this problem. [CSCai03193]
- "Add to status Monitor" is available from the Servers menu of the graphical user interface for items which cannot be added, such as scopes, clusters, and zones. [CSCai03263]
- If you have two (or more) clusters with different administrator passwords configured in the graphical user interface, you specify the correct password to connect to the first cluster, and then you click on "Cancel" when prompted for the username and password for the second cluster, the user interface will prompt you again for the username and password for the first cluster, rather than remembering that you have already specified it. [CSCai03307]
- The CLI Reference Guide now states that importing permanent leases will not work if you have disabled the permanent leases option. [CSCai03378]
- You cannot add MX records to a zone using the graphical user interface on Solaris. Either enter these records using the interface on Windows, or use the zone name addRR command in nrcmd. [CSCai03477]
- The on-line help for the extension command in nrcmd doesn't indicate that entry is a required argument. [CSCai03653]
- When you add the "policy-filters" option to a policy in the graphical user interface, you must specify a list of IP address/mask pairs. If you enter an odd number of entries (i.e., you enter an address without a mask or vice versa) and click "OK," the message "Invalid IP address" appears, even though all the IP addresses and masks you have specified are valid. Make sure you have entered pairs of IP addresses and masks for this option. [CSCai03712]
- If you supply an odd number of entries for the path-mtu-plateau-tables option when editing policy options in the graphical user interface, the error message "Invalid entry -- you must supply a value between 0 and 65535" appears, even though all the numbers you specified may have fallen within this range. Make sure you supply an even number of entries for this option. [CSCai03713]
- If the DHCP server fails to reload when told to do so, nrcmd may report the incorrect error "322 Feature not supported - dhcp reload {}." The failure to reload is real, but that particular error is incorrect. [CSCai03725]
- When you uninstall Network Registrar from a UNIX machine, the directory /opt/nwreg2 (or whatever non-default base installation directory you specified when installing Network Registrar) is deleted recursively without warning. [CSCai03735]
- The Network Registrar User's Guide appendix entitled "Network Registrar Logging and Debugging" now explains the size and rollover philosophy for server log files. [CSCai03737]
- The command lease withmacaddr 00:a0:2:2f:0e:bd in nrcmd returns the message "302 Not Found lease withmacaddr 00:a0:2:2f:0e:bd." [CSCai03744]
- In the graphical user interface, when you add a subzone or edit an existing subzone and change its name, do not supply a fully qualified domain name; instead, type only the relative name of the subzone. For example, to add the subzone "subzone1" to the zone "myzone.com.," type "subzone1" rather than "subzone1.myzone.com."
- If you enter the fully qualified domain name, the interface will create an improperly named subzone which cannot be removed using the Remove button on the Subzones tab. The subzone will appear to have been removed, but its resource records will not be removed and, after reloading, the subzone will reappear in the list on the Subzones tab. If this happens, go to the Resource Records tab and delete the relevant resource records individually. [CSCai03757]
- DHCP leases exported with the export leases -server command in nrcmd have incorrect client IDs. As a result, if the exported leases are imported back into the server, clients will not be able to renew them, and furthermore, they will be marked unavailable whenever a client attempts to renew. [CSCai03779]
- When the DHCP server reloads, it marks unavailable any lease which has the character "%" in any of its string-type options. The "%" character should only be prohibited in boot-file options, not in lease string-type options. [CSCai03784]
- The Network Registrar Server Agent may incorrectly log a "bad password" error the first time a server is reloaded from either the graphical or command-line interface. [CSCai03865]
- The Network Registrar DNS server compresses the data field of SRV records it returns to clients. Some older clients and name servers are unable to parse SRV records with compressed data fields. This will be fixed by disabling compression of SRV records' data fields. [CSCai03921]
- When the DNS server receives a request for an incremental zone transfer (i.e., an IXFR request) and replies that the client is up-to-date and does not need an update, it logs a misleading message which implies that a zone transfer has taken place or is about to take place. The misleading log message should be removed. [CSCai03923]
- When the DNS server receives an incremental zone transfer request with a serial number for which it has no history (e.g., if the zone is changing so quickly that the history has already been expired by the server), it leaks a chunk of memory. Over time, the leaked memory can cause the server image to grow very large. [CSCai03924]
- Any requested-options list sent by a client in an INFORM packet is ignored. Furthermore, the server doesn't send the options in the dhcp-reply-options list of a client's or a scope's policy. [CSCai04000]
- In certain rare cases (depending on the size of the option list), deleting an option from a policy corrupts the option list. [CSCai04032]
- The Maximum DNS Record Time To Live setting, configured in the Advanced DNS Tab of the DHCP Server dialog in the graphical interface or through the dns-max-ttl DHCP server option in nrcmd, is being ignored by the DHCP server. As a result, the time to live for Dynamic DNS records is always set to 1/3 of the DHCP lease time. [CSCai04074]
- nrcmd report and lease-notification commands no longer lock the cluster on which nrcmd is being run. This is consistent with the fact that additional specified remote clusters are not locked. [CSCai04104]
- If you upgrade from a previous Network Registrar release and preserve your existing database, the database schema file may not be replaced correctly with the new one in this release. In most cases this will be benign, but if suspicious behavior is observed, specifically if the DNS and/or DHCP servers start reloading continuously, please contact support for assistance. [CSCdk90275]
- The graphical user interface crashes if you mark a scope secondary in the Advanced scope tab when there are no primary scopes in the cluster. [CSCdk90437]
- The graphical user interface may crash if you use it to remove a forwarder from a DNS server. Use nrcmd to remove forwarders. This will be fixed in Network Registrar 3.0B1 and subsequent releases. [CSCdk91271]
- When running on Windows, the graphical user interface might fail to start and instead display an error about a missing file IMAGEHLP.DLL or MSVCIRT.DLL. [CSCdk91884]
- nrcmd may sometimes add an hour to the time specified for the authenticate-until client parameter. [CSCdm00594]
- If DHCP is configured to perform DNS updates, it will retry the update for a client if the client renews its lease before the initial DNS update is processed at the DNS server. This may increase the work load of the DNS server. [CSCdm16013]
- If your DHCP server is sending dynamic DNS updates to your primary DNS server, and you do not have a secondary DNS server doing occasional zone transfers from your primary server, the size of the primary DNS server in memory can grow without bound. [CSCdm17744]
- The graphical user interface allows the user to enter illegal characters in DNS zone names; it should enforce DNS protocol restrictions on characters in zone names. To avoid this problem, use only letters, numbers, hyphens and digits in zone names. [CSCai00138]
- The graphical user interface permits the entry of illegal IP addresses such as "0.1.1.1." [CSCai00511]
- If the mcdadmin command is used to recreate the Network Registrar database while it is open for use, any clients which have the database open will probably crash. The Network Registrar servers should be shut down before rebuilding the database, and even then, it should only be done under the direction of Technical Support. [CSCai00586]
- Network Registrar servers and clients may sometimes print or log errors as hexadecimal constants rather than explanatory strings. These constants should be translated into text before they are displayed or logged. [CSCai00602]
- The graphical user interface permits the entry into DHCP options of domain and host names that are too long or improperly formatted. [CSCai00660]
- When the lease time for a policy is changed and there are outstanding leases in the affected scope, their lease times will not be changed until they expire or are renewed, and the administrator receives no warning to that effect. [CSCai00674]
- Section 4.3.1 of IETF RFC 1531, which specifies the Dynamic Host Configuration Protocol, specifies that the DHCP server must, when responding to DHCPDISCOVER requests, return to the client "Any parameters from the existing binding that differ from the Host Requirements documents defaults." Network Registrar does not currently meet this requirement. [CSCai00698]
- When a cluster being edited by an administrator is unlocked by a second administrator, the first administrator does not receive any notification that the second administrator unlocked the cluster. Before unlocking a cluster, you should make certain that the cluster is not actively being edited. To help you do this, the message indicating that a cluster is locked provides the machine name and process identifier of the locking process. [CSCai00704]
- The Window menu in the graphical user interface continues to list the Server Manager window after it has been dismissed. [CSCai00710]
- In the graphical user interface, you can only enter DHCP lease times with a granularity of one minute. Use nrcmd to enter lease times which are not integral multiples of minutes. [CSCai00723]
- If you create DHCP options whose lengths total more than 246 bytes, some Microsoft clients may be unable to accept them. You should ensure that the total of the lengths of the following options is less than 246 bytes: Subnet Mask, Routers, Domain Name, Domain Name Servers, Netbios Name Servers, NetBios Datagram Distribution Servers, and NetBios Node Type. [CSCai00734]
- The graphical user interface can become confused when importing a BIND file whose SOA record is not at the top of the file or which contains multiple SOA records. To avoid this problem, ensure that any BIND file you import has only one SOA record, at the top of the file. [CSCai00773]
- In the graphical user interface, you cannot make the status monitor window as thin as the icons in it (i.e., the interface enforces a minimum width which is too large). Furthermore, you can make its height so small that you cannot see the icons. The minimum size on the window should be thinner and taller. [CSCai00978]
- If you edit the "lease properties" of a DHCP lease in the graphical user interface and check the "reserve lease" checkbox, but do not enter a MAC address, the reservation is not entered into the Network Registrar database and no error or warning is generated. Make sure that you enter a MAC address if you wish to reserve a lease. [CSCai01006]
- If you are connected to a cluster with the graphical user interface and someone stops and restarts the Network Registrar servers for that cluster, the user interface claims that all leases are available. To correct this problem, exit and restart the user interface. [CSCai01027]
- When the Network Registrar servers are stopped while the server is being configured, data may be corrupted. Do not stop the servers while they are being configured. [CSCai01041]
- If you are monitoring a remote server in the graphical user interface using the "Server Status Monitor," and the server stops responding to network requests (e.g., because it crashes, it is disconnected from the network, or its subnet becomes unreachable), the graphical user interface may hang. The interface should become responsive again within one minute after the server starts responding again. [CSCai01044]
- In the graphical user interface, if you select a scope, click on "reservation," add a mac address, select "deactivate," and restart the server, the "R" field is not checked.
- Furthermore, if you double-click on the lease again, the "reservations" check box is empty, and the address you typed into the "MAC Address" edit control now shows up in the "Leased to MAC address" edit box.
- To avoid this problem, add the "reservation" check and reload the server, then choose "deactivate." [CSCai01059]
- To use the graphical user interface or nrcmd from a Windows 95 machine, the machine must be configured with valid WINS servers. [CSCai01069]
- If you rename a server from the tree control of the graphical user interface, the interface does not display the reload indicator for the server. However, you must still reload the server for the new name to take effect. If you rename a server from the properties tab instead of the tree control, then the reload indicator is displayed correctly. [CSCai01083]
- The graphical user interface may allow you to assign to a host multiple MX records with the same host name. To avoid this problem, do not assign redundant MX records to a host, or if you do this and then subsequently realize it, delete the redundant records. [CSCai01110]
- When creating a DNS zone by importing a BIND file, the following resource record types are not supported: PX, LOC, KEY, NXT, and SIG. [CSCai01206]
- The graphical user interface permits users to create DNS names which are longer than the maximum of 253 characters. In some cases, fully qualified domain names which are greater than this length will be discarded. In other cases, these long fully qualified domain names will prevent the DNS servers from being reloaded. Avoid using fully qualified domain names longer than 253 characters. [CSCai01215]
- Network Registrar does not verify the data in WKS records. Be sure to enter this data correctly. [CSCai01223]
- It is possible to enter multiple WKS records with the same name, address, and port #. However, the server will only send one of these WKS records in response to a query. [CSCai01224]
- When you add a resource record of type AFSDB, MG, MB, MR, MX or RT in the DNS zone properties dialog of the graphical user interface, the value of the record is silently truncated to 63 characters. [CSCai01264]
- If you change the TTL of a Name Server record through the Resource Records tab of the graphical user interface, the change is not immediately captured on the Name Servers tab. You must close and re-open the zone to see the new TTL value. [CSCai01265]
- When you click on the "View recent log events" button of the graphical user interface after reloading the server, the title of that window, "Validation errors," is inaccurate. The window may contain activity, informational and/or error messages, not just validation errors. [CSCai01327]
- When you reload a server through the graphical user interface, the confirmation message says that the server has been "restarted." The message should instead say that the server has been reloaded. The server behavior is correct. [CSCai01328]
- When the Network Registrar database is large and the disk on which it resides is slow or very busy, the nightly "shadow backup" of the database may cause noticeable service interruption. The database cannot be modified while it is being backed up, so servers and clients attempting to write to it (e.g., to save information about a newly granted lease) are delayed until the backup completes. To mitigate this problem, ensure that the shadow backup is scheduled for a different time on each of your Network Registrar servers. [CSCai01348]
- If a DHCP server logs an error containing "80010023," the server's socket buffer may be filling up. To solve this problem, increase the "min-buffer-size" DHCP interface setting using nrcmd. For example:
nrcmd> session set visibility=3
100 OK
visibility=3
nrcmd> dhcp-interface default get min-buffer-size
100 OK
min-buffer-size=32768
nrcmd> dhcp-interface default set min-buffer-size=65536
100 OK
min-buffer-size=65536
nrcmd> dhcp reload
100 OK
- [CSCai01366]
- If you enter a zone serial number greater than 2147483648 into nrcmd, it will be changed to 2147483648. [CSCai01448]
- In the graphical user interface, if you click on the "New" button while editing a policy and specify that you wish to copy the new policy from the one you were editing, any unapplied changes you made to the old policy will not be copied to the new one. Before copying a new policy from one you are editing, click on "Apply" to ensure that your changes are copied to the new policy. [CSCai01476]
- In the graphical user interface, if you reserve, activate, or deactivate a lease in the Lease Properties dialog and then click "OK," the "Apply" button should be enabled but is not. Your changes have been applied to the database, but you must reload the DHCP server for them to take effect. [CSCai01526]
- In the graphical user interface, using the "tab" key to move between fields behaves inconsistently on different screens. [CSCai01547]
- If you create a CNAME record whose name is outside of the zone, the graphical user interface will continue to display the name as entered. However, the DNS server will append the name of the zone to the name of the CNAME. For example, if you enter a CNAME called "host1.local.com." in zone "local2.com," the server will interpret this name as "host1.local.com.local2.com." [CSCai01595]
- If the DHCP server receives an ICMP error message in response to an ICMP echo (i.e., ping) while attempting to allocate an IP address for a client, the server will allocate the IP address to the client. The log message when this occurs implies that the IP address was marked unreachable in the database, but it was not. [CSCai01639]
- Changes you make to column widths on the "Lease properties" screen of the graphical user interface are forgotten when you close the screen. [CSCai01652]
- In the graphical user interface, the last line of data in the Subzones tab of the Zone Properties dialog may be displayed incorrectly when the tab displays more than one screen of data. This does not affect the integrity of the data in the database. [CSCai01671]
- When using the graphical user interface on Solaris, activating the vertical scrollbar in the DHCP leases display causes the list of leases to be redrawn twice. [CSCai01704]
- If the server suffers a power failure while configuration changes are being written to the database, the database may become corrupted. To avoid this problem, Network Registrar servers should be run on uninterruptable power supplies, and when power is lost, the servers should be shut down cleanly before the UPS is drained. [CSCai01720]
- On a Network Registrar server machine, numerous files are added over time to the data/db/logs directory, and some of them are neither deleted nor rewritten. Because these files contain information which may be essential for proper data recovery, they may be deleted only after the Network Registrar servers have been started and stopped without error. In no case should they be deleted while the product is running or directly after a crash or other malfunction. [CSCai01753]
- The client host names in the output of the nrcmd export leases command are the names requested by clients, not the names entered into DNS for the leases. You can find the actual DNS host name for a given lease by resolving its IP address through the DNS server. [CSCai01767]
- If you run the graphical user interface on the server host, it will allow you to make multiple simultaneous connections to the server by connecting to it using both "localhost" and the server's real host name. Confusion or data corruption may result if you do this and then make changes using both connections to the server. To avoid this problem, do not connect to the same server multiple times using different host names. [CSCai01776]
- In some cases, you may be able to connect to a database from two different workstations without receiving a warning that the database is already in use. [CSCai01777]
- In the graphical user interface, if you configure a DNS resolution exception, save and reload your zone, and then delete the resolution exception, it may not be removed from the DNS configuration. If this occurs, you can delete the exception by first disconnecting from and reconnecting to the cluster. [CSCai01779]
- When using root-path substitution in the boot-file DHCP option (i.e., when using @%root-path%), if the resulting string is too large to fit into the boot-file field (127 bytes), then it is truncated and no log message is issued. To work around this problem, ensure that after substitution the resulting string is less than 127 bytes long.
- This problem does occur with the boot-file option. In that case, if the string becomes too large after substitution to fit into a DHCP option (255 bytes), then the option is returned to its pre-substitution state and a log message is generated. [CSCai01809]
- When a machine is dual-boot, from a DHCP standpoint it has two personalities. Each of these personalities may have a separate DHCP address, sometimes (when the servers are about equally fast) on different DHCP servers. If this is the case, then when the user goes back and forth between operating systems, the A record in DNS will be wrong sometimes.
- This only occurs when there are two DHCP servers servicing the same subnet (typically for reliability). It is also only the case when the two different operating systems that are being used have the same host name (which is configurable) and the same client-id (which would be the case for Windows 95 and Window NT 4.0 running on the same hardware).
- One workaround is to make sure that each operating system on a machine is configured with a different host name. [CSCai01820]
- The lease time returned by the DHCP server in a BOOTP packet when the client specifies the lease-time option is incorrect. BOOTP leases are permanent, but rather than indicating a permanent lease, the server returns the lease expiration time for non-BOOTP leases in the client's scope. [CSCai01831]
- In the Scope Properties DNS tab of the graphical user interface, if you enter an invalid Reverse zone, you may get the error, "Please enter an integer between 1 and 3." This error is misleading. The error should read "Please edit the reverse zone name." [CSCai01835]
- In the graphical user interface, the tab key cannot be used to traverse all input fields on the Root Name Servers tab of the DNS Server Property Page or the DHCP tab of the Zone Property Page. Use mouse clicks to access fields to which you cannot tab. [CSCai01860]
- In the graphical user interface, sorting on the "Data" column of the Resource Records tab of the DNS Zone Properties dialog does not sort in alphanumeric order. [CSCai01864]
- In the graphical user interface, if you are editing data in a table, you click on the "?" context-sensitive help button, and then you click directly on the table cell which you are editing, you may see the message, "No help topic is associated with this item." To see the appropriate help text for the table, click on a cell other than the one you are editing. [CSCai02158]
- If you try to connect to a cluster from the graphical user interface very soon after the Network Registrar servers have started, you may get an "Unable to connect" message. If you get this message, exit and restart the user interface. [CSCai02173]
- If you change the policy for a DHCP scope in the graphical user interface, and then you press the ESC key or hit the cancel button, the changes are discarded without warning. [CSCai02174]
- When you are entering data into the SOA tab while creating a new zone, the graphical user interface may allow you to enter illegal characters for some data. In particular, the interface will allow you to enter the character '@' in the "contact e-mail address" field. You should replace the '@' with '.' in the address you place in this field. [CSCai02179]
- In the graphical user interface, the left and right arrow keys do not reliably switch among tabs on dialogs with multiple tabs. Use the mouse to switch tabs, or ensure that keyboard focus is on the tab itself, rather than on one of its controls, before using the arrow keys to switch tabs. [CSCai02181]
- If you edit the name of an existing host in a zone using the "Hosts" dialog tab and change the case of one letter from lower to upper, the change is not reflected in the Resource Records dialog, although you will see the change on the Hosts tab. [CSCai02241]
- In the graphical user interface on Solaris, if you have a large number of entries in the Server Manager tree, you may have difficulty scrolling to the bottom of the list. When this happens, a second, smaller vertical scrollbar appears inside the Server Manager frame. This second scrollbar may be difficult to see when using the OpenWindows Desktop; we recommend using the Solaris graphical user interface under the Common Desktop Environment (CDE) instead of OpenWindows.
- To work around this problem, make the graphical user interface's top-level window and the Server Manager window as large as possible, which may make the second vertical scrollbar disappear. If it does not, scroll to the bottom of both scrollbars to reach the bottom of the data. [CSCai02465]
- In the graphical user interface on Solaris, if you have a large number of entries in the Server Manager tree and a second, smaller vertical scrollbar appears, dragging and dropping the inner scrollbar rapidly may cause sections of the tree to be redrawn improperly. This is a display problem only. To remedy the problem, select View > Server Manager from the menu to close the Server Manager window, and then select View > Server Manager again to re-open it. [CSCai02467]
- On an NT host with Remote Access services enabled, the AIC Server Agent may crash on shutdown with the error "aicservagt.exe --- DLL Initialization Failed Initialization of dynamic link library c:\winnt\system32\rasman.dll failed. Process is terminating abnormally." This problem can be worked around by stopping the AIC Server Agent in the control panel before system shutdown. This problem exists in Windows NT 3.51 or 4.0; however, for NT 3.51, it is fixed by installing Microsoft NT 3.51 Service Pack 5. We believe that this is a bug in Microsoft NT, and will be working with Microsoft to resolve the issue. [CSCai02507]
- In the graphical user interface, you cannot swap the name and an alias of a host in a single operation. To switch a host's name with one of its aliases, bring up the Host edit dialog, remove the alias and click "OK," then bring up the Host edit dialog again, rename the host and add its old name as an alias. [CSCai02539]
- If a host's name is the same as the name of the zone, the zone will be difficult to modify. Do not create hosts with the same name as the zone that they reside in. [CSCai02619]
- Extra command arguments are silently ignored by nrcmd. For example, extension list foo bar behaves the same way as extension list. [CSCai02663]
- In the graphical user interface, unchecking a reserved lease may result in "garbage" text being displayed. To avoid this, do not perform the following sequence of steps:
- On the "Leases" tab of the DHCP Scope Properties dialog, select a reserved lease that is outside the set of address ranges listed for the scope.
- Click the "Lease Properties" button, and then uncheck the "Reserve a Lease" checkbox.
- Click OK.
- Back on the "Leases" tab, click "Refresh List."
- You may now see garbage text for that lease entry.
- If this happens, exit the DHCP Scope Properties dialog (using "OK") and re-open it to synchronize the display with your changes. If, instead, you click "Cancel" and then disconnect from the cluster or exit from the user interface, you will see a dialog stating that the DHCP server has configuration data that has not been saved. If you click "Yes" or hit Return to indicate that you want to save the data, then the lease will be removed as you intended. If you click on "No," the lease will not be removed. In either case, the user interface display will be correct when you reconnect to the cluster. [CSCai02690]
- If a client's host-name token is invalid, all requests from that client will be dropped, after logging a message like this:
03/24/1998 17:26:59 name/dhcp/1 Warning Configuration 0 While processing a packet from Client MAC: '1,6,01:55:55:55:55:55', an unrecognized host name token '@st-name-jreio' was found in the client. Packet dropped'.
- Do not configure a host name with a '@' on a DHCP client, and be sure when configuring host name tokens that they are correctly formatted. [CSCai02707]
- If you specify an incorrectly formatted file to the import leases command, the command will incorrectly return a 100 OK response, followed by one or more 316 error responses. The bad leases will not be imported. [CSCai02791]
- If you specify a non-existent address to the lease addr force-available command in nrcmd, you will get an internal error rather than a standard "Not found" error. [CSCai02793]
- If you attempt to connect to a cluster for which the server is not started, you get the message "Unable to connect to cluster clustername, item not found." The phrase "item not found" should not appear. [CSCai02827]
- If the Network Registrar Server Agent is restarted when a graphical user interface is connected to it, some user interface commands (e.g., stats refresh) won't work until you disconnect from and then reconnect to the cluster. Otherwise, you may get the error message "Unable to communicate with server." [CSCai02833]
- The DNS NOTIFY feature can cause a significant load on the server in certain cases. This is primarily a problem when many, many zones are changing very quickly and attempting to generate frequent NOTIFY packets, coupled with zone name servers which are non-existent or whose names are not within the zone, and whose IP address cannot be resolved. [CSCai02902]
- If the DHCP server is not running, the nrcmd export leases command will fail with the error "502 Server Failure." You should ensure that the DHCP server is running before exporting lease information. [CSCai02973]
- In nrcmd, the commands unset, enable and disable will not permanently affect the dhcp-interface ignore attribute. If you use unset or disable to change the value of ignore from enabled to disabled, the change will not be saved to the configuration database.
- The set command will cause the change to be persistent. For example, to enable an interface 10.1.1.1/24, you would use the command:
nrcmd> dhcp-interface 10.1.1.1/24 set ignore=false
- To disable this interface, use the command:
nrcmd> dhcp-interface 10.1.1.1/24 set ignore=true
- [CSCai03064]
- In the graphical user interface, reserving an active lease for a client other than the one already associated with it displays incorrect data in the "Reserved for MAC address" field of the Lease Properties dialog. The original client's MAC address will be displayed until that client acquires a new lease or the new client acquires the reserved address. The Scope Reservations property page shows the correct information. [CSCai03066]
- In nrcmd, the zone SOA properties refresh, retry, expire, and minttl cannot be set using the weeks, days, hours, minutes syntax for time values. You must specify a decimal number of seconds for these values. For example, the command "zone zone set minttl=2h" will incorrectly set the minttl property to two seconds. The command "zone zone set minttl=7200" will correctly set the value to 7200 seconds, i.e., two hours. [CSCai03068]
- A DNS server with secondary zones may be slow to stop or reload in some cases while trying to connect to a nonexistent remote DNS server for zone transfer. [CSCai03071]
- The output of the nrcmd listaddr command is:
100 OK
list of addresses
100 OK
- This is different from other commands, which print "100 OK" only once. [CSCai03088]
- When you exit from nrcmd, the lock for the cluster is forcibly removed, even if it has been stolen by another user. This will leave the cluster unlocked, even though a user interface session may think it has it locked. To avoid this problem, do not use the force-lock command in nrcmd unless you are sure no one else is editing the database. [CSCai03090]
- In nrcmd, the dynupdate-set property can be edited and viewed for secondary zone configurations. This is incorrect, and the configured data will have no affect on the behavior of the DNS server with respect to this zone. [CSCai03097]
- In the graphical user interface, if you import a BIND file containing A records with 64-character labels, the Zone Import Errors dialog correctly appears to warn of the overly long names, but the incorrect records are imported anyway.
- Under normal circumstances, you should never encounter BIND files with 64-character names. Consult RFC 1035 sections 2.3.1 and 2.3.4 (available at http://www.ietf.org/rfc/rfc1035.txt) for the definitive specification of name length limits. You should avoid importing BIND files with too-long names, because they may cause database corruption. [CSCai03098]
- In nrcmd, you can create a scope with a mask that is not consistent with the network number for the scope. A mask which has a zero bit where there is a one bit in the network number should not be allowed. [CSCai03103]
- A Network Registrar server running on Windows NT may abort if a disk backup job accesses the MCD database files while Network Registrar is running. The Network Registrar server will immediately be reloaded by the AIC Server Agent and Network Registrar will once again be fully operational. To avoid this problem, all backup jobs must be configured to exclude the MCD database directory, which is typically "C:\Program Files\Network Registrar\data\db." [CSCai03112]
- In nrcmd, issuing the command zone listrr static to a secondary zone fails with a confusing error message. The error message should indicate that this command cannot be issued on a secondary zone. Use export zone zonename dynamic|both when querying a secondary zone. [CSCai03114]
- The help text for the dhcp command in nrcmd mentions the scope-selection-tags property. This property can only be manipulated with the scope-selection-tag command. You cannot edit or view the master list of scope selection tags by setting or getting the value of this property. [CSCai03115]
- The help command in nrcmd does not give an error message when a nonexistent help section is specified. Specifying a nonexistent section name will cause all of the help sections to be suppressed. For example, the command help dhcp log-settings will display only the heading of the dhcp help text and will return a status of "100 OK." [CSCai03116]
- C-language DHCP extension points should not call any of the getxxx() or putxxx() APIs with a pKey parameter that is NULL. Doing so may cause the DHCP server to crash. [CSCai03120]
- IP addresses which have not yet been leased by the DHCP server have a hostname of "1,6," on the leases tab. They should be "." [CSCai03124]
- When a malformed MAC address is entered into the Reservations tab of the graphical user interface, the following error message is displayed:
- The address provided is not a valid MAC address. Please provide an address in the form xx:xx:xx where x is a hexidecimal digit.
- This error message is insufficient for two reasons:
- It does not mention that you can include "htype,len," at the beginning of the address.
- It does not explain why the specified address is malformed. [CSCai03125]
- The graphical user interface incorrectly allows you to enter hostnames with spaces in them. This will cause problems when the zone is transferred to some BIND servers. They may log errors like this:
secondary/net-171-78-star.zone: line 1675: database format error (foo.com.)
Line 1675 was:
97 IN PTR gilligan .foo.com.
- Note the whitespace between the host name, "gilligan," and its zone name, ".foo.com.." To avoid this problem, do not create host names with spaces in them. [CSCai03132]
- Entering an invalid MAC address causes nrcmd to print an overly simplistic error message. For example: scope myscope addReservation 5.5.5.5 6,5,11:22:33:44 causes the message, "316 Invalid - scope addReservation {5.5.5.5 6,5,11:22:33:44}." A more informative message would be, "316 Invalid - not enough bytes in mac address 6,5,...." [CSCai03153]
- If you use option numbers instead of names to populate the Dhcp-reply-options or Bootp-reply-options fields in nrcmd, it will return a "100 OK" status, but the options will not be saved. Use option names rather than numbers when populating these fields. [CSCai03171]
- In nrcmd, if you change the dns-host-bytes setting in a scope, you must also edit the dns-reverse-zone-name setting, in order for these two settings to be consistent. For example, if the dns-host-bytes setting is 3 and the dns-reverse-zone-name is "3.2.1.in-addr.arpa.," and you change the dns-host-bytes setting to 2, you must also change the dns-reverse-zone-name setting to "2.1.in-addr.arpa.." [CSCai03205]
- Database exports from a "Join" DHCP server do not include client ID data. Therefore, when you import leases exported from a Join server into a Network Registrar DHCP server, clients which supply client IDs will not be identified consistently. To work around this, configure the DHCP server to use only MAC addresses when representing clients, as follows:
nrcmd> dhcp enable use-macaddress-only
- Note that disabling this feature later will prevent all clients who send client IDs from renewing their leases. The DHCP server will have to assign them new addresses. [CSCai03207]
- If you delete a zone in nrcmd using zone delete zonename, you must use the save command before attempting to recreate the same zone, or data will be lost. [CSCai03244]
- In nrcmd, the command zone zone-name addRR ns server-name incorrectly allows an IP address to be specified as the server-name argument. Do not specify an IP address when adding a nameserver to a zone. [CSCai03249]
- If two scopes serve the same subnet, the nrcmd listaddr command will print unallocated addresses twice for any address not in the range of either scope. [CSCai03256]
- Network Registrar does not ensure that database files have the correct permissions after installing the servers on Windows NT. The database files need be readable and writable only by the Administrator account. If you are concerned about this, you may use the Windows NT file properties dialog to change the access permissions of the database directory, which is typically "C:\Program Files\Network Registrar\Data\Db." [CSCai03266]
- When the graphical user interface displays TTL values in the Name Servers tab or RR tab of the Edit/Add RR dialog, it gives no indication that the time is displayed in seconds. [CSCai03269]
- If continuous ranges are added to a scope, but are not added in a contiguous order, then the scope range will not be merged into one continuous range. [CSCai03270]
- The user interfaces should prevent you from deleting DHCP scope address ranges which have active leases, but they do not. Before deleting an address range from a scope, you should check the active leases on the scope to confirm that the range is not being used. [CSCai03282]
- The graphical user interface should sort the ranges displayed for a DHCP scope, but it does not. [CSCai03283]
- If a server runs out of virtual memory or swap space on NT, the server may stop functioning. [CSCai03299]
- The message "Server DNS@XYZ has configuration data that has not been saved to the database. Do you want to save it now?" may sometimes appear when exiting from the graphical user interface, even after the server has been reloaded. If this occurs, then you should exit from the user interface, restart it, and reload the server. [CSCai03316]
- Sorting DHCP leases by "state" in the graphical user interface does not work properly. [CSCai03333]
- When you modify a DHCP policy option using nrcmd, the modified option will appear at the end of the list of options for the policy (out of numerical order). Exit from nrcmd and restart it to restore the sorting. [CSCai03404]
- In nrcmd, the command policy name unsetoption option returns "100 OK" Even if option does not exist or is not set in the policy. [CSCai03405]
- The nrcmd commands policy help and zone help are invalid, but they produce a result code of "100 OK" instead of an error. Use help policy or help zone to display help for these commands. [CSCai03406]
- Removing a large zone from your configuration may cause other services to stop responding for brief periods of time. [CSCai03409]
- In nrcmd, the command lease ip-address get property returns "306 Unknown command." [CSCai03412]
- In nrcmd, if you run the command export leases -server when connected to a cluster running Network Registrar release 2.0 or to a cluster with no scopes, you get the error message "500 -0x1." The error message should indicate either that the server version does not support the command or that there are no leases to export. [CSCai03420]
- On very rare occasions, The Network Registrar servers may fail to start at boot-time on a Windows NT server. If this occurs, restart the servers manually from the Services control panel. [CSCai03424]
- nrcmd commands which require quotes around strings, e.g., ldap dirserver set username="cn=Directory Manager, o=Ace Industry, c=US," will fail if specified on the shell command line (that is, in one-shot mode). To avoid this problem, enter such commands in interactive mode. [CSCai03452]
- When using the graphical user interface from one release of Network Registrar to connect to a cluster whose servers are from a different release of Network Registrar, the lease expiration times displayed by the user interface may be three and a half minutes later than the correct times. [CSCai03454]
- When you create a primary zone in nrcmd, an NS record for the zone is created automatically, but the listRR command does not show the NS record until you exit from nrcmd and restart it. Furthermore, if you create the NS record using addRR because you believe that it has not yet been created, the zone will end up with duplicate NS records. To avoid this problem, exit and restart nrcmd after using it to create a primary zone. [CSCai03483]
- In the graphical user interface, if you set the "Authenticate-until" feature for a client using the "+[weeks]w[days]d[hours]h" syntax, one hour will be added to the specified duration each time you view the client and click on "OK." Click on "Cancel" to prevent this from occurring. [CSCai03487]
- In some cases, nrcmd returns the error "308 Unknown Parameter" when in fact the parameter is valid but the value specified for it is not. nrcmd should distinguish between these two problems in its error messages. [CSCai03488]
- The Network Registrar DHCP server uses specific DynDNS prerequisite tests to detect and avoid client hostname collisions. This functionality needs to be better described to avoid confusion. [CSCai03500]
- if nrcmd returns the error "309 Too many arguments - extra args: arg," it may be because arg is invalid or misspelled rather than "extra." [CSCai03506]
- If a scope is deactivated, the nrcmd lease list command should show all leases as "deactivated," not "available." [CSCai03509]
- If you are editing the Name Servers tab for a zone in the graphical user interface and you switch from the user interface to another application and then back, you may see "<placeholder>" instead of the appropriate text on the screen. [CSCai03518]
- On Solaris systems, attempts to uninstall the product might sometimes produce errors because the Network Registrar servers failed to shut down properly during the uninstall. After uninstalling the product, run ps -ef | grep nwreg2 to see if any of the servers are still running; if they are, use the kill command to send them a kill signal, and if they are still running afterwards, use kill -9 to force them to exit. [CSCai03548]
- Network Registrar will report an error when importing a BIND file if it encounters a WKS record in the file with an empty list of services. Although such a record is valid, it will not be imported. You can add the record to the database manually after completing the import. [CSCai03560]
- While adding a zone in the graphical user interface, the "Cancel" button becomes disabled once data is entered into the SOA tab. [CSCai03575]
- The command license get expiration in nrcmd returns an expiration date even when the license key is permanent. [CSCai03639]
- You may see the error "Removing user registry keys: The directory is not empty" when you uninstall Network Registrar from a Windows machine. You can safely ignore the error. [CSCai03652]
- Creating a scope with a badly-formed network mask (e.g., "ff.ff.ff.00") in nrcmd will fail silently. If this occurs, you must delete the scope and create it again with a correct network mask. [CSCai03655]
- When editing a DHCP policy's options in the graphical user interface, if you select an option from the "available" list and then select an option from the "active" list, you can't go back and reselect the option in the "available" list. To work around this problem, select another available option and then reselect the previous one. [CSCai03665]
- When editing a policy in the graphical user interface, if you accidentally add an option to the active list, you cannot remove it until you give it a value, click on another option, and then reselect the option you wish to remove. [CSCai03666]
- When editing a policy's options in the graphical user interface, if you select a multi-valued option, the text edit field appears active (i.e., editable) when in fact it should be grayed out and non-editable. [CSCai03671]
- In the graphical user interface, if you select a custom policy option that has been renamed or removed and then click "Edit Options," an error dialog containing the message "Unable to retrieve option data" will appear; when you acknowledge the error, the Edit Options dialog will be correctly displayed. You can ignore the error. [CSCai03672]
- If a policy is created using an illegal character (e.g. ") then the policy section of the database may become corrupted. [CSCai03696]
- Although the user interfaces may allow you to enter discontiguous netmasks (i.e., netmasks whose bits contain zeroes interspersed with ones rather than a block of ones followed by zeroes for the remainder of the mask), the use of such netmasks is deprecated, and the DHCP server may function improperly if they are used. Do not use discontiguous netmasks in DHCP scopes. [CSCai03707]
- On Solaris, a multi-homed server cannot use an FDDI interface. If you need to support an FDDI interface, you must configure DHCP to use only that interface. [CSCai03716]
- You may experience the following difficulties using the Help Find feature of the graphical user interface on Solaris: help topics called "<<Untitled>>" appear; when you select a help topic and click Display, the wrong topic appears. If this occurs, delete the file ntwkreg_exe.fts from your $HOME/.hh directory, and Help Find should then work correctly. [CSCai03717]
- If you cancel an import zone operation in the graphical user interface, the server icon will indicate it needs to be reloaded even though no changes have been made. You can reload the DNS server to turn off the reload indicator, but the server will continue to function properly even if you do not reload it. [CSCai03718]
- DHCP options which are character strings must not be empty strings; the DHCP server may crash if an empty string is entered. [CSCai03721]
- If you add a host to a zone in the graphical user interface and leave the "generate reverse mapping records" unchecked, reverse records will nevertheless be created for the IP addresses, if any, specified after the first one. [CSCai03954]
- In nrcmd, the command "zone name removeRR rrName type ipaddr" will fail silently (returning "100 OK") if the specified ipaddr is invalid. [CSCai03955]
- The list of scope selection tags displayed by the graphical user interface or nrcmd includes the special tag "<none>." This tag is used internally by Network Registrar and should never be deleted. [CSCai04100]
- If a DHCP server has only two scopes, with one secondary to the other, and you delete the primary scope, the Advanced tab of the graphical user interface's Scope Property Sheet will show that the secondary scope's primary is "<Scope Not Found>." Furthermore, if you then uncheck the "Make this scope a secondary" checkbox, the next time you display the scope's properties, you will find that it has reverted to a secondary whose primary is "<Scope Not Found>."
- You can avoid this situation by being careful not to delete any scopes that have secondary scopes. If you do get into the situation where you are unable to uncheck a scope's "Make this scope a secondary" checkbox, do the following: create a temporary, empty scope; make the temporary scope the new primary scope of the problem scope; save the data and reload the server; uncheck the "Make this scope a secondary" checkbox of the problem scope; save and reload the server again; and delete the temporary scope. [CSCdk90451]
- nrcmd does not display the primary-scope property of a scope when the scope scopename get primary-scope command is entered. The primary-scope value can be viewed with the scope scopename show command. [CSCdk90459]
- For the Web user interface to function properly, you must stop and restart your Web server after installing or upgrading Network Registrar. [CSCdk91356]
- The report command in nrcmd may not report correct information when safe failover is in use and changes have been made recently to the configuration on the main or backup server. To ensure that generated reports are correct after making changes to the configuration of the main or backup server, reload both of them before generating a report. [CSCdk92523]
- When safe failover is enabled, the number of leases allocated to the backup server based on the backup-percentage property may be higher than specified when a symmetric configuration is in use. This will not effect the operation of leases the backup server currently has allocated. [CSCdk93046]
- The Edit Options window of the graphical user interface rounds down displayed lease times to integral multiples of minutes. For example, a lease time of 96 seconds is displayed as 60 seconds. The correct data is saved to the server. [CSCdk93351]
- If nrcmd cannot connect to the mail host specified in the mail-host parameter of the lease-notification command, it will correctly display an error about the failure to connect, but subsequent commands will behave incorrectly. To avoid this problem, ensure that you specify the correct mail host and that it is currently accepting SMTP (mail delivery) connections. If you encounter the problem, exit and restart nrcmd. [CSCdk93927]
- If you use nrcmd to create a new zone and then immediately attempt to add a host to it, nrcmd may print an error indicating that the zone was not found. To avoid this problem, reload the DNS server after creating a new zone in nrcmd before attempting to add hosts to it. [CSCdm00608]
- If the cluster you specify with the -C argument to nrcmd is listed in the middle of the cluster list in the configuration file for the nrcmd export addresses, report or lease-notification command, the commands will fail and may leave a stale lock. To avoid this problem, either ensure that the cluster you specify with -C is listed first in the cluster list in the configuration file, or use localhost as the logon cluster (if you are connecting to servers on the local host). [CSCdm01288]
- In rare cases, the Network Registrar server agent on Solaris systems may not restart the DHCP server when it is killed or crashes. If the DHCP server fails to restart, stop and restart the server agent by running /etc/init.d/aicservagt stop and then /etc/init.d/aicservagt start as root. [CSCdm07436]
- The graphical user interface and nrcmd both allow the user to create empty DHCP options with length zero (i.e., a DHCP option with a name but an empty value). Zero-length options are not permitted by the DHCP protocol. While the DHCP server will ignore such options, both the graphical user interface and nrcmd should display an error rather than allowing them to be created. [CSCdm08564]
- In nrcmd, IP addresses with leading zeroes are incorrectly reported by the export addresses command. For example, in the ip_address column, lease 1.1.1.1 is incorrectly reported as 0x1010101 instead of 0x01010101. [CSCdm09165]
- In rare cases on Solaris systems, the DHCP server will wait several minutes after startup before it starts providing DHCP services. [CSCdm11677]
- A zone's serial number cannot be modified using nrcmd. The command "zone zone-name set serial=number" appears to work, but it has no affect on the serial number shown in the SOA record for the zone. [CSCdm15489]
- When the "health" of the DHCP server is low, e.g., when there are no scopes defined, some nrcmd commands, including getRelatedServers, setPartnerDown, and export leases, will return:
312 Unexpected response from server - (0x1)
- The message should state more clearly that the requested command could not be executed because of the current state of the server; however, there is no real loss in server functionality. To eliminate the spurious errors, resolve the problem that is causing the low server health. [CSCdm16250]
- If you create a custom DHCP option to override a default value for that option, and then you remove the custom option, its default value is not restored. To avoid this problem, if you wish to restore the default value to an option, you should explicitly set its value to the default rather than removing the custom option. [CSCdm16632]
- When DHCP safe failover is in use and you deactivate a lease with either the graphical user interface or nrcmd, it is deactivated only in the cluster to which you are connected. If you wish the lease to be deactivated on both the main and backup servers, you must connect to and deactivate the lease in each server's cluster. [CSCdm16657]
- In the graphical user interface, if the state of the "Enable Dynamic DNS Updates" checkbox is changed more than once in a single session, the user interface will report that it was unable to save the data to the database. The correct data will, however, be saved. [CSCdm19865]
- When using DHCP safe failover, if you reverse the roles of the main and backup servers within nrcmd, you must exit and restart nrcmd after reloading the DHCP servers with the new failover roles, or nrcmd will give incorrect lease state information for the main and backup servers. [CSCdm23176]
- When using failover, a lease can sometimes get "stuck" in the pending-available state, where one server (typically the main) is trying to get the backup to acknowledge that this lease is now available, and the backup is NAKing the binding update from the main since the backup believes that is has more recent binding information (i.e., the NAK code is "outdate binding information").
- When this occurs, it is usually associated with a backup server which is overloaded and cannot keep up with the volume of updates being offered to it by the main server. The backup server needs to get considerably behind in its processing of failover updates for this to occur.
- When a client interacts with the main server, this problem will typically resolve itself. However, when a client is no longer present on the network and has allowed its lease to expire or has explicitly released its lease, and if the backup server was very busy when this happened, then sometimes a lease will get stuck in the pending-available state.
- You can recover a lease stuck in the pending-available state by forcing it available using either the graphical user interface or nrcmd. You should force it available on the server that believes that it has more recent binding information and is NAKing the other server, which will almost always be the backup server. [CSCdm23864]
- The Network Registrar 3.0 graphical user interface has a different format for its "GUI cache," which it uses to display DNS resource records. Because of this, you may have problems using both the Network Registrar 2.5 and 3.0 graphical user interfaces under the following circumstances:
- The server version is 2.5.
- You connect with a 3.0 graphical user interface (allowed for backward compatibility).
- You change one or more Hosts or resource records on the 2.5 server.
- You subsequently connect with a 2.5 graphical user interface.
- If you encounter this situation, run the "Rebuild Resource Record Indexes" command from the Advanced tab of DNS Server Properties dialog in the user interface (ignore the warning message). It is then safe to add, delete, or modify hosts. [CSCdm24250]
- Certain errors in safe failover configuration could cause lost failover capability. Such errors include a mismatched MCLT, a disagreement between the main and backup servers on their main or backup designations, and a mismatch in the Address ranges for a scope on the main and backup servers.
- Currently, Network Registrar does not notify the administrator of such configuration errors. We will soon be releasing a utility to compare the failover configurations on two servers and report on any discrepancies. A notice will be sent to all registered customers at that time. [CSCdm24264] .








Posted: Thu Nov 18 12:34:15 PST 1999
Copyright 1989-1999©Cisco Systems Inc.