This appendix describes the changes made to previously released versions of the Catalyst Switch software.
- When you set the Internet Protocol (IP) address using the ifconfig command, you must explicitly set the netmask and broadcast addresses to the desired values. If this is not done, network management applications do not recognize those values.
 | Caution If the Catalyst 1200 switch is reset, the netmask changes to the default netmask for the assigned class address. You must then use the ifconfig command to reset the netmask and broadcast addresses to the desired values. |
- Giant packet counters are not incremented in the display mac command. The display mac command displays parameters for the Media Access Control (MAC) layer for each port of the switch.
- When you set the IP address using the ifconfig command, you must explicitly set the netmask and broadcast addresses to the desired values. If this is not done, network management applications do not recognize those values.
| Caution If the Catalyst 1200 switch is reset, the netmask changes to the default netmask for the assigned class address. You must then use the ifconfig command to reset the netmask and broadcast addresses to the desired values. |
- Giant packet counters are not incremented in the display mac command. The display mac command displays parameters for the MAC layer for each port of the switch.
- Data frames on FDDI networks less than 25 bytes long may be discarded. This may cause network connection loss on FDDI-attached stations using protocols that use less than 25 byte frames, such as NETBIOS and NetBEUI.
Note This does not affect Ethernet-to-Ethernet through FDDI connections.
- If the netmask is not set using the ifconfig command, the Catalyst 1200 switch is unable to communicate through the IP network using, for example, the ping, download, and other commands.
- Enabling a disabled port when spanning tree is active causes that port to change to a permanent blocked state. To eliminate the blocked state after enabling the port, disable then enable spanning tree.
The following issues were resolved during this revision of the Catalyst Series software:
- Subnet mask changes set using commands other than ifconfig would not allow communication through the IP network.
- DMP and NMP Versions 2.3--If a port is disabled, the show port command displays the status of the disabled port as ok instead of the correct status of disabled.
Workaround:
The show port command displays the actual hardware status of all the ports in the switch. The disable port command available through the admin. interface does not change the hardware status of the ports and all the ports are still shown as ok. However, to get the software status of the ports, use the show bridge command.
- DMP and NMP Versions 2.3--When an FDDI or Ethernet port is physically disconnected or when the disconnect command is used to disconnect a switch port from the ring, spanning tree still considers the port to be in forwarding state. Therefore the show bridge command displays the port to be in forwarding state.
Workaround:
The disconnect command temporarily disconnects a switch from the FDDI ring. Therefore, this information is not passed to the spanning tree and the show bridge command shows that the port is still in forwarding state. Spanning tree is informed that the port is being disabled only when the user uses the disable port command.
- DMP and NMP Versions 2.3--When the switch receives fragmented ping packets destined to itself on an FDDI interface, the terminal admin. interface session stops. However, the switch still responds to any ping and Telnet requests.
Workaround:
This problem can be eliminated by not sending fragmented ping packets to the switch on the FDDI ring. Reset the switch to clear the terminal stop situation.
- DMP and NMP Versions 2.3--When the IP address of the switch is changed from one class to another, you may not be able to delete the static routes defined for the previous class.
Workaround:
The only way to delete those entries is to reboot the switch.
- DMP and NMP Versions 2.3--When the switch Content Addressable Memory (CAM) table has many MAC entries and the switch is very busy bridging, the show cam command displays the CAM entries on the screen at a slower rate. This is because the admin. interface receives a lower priority and the screen display can become slower as a result.
Workaround:
If your CAM table has many entries, you should use the cam lookup command to track any of the MAC addresses in the CAM table. You should not use the show cam command under very busy switch conditions.
- DMP and NMP Versions 2.3--This release of the switch software allows users to control the display screen scrolling by using the set length command. However, if the screen length is set to a finite value, the switch CAM table is full, and the user issues the show cam command the switch might truncate the display on the screen because of its limited buffer allocation for the screen display.
Workaround:
You should use the cam lookup command to look up for any MAC entries in the cam table instead of using the show cam command which dumps all the MAC entries on the screen. If you use the show cam command, you should set the length to zero using the set length command.
- Giant Packets--When any Ethernet or FDDI port receives a giant packet (greater than 1,714 bytes), the switch resets. To eliminate this problem perform, set the encheck parameter to on.
Note The largest valid frame is a 1514-byte + 4 byte cyclic redundancy check (CRC). If your network does not use frames larger than 1714 bytes, you do not need to modify this parameter.
By default, the encheck parameter is turned off. To turn the encheck parameter on, perform the following steps:
Note Be sure that you set up a terminal session into the switch by connecting a terminal to the EIA/TIA-232 port of the switch. Do not use a Telnet session to the switch to modify this parameter.
Step 1 At the Catalyst user prompt, change to privileged mode by entering enable.
Step 2 At the prompt, enter show optimization and press Return.
The system displays the following:
Console> (enable) show optimization
encheck = off
enrx = 0x20
entx = 0x10
fddirx = 0x100
fdditx = 0x10
Console> (enable)
Step 3 Check the encheck parameter. If the parameter is set to on, the system handles giant packets efficiently without resetting the switch. If the encheck parameter is set to off, proceed to the next step.
Step 4 At the prompt, enter set optimization encheck on and press Return to change the encheck parameter. Following is an example:
Console> (enable) set optimization ?
Usage: set optimization encheck <on|off>
set optimization enrx <value>
set optimization entx <value>
set optimization fddirx <value>
set optimization fdditx <value>
set optimization default
Console> (enable) set optimization encheck on
The system responds with the following confirmation message:
This command will reset and restart the system for Ethernet re-configuration.
Do you want to continue (y/n) [n]?
Step 5 Enter y to continue.
This enables the encheck parameter and the switch reboots itself to make the change in the system configuration. This change prevents the switch from resetting when it receives a giant packet in the future.
The following issues were resolved during this revision of the Catalyst 1200 switch software:
- Under certain circumstances the Catalyst switch would report a large number of giant packets on a port causing that port to reinitialize frequently. This problem occurred due to a race condition between the DMP RISC processor and the SONIC processor while processing the received packets on that port.
- When a Catalyst switch received FDDI subnetwork access protocol (SNAP) frames other than IP, Internet Protocol Exchange (IPX), or AppleTalk, the switch was converting them into an incorrect format.
- When using the set cam command to enter static entries into the CAM table, any MAC address with the 00 as the last byte is treated as an invalid address.
- When using the set route command to add routing table entries with a metric value 1 the Catalyst switch defaults to a metric value 0 although the default is 1.
Workaround:
When entering a metric value 1, always enter the desired metric value.
- If a port is disabled, the show port command displays the status of the disabled port as ok instead of the correct status of disabled.
Workaround:
The show port command displays the actual hardware status of all the ports in the switch. The disable port command available through the admin. interface does not change the hardware status of the ports, and all the ports are still shown as ok. However, to get the software status of the ports, use the show bridge command.
- When an FDDI or Ethernet port is physically disconnected, or when the disconnect command is used to disconnect a switch port from the ring, the spanning tree still considers the port to be in the forwarding state. Therefore the show bridge command displays the port as in the forwarding state.
Workaround:
The disconnect command temporarily disconnects a switch from the FDDI ring. Therefore, this information is not passed to the spanning tree and the show bridge command still shows that the port is in forwarding state. Spanning tree is informed that the port is being disabled only when you use the disable port command.
- When the switch receives fragmented ping packets destined to itself on an FDDI interface, the terminal admin. interface session stops. However, the switch still responds to any ping and Telnet requests.
Workaround:
This problem can be eliminated by not sending fragmented ping packets to the switch on the FDDI ring. Reset the switch must to clear the terminal stop situation.
- When the IP address of the switch is changed from one class to another, you may not be able to delete the static routes defined for the previous class.
Workaround:
The only way to delete those entries is to reboot the switch.
- When the switch CAM table has many MAC entries and the switch is very busy bridging, the show cam command displays the CAM entries on the screen at a slower rate. This is because the admin. interface receives a lower priority and the screen display can become slower as a result.
Workaround:
If your CAM table has many entries, you should use the cam lookup command to track any of the MAC addresses in the CAM table. You should not use the show cam command under very busy switch conditions.
- This release of the switch software allows users to control the display screen scrolling by using the set length command. However, if the screen length is set to a finite value, the switch CAM table is full, and the user issues the command show cam, the switch might truncate the display on the screen because of its limited buffer allocation for the screen display.
Workaround:
You should use the cam lookup command to look up any MAC entries in the CAM table instead of using the show cam command which dumps all the MAC entries on the screen. If you choose to use the show cam command, you should set the length to zero using the set length command.
- Giant Packets--When any Ethernet port receives a giant packet (greater than 1,714 bytes), the switch resets. To eliminate this problem, set the encheck parameter to on.
Note The largest valid frame is a 1,514-byte + 4-byte cyclic redundancy check (CRC). If your network does not use frames larger than 1,714 bytes, you do not need to modify this parameter.
By default the encheck parameter is turned off. To turn the encheck parameter on, perform the following steps:
Note Be sure that you set up a terminal session into the switch by connecting a terminal to the EIA/TIA-232 port of the switch. Do not use a Telnet session to the switch to modify this parameter.
Step 1 At the Catalyst user prompt, change to privileged mode by entering enable.
Step 2 At the prompt, enter show optimization and press Return.
The system displays the following:
Console> (enable) show optimization
encheck = off
enrx = 0x20
entx = 0x10
fddirx = 0x80
fdditx = 0x80
Step 3 Check the encheck parameter. If the parameter is set to on, the system handles giant packets efficiently without resetting the switch. If the encheck parameter is set to off, proceed to the next step.
Step 4 At the prompt, enter set optimization encheck on and press Return to change the encheck parameter. Following is an example:
Console> (enable) set optimization ?
Usage: set optimization encheck <on|off>
set optimization enrx <value>
set optimization entx <value>
set optimization fddirx <value>
set optimization fdditx <value>
set optimization default
Console> (enable) set optimization encheck on
The system responds with the following confirmation message:
This command will reset and restart the system for Ethernet re-configuration.
Do you want to continue (y/n) [n]?
Step 5 Enter y to continue.
This enables the encheck parameter and the switch reboots itself to make the change in the system configuration. This change prevents the switch from resetting when it receives a giant packet in the future.
The following modifications have been made since the previous release:
- When the set optimization command variable encheck was set to on, false giant packets were being reported, causing poor switch performance. With this version of Flash code, if giant packets are received, the encheck parameter is automatically set to on, eliminating the problem.
- The set tlmin command has been added to set the time required for Physical Layer Protocol (PHY) hardware to transmit a given line state before advancing to the next physical connection management (PCM) state at the station management (SMT) level. The port_num should be 1 or 2, and the hexvalue should be between 0 and 0xffff. The tl_min setting is stored in the TL_MIN register (also known as the LS_MAX register) as part of the SMT management information base (MIB) structure in nonvolatile memory, and is used for initializing the PHY hardware setting each time the switch is rebooted.
Console> (enable) set tlmin ?
Usage: set tlmin <port_num> <hexvalue>
(hexvalue is in 2's complement)
Console> (enable) set tlmin 1 fdfd
Port 1 tlmin set to 0xfdfd.
Console> (enable)
- Some FDDI SNAP frames with the proprietary organizationally unique identifier (OUI) field were not being translated correctly. This version of Flash code corrects this problem.
- Performance enhancements have been made to boost the Catalyst switch performance significantly while handling broadcast frames at wire speed.
- All non-ODI (open data-link interface) IPX drivers and some other LAN drivers that were sensitive to 802.3 frame length were experiencing connectivity problems with the Catalyst switch. This version of Flash code corrects this problem.
- Static CAM entries were not allowed with the last byte set to 00. This version of Flash code corrects this problem.
- Static CAM entries were not being handled correctly. This version of Flash code corrects this problem.
The following caveats have been identified and workarounds provided:
- The SUM port (sc0) IP address scheme has changed with Catalyst Switch Software Release 3.1. The switch IP address previously assigned to the SUM port will now, by default, be assigned to route group 1. Route group 1 will include all ports of the switch including the FDDI port. The SUM port (sc0) will be set to IP address 0.0.0.0. The SUM port will not be part of the default route group 1, and will not be able to communicate with the switch.
Workaround:
The SUM port must be assigned a unique IP address to allow communication with the switch.
- Aborting serial download will cause an NMP exception error when you enter the show interface command at the admin port.
Workaround:
Do not abort a serial download. Let the serial download complete. If you do abort a serial download, reset the switch and reinitiate the serial download.
- The Catalyst switch may continue to learn routes using the router interface protocol (RIP), but cannot route packets on interfaces that are administratively set to down using the set interface port# down command.
Note The switch can still bridge data on interfaces administratively configured as down.
Workaround:
To stop learning RIP updates on an interface, disable the port using the set port port# disable command.
- The show bridge command displays an FDDI port to be forwarding even though it is disconnected physically or disconnected using the disconnect fddi command. The erroneous status is displayed because the bridge still considers the FDDI port to be in the forwarding state. This is because the disconnect fddi command only temporarily disconnects the switch from the FDDI ring.
Workaround:
The spanning tree is informed that the FDDI port is disabled only when you use the disable port 1 command.
- Use the connect fddi and disconnect fddi commands carefully while the switch is bridging and routing traffic. If you use the connect fddi command, the connect state LED may change to orange, indicating a minor alarm.
Workaround:
Reissue the disconnect fddi command and then reconnect the port using the connect fddi command until the connect state LED changes to GREEN, indicating no alarms.
- Terminal sessions may freeze if the switch receives fragmented Internet Control Message Protocol (ICMP) packets destined to itself on the FDDI ring. The switch will still respond to any ping and Telnet requests destined to it, and bridge or route any packets.
Workaround:
Reset the switch to eliminate the frozen terminal session.
- The switch does not automatically delete the default route if the routing group configuration is changed using the set routegroup command. For example:
- The Resource Manager, Remote Login, and Token Ring Main screen icons are not supported by the NetScout Manager software and should not be used.
- If the screen appears frozen or unresponsive, this may be due to a series of unintentional actions by the user. For example, if both the Main screen and the Add Agent submenu screen are opened at the same time and you click on an inappropriate button (for example, cancel) the Main screen will move to the front, obscuring the Add Agent submenu. However, the Add Agent submenu screen will remain active in the background, making the Main screen inaccessible.
You must complete all submenu actions items before you can activate other Main menu items.
Workaround:
Reduce the frozen screen(s) to icons to find the active submenu screen hidden behind them. You must close the active submenu screen before you can make another Main menu selection.
- From the Main screen menu, all other submenus require only one click of the mouse to activate. If you click twice on a selection, two instances of the same selection will be activated, causing multiple instances of the same function with overlapping screens. Multiple instances of the same submenu are a drain on the Client and Agent resources, and should be closed when not in use.
Note The submenus of the Main Screen are independent and can be activated from the Main menu screen with a single click of the mouse button.
- The Client may run out of resources because you have too many processes open at the same time for the amount of RAM and swap space configured in the Client workstation. The agent may also run out of resources due to limited memory availability.
The Catalyst Switch will attempt to install all possible resources using the available memory. If the switch cannot successfully install all of the enabled groups for the chosen domain, it will install as many as possible and display the following error messages:
no resources...
could not install
- If the system appears slow to respond to a command, the possible causes are:
- The network is very busy, causing the Client to retransmit the messages to the agent for proper responses.
- The screens that generate graphic displays, for example, pie charts and bar graphs may be compiling data needed to generate the graphic display. Initial screen updates may take up to a minute or longer while the system command is sent to the probe, and an initial period of 15 seconds is required for the agent to wait for updated information before returning a screen update.
- If you use the Config option from the Main screen to configure RMON, the system does not respond when the configuration has been completed.
Note RMON configuration may take up to two minutes to complete.
Workaround:
Use the Domain Management menu to confirm the completion of the RMON configuration, or use the Domain Manager submenu from the Main menu screen to install logical agents.
- While communicating with an agent, an error message similar to the following may appear:
-- Error accessing agent < agent name >
While retrieving Domains
Error: cannot communicate with agent
This error message does not indicate that the Catalyst switch is down. This error message may be caused by one or more of the following conditions:
- You should not modify the sample configuration files with the file extension .cfg, stored in the usr/nohome/samples directory. These files should be copied to the user's working directory and modified to create user-specific configuration files.
| Caution If the sample configuration files become corrupted, you must reinstall the RMON to create new sample configuration files. |
- The two terms segment and domain have different meanings in the NetScout Manager context:
- Domain means an aggregate of protocol filters supported by the NetScout Manager.
- Segment means the network itself, or the subnet, which is fully connected.
Note For a subnet of nodes to form a segment, the nodes do not have to be separated by bridges, routers, or gateways.
- To specify segments, a drawing of all nodes including bridges, routers, and gateways, should be made. This facilitates interpretation of the collected data and the comparison of the various segments in terms of various data collection activities and remote monitoring of the MIBs.
- When creating logical agents using the Edit window, any of the following entries will generate an SNMP error code 3:
Note The correct interface number range is 3 through 10.
