Release Notes for the Catalyst 3750, 3560, and 2970 Switches, Cisco IOS Release 12.2(20)SE (Revised Aug. 18, 2004)

Hardware Supported

Table 1 lists the hardware supported on Cisco IOS Release 12.2SE.

 

1.SFP = small form-factor pluggable 2.PoE = Power over Ethernet 3.CWDM = coarse wavelength-division multiplexer

Software Compatibility

For hardware requirements, operating system, and browser recommendations for running the Cluster Management Suite (CMS), refer to the “Getting Started with CMS” chapter in the software configuration guide.

Windows

This release uses a CMS plug-in to run CMS. You can download the latest CMS plug-in for Windows from this URL:

http://www.cisco.com/pcgi-bin/Support/ClusterMgmtSuite/cms_plugin_redirect.cgi?platform=windows&version=1.1

Solaris

This release uses a CMS plug-in that replaces the Java plug-in. You can download the latest CMS plug-in for Solaris from this URL:

http://www.cisco.com/pcgi-bin/Support/ClusterMgmtSuite/cms_plugin_redirect.cgi?platform=solaris&version=1.1

Cluster Compatibility

This section describes how to choose command and standby command switches when a cluster consists of a mixture of Catalyst switches. When creating a switch cluster or adding a switch to a cluster, follow these guidelines:

• When you create a switch cluster, Cisco recommends configuring the highest-end switch in your cluster as the command switch. Table 2 lists the cluster capabilities and Cisco IOS releases for the switches. The switches are listed from the highest to lowest end.
• If you are managing the cluster through CMS, the switch that has the latest software should be the command switch, unless your command switch is running Cisco IOS Release 12.1(19)EA1 or later.
• The standby command switch must be the same type as the command switch. For example, if the command switch is a Catalyst 3750 switch, all standby command switches must be Catalyst 3750 switches.

 

4.Catalyst 2900 XL (4-MB) switches appear in the front-panel and topology views of the CMS. However, CMS does not support the configuration or the monitoring of these switches.

CMS is not forward-compatible on command switches running Cisco Release IOS 12.1(14)EA1 and earlier. This means that if a member switch is running a release that is earlier than the release running on the command switch, the new features are not available on the member switch. If the member switch is a new device running a release that is later than the release on the command switch, the command switch cannot recognize the member switch, and the Front Panel view displays it as an unknown device. You cannot configure any parameters or generate a report through CMS for that member; instead, you must launch the Device Manager application to configure and to obtain reports for that member.

If you have a cluster with switches that are running different versions of Cisco IOS software, features added on the latest release might not be reflected on switches running the older releases. For example, if you start CMS on a Catalyst 2900 XL switch running Cisco IOS Release 11.2(8)SA6, the windows and functionality can be different from a switch running Cisco IOS Release 12.0(5)WC(1) or later.

Some early Cisco IOS releases do not support clustering.

For more information about clustering and CMS, refer to the software configuration guide.

Finding the Software Version and Feature Set

The Cisco IOS image is stored as a . bin file in a directory that is named with the Cisco IOS release. A subdirectory contains the files needed for web management. The image is stored on the system board flash device (flash:).

You can use the show version privileged EXEC command to see the software version that is running on your switch. The second line of the display shows the version.

---

Note For Catalyst 3750 and 3560 switches, although the show version output always shows the software image running on the switch, the model name shown at the end of this display is the factory configuration (standard multilayer image [SMI] or enhanced multilayer image [EMI]) and does not change if you upgrade the software image.

You also can use the dir filesystem : privileged EXEC command to see the directory names of other software images that you might have stored in flash memory.

Deciding Which Files to Use

The upgrade procedures in these release notes describe how to perform the upgrade by using a combined tar file. This file contains both the Cisco IOS image file and the files needed for CMS. You must use the combined tar file to upgrade the switch through CMS. To upgrade the switch through the command-line interface (CLI), use the tar file and the archive download-sw privileged EXEC command.

Table 3 lists the filenames for this software release.

 

5.SSH = Secure Shell

Upgrading a Switch by Using CMS

You can upgrade switch software by using CMS. From the feature bar, choose Administration > Software Upgrade. For detailed instructions, click Help.

---

Note When using HTTP to upgrade member switches, the command switch must be running either Cisco IOS 12.1(20)EA2 or Cisco IOS 12.2(20)SE or later. The cluster members that are upgraded must be running Cisco IOS 12.2(20)SE or later.

Upgrading a Switch by Using the CLI

This procedure is for copying the combined tar file to the switch. You copy the file to the switch from a TFTP server and extract the files. You can download an image file and replace or keep the current image.

To download software, follow these steps:

---

Step 1 Use Table 3 to identify the file that you want to download.

Step 2 Download the software image file.

• If you have a SmartNet support contract, go to this URL, and log in to download the appropriate files:

http://www.cisco.com/kobayashi/sw-center/sw-lan.shtml

• If you do not have a SmartNet contract, go to this URL, and follow the instructions to register on Cisco.com and download the appropriate files:

http://www.cisco.com/public/sw-center/sw-lan.shtml

To download the image for a Catalyst 2970 switch, click Catalyst 2970 software. To obtain authorization and to download the cryptographic software files, click Catalyst 2970 3DES Cryptographic Software.

To download the EMI or SMI files for a Catalyst 3560 switch, click Catalyst 3560 software. To obtain authorization and to download the cryptographic software files, click Catalyst 3560 3DES Cryptographic Software.

To download the EMI or SMI files for a Catalyst 3750 switch, click Catalyst 3750 software. To obtain authorization and to download the cryptographic software files, click Catalyst 3750 3DES Cryptographic Software.

---

Step 3 Copy the image to the appropriate TFTP directory on the workstation, and make sure that the TFTP server is properly configured.

For more information, refer to Appendix B in the software configuration guide for this release.

Step 4 Log into the switch through the console port or a Telnet session.

Step 5 (Optional) Ensure that you have IP connectivity to the TFTP server by entering this privileged EXEC command:

For more information about assigning an IP address and default gateway to the switch, refer to the software configuration guide for this release.

Step 6 Download the image file from the TFTP server to the switch. If you are installing the same version of software that is currently on the switch, overwrite the current image by entering this privileged EXEC command:

The /overwrite option overwrites the software image in flash memory with the downloaded one.

The /reload option reloads the system after downloading the image unless the configuration has been changed and not saved.

For // location, specify the IP address of the TFTP server.

For / directory / image-name .tar, specify the directory (optional) and the image to download. Directory and image names are case sensitive.

This example shows how to download an image from a TFTP server at 198.30.20.19 and to overwrite the image on the switch:

You also can download the image file from the TFTP server to the switch and keep the current image by replacing the /overwrite option with the /leave-old-sw option.

Recovering from a Software Failure

For detailed recovery procedures, refer to the “Troubleshooting” chapter in the software configuration guide for this release.

---

New Hardware Features

For a list of all supported hardware, see the “Hardware Supported” section.

New Software Features

For a list of default settings after initial switch configuration, including default settings that are changed after the software migration from Cisco IOS Release 12.1 to 12.2, refer to Table 1-1 in Chapter 1 of the software configuration guide.

For a list of commands that have the same function in Cisco IOS Release 12.1(19)EA1 or earlier but different syntax after the 12.2 migration, refer to Table 1-2 in Chapter 1 of the command reference.

New Features for the Catalyst 3750, 3560, and 2970 Switches

This release contains these new Catalyst 3750, 3560, and 2970 switch features or enhancements (available in all software images):

• Automatic QoS (auto-QoS) when traffic is received on trusted ports and when your switch is connected to a device running the Cisco SoftPhone application
• DHCP snooping binding database configuration
• 802.1x accounting tracks network usage and supports these 802.1x port access entities (PAEs) MIBs: IEEE8021-PAE-MIB and CISCO-PAE-MIB
• Flex Links provides basic link redundancy for Layer 2 interfaces as an alternative to spanning tree
• New Cisco-default Smartports global macro simplifies switch configuration across the network
• CISCO-UDLDP-MIB support for CiscoWorks
• Support in CMS for upgrading your switch by using HTTP (no TFTP server is necessary)
• SFP diagnostic management interface monitors the physical or the operational status of an SFP module

New EMI Features for the Catalyst 3750 and 3560 Switches

This release contains these new Catalyst 3750 and 3560 switch features or enhancements (available only in the EMI):

• Dynamic Address Resolution Protocol (ARP) inspection prevents malicious attacks on the switch by not relaying invalid ARP requests and responses to other ports in the same VLAN.
• IP source guard restricts traffic on nonrouted interfaces by filtering traffic based on the DHCP snooping database and IP source bindings.
• Private VLAN extends the features of normal VLANs, provides a more controlled IP address allocation, and allows Layer 2 port isolation from other ports on the switch.

New Stacking Features for the Catalyst 3750 Switch

This release contains these new Catalyst 3750 switch stacking features or enhancements (available in all software images):

• Switch stack offline configuration provisions a new member for a switch stack. You configure, in advance, the stack member number, the switch type, and the interfaces associated with a switch that is not currently part of the stack.
• Stack-ring activity statistics displays the number of frames per stack member that are sent to the stack ring.

Table 4 Catalyst 3750, 3560, and 2970 Switch Features and the Minimum Cisco IOS Release Required

Feature	Minimum Cisco IOS Release Required	Catalyst Switch Support
802.1x accounting and MIBs (IEEE8021-PAE-MIB and CISCO-PAE-MIB)	12.2(20)SE	3750, 3560, 2970
Dynamic ARP inspection (EMI only)	12.2(20)SE	3750, 3560
Flex Links	12.2(20)SE	3750, 3560, 2970
HTTP upgrade (CMS only)	12.2(20)SE	3750, 3560, 2970
IP source guard (EMI only)	12.2(20)SE	3750, 3560
Private VLAN (EMI only)	12.2(20)SE	3750, 3560
SFP diagnostic management interface	12.2(20)SE	3750, 3560, 2970
Switch stack offline configuration	12.2(20)SE	3750
Stack-ring activity statistics	12.2(20)SE	3750
Smartports macros	12.2(18)SE	3750, 3560, 2970

Cisco IOS Limitations and Restrictions

Unless otherwise noted, these limitations apply to the Catalyst 3750, 3560, and 2970 switches:

• “Configuration” section
• “Ethernet” section
• “Fallback Bridging” section
• “HSRP” section
• “IP” section
• “IP Telephony” section
• “MAC Addressing” section
• “Multicasting” section
• “QoS” section
• “Routing” section
• “SPAN and RSPAN” section
• “Stacking (Catalyst 3750 switch stack only)” section
• “Trunking” section
• “VLAN” section

Configuration

These are the configuration limitations:

• A static IP address might be removed when the previously acquired DHCP IP address lease expires.

This problem occurs under these conditions:

– When the switch is booted without a configuration (no config.text file in flash memory).

– When the switch is connected to a DHCP server that is configured to give an address to it (the dynamic IP address is assigned to VLAN 1).

– When an IP address is configured on VLAN 1 before the dynamic address lease assigned to VLAN 1 expires.

The workaround is to reconfigure the static IP address. (CSCea71176 and CSCdz11708)

• Certain combinations of features and switches create conflicts with the port security feature. In Table 5 , No means that port security cannot be enabled on a port on the referenced switch if the referenced feature is also running on the same port. Yes means that both port security and the referenced feature can be enabled on the same port on a switch at the same time. A dash means not applicable.

6.DTP = Dynamic Trunking Protocol 7.A port configured with the switchport mode dynamic interface configuration command. 8.A VLAN Query Protocol (VQP) port configured with the switchport access vlan dynamic interface configuration command. 9.The switch must be running the enhanced software image (EI). 10.You must set the maximum allowed secure addresses on the port to two plus the maximum number of secure addresses allowed on the access VLAN. 11.The switch must be running the enhanced multilayer image (EMI). 12.ARP = Address Resolution Protocol

• (Catalyst 3750 or 3560 switches) If a 10/100BASE-T port configured for forced 100 Mbps full-duplex mode is connected to a link partner that is auto-negotiating, the link partner comes up in 100 Mbps full-duplex mode. However, if the same link partner is connected to a Gigabit port configured for forced 100 Mbps full-duplex mode, the link comes up in 100 Mbps half-duplex mode. The reason for this inconsistent behavior is that the 10/100 port auto-negotiates even if both speed and duplex mode are fixed, whereas the Gigabit port does not. This is a hardware limitation. No workaround is necessary. (CSCeb14068)
• (Catalyst 3750 or 3560 switches) When the show interface privileged EXEC is entered on a port that is running 802.1Q, inconsistent statistics from ports running 802.1Q might be reported. The workaround is to upgrade to Cisco IOS Release 12.1(20)EA1. (CSCec35100)
• (Catalyst 3750 or 3560 switches) When you change a port from a nonrouted port to a routed port or the reverse, the applied auto-QoS setting is not changed or updated when you verify it by using the show running interface or show mls qos interface user EXEC commands. These are the workarounds:

1. Disable auto-QoS on the interface.

2. Change the routed port to a nonrouted port or the reverse.

3. Re-enable auto-QoS on the interface. (CSCec44169)

• The DHCP snooping binding database is not written to flash or a remote file in any of these situations:

– (Catalyst 3750 switch) When the Network Time Protocol (NTP) is configured, but the NTP clock is not synchronized. You can check the clock status by entering the show NTP status privileged EXEC command and verifying that the network connection to the NTP server and peer work correctly.

– (Catalyst 3750, 3560, or 2970 switches) The DHCP snooping database file is manually removed from the file system. After enabling the DHCP snooping database by configuring a database URL, a database file is created. If the file is removed manually from the file system, the DHCP snooping database does not create another database file. You need to disable the DHCP snooping database and enable it again to create the database file.

– (Catalyst 3750, 3560, or 2970 switches) The URL for the configured DHCP snooping database was replaced because the original URL is not accessible. The new URL might not take effect after the timeout of the old URL.

No workaround is necessary; these are the designed behaviors. (CSCed50819)

• (Catalyst 3750 or 3560 switches) When dynamic ARP inspection is enabled on a switch or switch stack, ARP and RARP packets greater than 2016 bytes are dropped by the switch or switch stack. This is a hardware limitation.

However, when dynamic ARP inspection is not enabled and jumbo MTU is configured, ARP and RARP packets are correctly bridged in hardware. (CSCed79734)

• CSCed88768

When you are upgrading a switch from a Cisco IOS Release 12.1 image to Cisco IOS Release 12.2(20)SE and you are using a type-7 encrypted password after the upgrade, you no longer need to configure the switch with the same password that you were using before the upgrade.

You might still need to re-configure the switch with the same password in these situations:

– Upgrading from a Cisco IOS 12.1 image to Cisco IOS Release 12.2(18)SE

– Upgrading from Cisco IOS Release 12.2(18)SE to Cisco IOS Release 12.2(20)SE

Ethernet

These are the Ethernet limitations:

• Subnetwork Access Protocol (SNAP) encapsulated IP packets are dropped without an error message being reported at the interface. The switch does not support SNAP-encapsulated IP packets. There is no workaround. (CSCdz89142)
• (Catalyst 3750 or 2970 switches) A Gigabit Ethernet connection between a Serial Gigabit Media Independent Interface SGMII) port (3/4, 7/8, 11/12, 15/16, 19/20, and 23/24) and an Intel Pro/1000T Server Adapter NIC might lose connectivity on the Catalyst 3750G-24T and 3750G-24TS switches. The link activates correctly, but might subsequently stop exchanging data. This is an Intel product defect. The workaround is to use Reduced Gigabit Media Independent Interface (RGMII) ports (1/2, 5/6, 9/10, 13/14, 17/18, and 21/22) instead of SGMII ports. You can also use the speed 1000 interface configuration command to force the speed of the port to 1000 Mbps. (CSCea77032)

Fallback Bridging

These are the fallback bridging limitations:

• (Catalyst 3750 or 3560 switches) If a bridge group contains a VLAN to which a static MAC address is configured, all non-IP traffic in the bridge group with this MAC address destination is sent to all ports in the bridge group. The workaround is to remove the VLAN from the bridge group or to remove the static MAC address from the VLAN. (CSCdw81955)
• (Catalyst 3750 or 3560 switches) Known unicast (secured) addresses are flooded within a bridge group if secure addresses are learned or configured on a port and the VLAN on this port is part of a bridge group. Non-IP traffic destined to the secure addresses is flooded within the bridge group. The workaround is to disable fallback bridging or to disable port security on all ports in all VLANs participating in fallback bridging. To remove an interface from a bridge group and to remove the bridge group, use the no bridge-group bridge-group interface configuration command. To disable port security on all ports in all VLANs participating in fallback bridging, use the no switchport port-security interface configuration command. (CSCdz80499)

HSRP

This is the Hot Standby Routing Protocol (HSRP) limitation:

When the active switch fails in a switch cluster that uses HSRP redundancy, the new active switch might not contain a full cluster member list. The workaround is to ensure that the ports on the standby cluster members are not in the spanning-tree blocking state. To verify that these ports are not in the blocking state, refer to the “Configuring STP” chapter in the software configuration guide. (CSCec76893)

IP

These are the IP limitations:

• (Catalyst 3750 or 3560 switches) The switch does not create an adjacent table entry when the ARP timeout value is 15 seconds and the ARP request times out. The workaround is to not set an ARP timeout value lower than 120 seconds. (CSCea21674)
• When the rate of received DHCP requests exceeds 2,000 packets per minute for a long time, the response time might be slow when you are using the console. The workaround is to use rate limiting on DHCP traffic to prevent a denial of service attack from occurring. (CSCeb59166)
• Some access point (AP)-350 devices are incorrectly discovered as IEEE 802.3af Class 1 devices. These APs should be discovered as Cisco pre-standard devices. The show power inline user EXEC command shows the AP-350 as an IEEE Class 1 device. The workaround is to power the AP by using an AC wall adaptor. (CSCin69533)

IP Telephony

These are the IP telephony limitations:

• When a Cisco IP Phone is connected to the switch, the port VLAN ID (PVID) and the voice VLAN ID (VVID) both learn its MAC address. However, after dynamic MAC addresses are deleted, only the VVID relearns the phone MAC address. MAC addresses are manually or automatically deleted when a topology change occurs or when port security or an 802.1x feature is enabled or disabled. There is no workaround. (CSCea80105)
• After you change the access VLAN on a port that has 802.1x enabled, the IP Phone address is removed. Because learning is restricted on 802.1x capable ports, it takes approximately 30 seconds before the address is relearned. No workaround is necessary. (CSCea85312)

MAC Addressing

This is the MAC addressing limitation:

(Catalyst 3750 or 3560 switches) When a MAC address is configured for filtering on the internal VLAN of a routed port, incoming packets from the MAC address to the routed port are not dropped. (CSCeb67937)

Multicasting

These are the multicasting limitations:

• (Catalyst 3750 or 3560 switches) The switch does not support tunnel interfaces for unicast routed traffic. Only Distance Vector Multicast Routing Protocol (DVMRP) tunnel interfaces are supported for multicast routing.
• (Catalyst 3750 or 3560 switches) Nonreverse-path forwarded (RPF) IP multicast traffic to a group that is bridged in a VLAN is leaked onto a trunk port in the VLAN even if the port is not a member of the group in the VLAN, but it is a member of the group in another VLAN. Because unnecessary traffic is sent on the trunk port, it reduces the bandwidth of the port. There is no workaround for this problem because non-RPF traffic is continuous in certain topologies. As long as the trunk port is a member on a trunk port in at least one VLAN, this problem for the non-RPF traffic occurs. (CSCdu25219)
• If the number of multicast routes and Internet Group Management Protocol (IGMP) groups are more than the maximum number specified by the show sdm prefer global configuration command, the traffic received on unknown groups is flooded in the received VLAN even though the show ip igmp snooping multicast-table privileged EXEC command output shows otherwise. The workaround is to reduce the number of multicast routes and IGMP snooping groups to less than the maximum supported value. (CSCdy09008)
• IGMP filtering is applied to packets that are forwarded through hardware. It is not applied to packets that are forwarded through software. Hence, with multicast routing enabled, the first few packets are sent from a port even when IGMP filtering is set to deny those groups on that port. There is no workaround. (CSCdy82818)
• (Catalyst 3750 or 3560 switches) When you use the ip access-group interface configuration command with a router access control list (ACL) to deny access to a group in a VLAN, multicast data to the group that is received in the VLAN is always flooded in the VLAN, regardless of IGMP group membership in the VLAN. This provides reachability to directly connected clients, if any, in the VLAN. The workaround is to not apply a router ACL set to deny access to a VLAN interface. Apply the security through other means; for example, apply VLAN maps to the VLAN instead of using a router ACL for the group. (CSCdz86110)
• (Catalyst 3750 switch stack) If the stack master is power cycled immediately after the ip mroute global configuration command is entered, there is a slight chance that this configuration change might be lost after the stack master changes. This occurs because the stack master did not have time to propagate the running configuration to all the stack members before it was powered down. This problem might also affect other configuration commands. There is no workaround. (CSCea71255)
• (Catalyst 3750 switches) When IP Protocol-Independent Multicast (PIM) is enabled on a tunnel interface, the switch incorrectly displays the Multicast is not supported on tunnel interfaces error message. IP PIM is not supported on tunnel interfaces. There is no workaround. (CSCeb75366)
• If an IGMP report packet has two multicast group records, the switch removes or adds interfaces depending on the order of the records in the packet:

– If the ALLOW_NEW_SOURCE record is before the BLOCK_OLD_SOURCE record, the switch removes the port from the group.

– If the BLOCK_OLD_SOURCE record is before the ALLOW_NEW_SOURCE record, the switch adds the port to the group.

There is no workaround. (CSCec20128)

• (Catalyst 3750 switches) Although the switchport block multicast interface configuration command appears in the CLI, it is not supported on the Catalyst 3750 switches. There is no workaround. (CSCee16865)

QoS

These are the quality of service (QoS) limitations:

• Some switch queues are disabled if the buffer size or threshold level is set too low with the mls qos queue-set output global configuration command. The ratio of buffer size to threshold level should be greater than 10 to avoid disabling the queue. The workaround is to choose compatible buffer sizes and threshold levels. (CSCea76893)
• When auto-QoS is enabled on the switch, priority queuing is not enabled. Instead, the switch uses shaped round robin (SRR) as the queuing mechanism. The auto-QoS feature is designed on each platform based on the feature set and hardware limitations, and the queuing mechanism supported on each platform might be different. There is no workaround. (CSCee22591)

Routing

These are the routing limitations:

• (Catalyst 3750 or 3560 switches) The switch does not support tunnel interfaces for unicast routed traffic. Only Distance Vector Multicast Routing Protocol (DVMRP) tunnel interfaces are supported for multicast routing.
• (Catalyst 3750 or 3560 switches) A route map that has an ACL with a Differentiated Services Code Point (DSCP) clause cannot be applied to a Layer 3 interface. The switch rejects this configuration and displays a message that the route map is unsupported. There is no workaround. (CSCea52915)
• On a Catalyst 3750 switch stack with a large number of switched virtual interfaces (SVIs), routes, or both on a fully populated nine-member switch stack, this message might appear when you reload the switch stack or add a switch to the stack:

This error message means there is a temporary memory shortage that normally recovers by itself. You can verify that the switch stack has recovered by entering the show cef line user EXEC command and verifying that the line card states are up and sync. No workaround is required because the problem is self-correcting. (CSCea71611)

SPAN and RSPAN

These are the SPAN and Remote SPAN (RSPAN) limitations:

• (Catalyst 3750 or 3560 switches) An egress SPAN copy of routed unicast traffic might show an incorrect destination MAC address on both local and remote SPAN sessions. This limitation does not apply to bridged packets. The workaround for local SPAN is to use the replicate option. For a remote SPAN session, there is no workaround. This is a hardware limitation. (CSCdy72835)
• (Catalyst 3750 or 3560 switches) Egress SPAN routed packets (both unicast and multicast) show the incorrect source MAC address. For remote SPAN packets, the source MAC address should be the MAC address of the egress VLAN, but instead the packet shows the MAC address of the RSPAN VLAN. For local SPAN packets with native encapsulation on the destination port, the packet shows the MAC address of VLAN 1. This problem does not appear with local SPAN when the encapsulation replicate option is used. This limitation does not apply to bridged packets. The workaround is to use the encapsulate replicate keywords in the monitor session global configuration command. Otherwise, there is no workaround. This is a hardware limitation. (CSCdy81521)
• During periods of very high traffic, when two RSPAN source sessions are configured, the VLAN ID of packets in one RSPAN session might overwrite the VLAN ID of the other RSPAN session. If this occurs, packets intended for one RSPAN VLAN are incorrectly sent to the other RSPAN VLAN. This problem does not affect RSPAN destination sessions. The workaround is to configure only one RSPAN source session. (CSCea72326)
• (Catalyst 3750 or 3560 switches) The egress SPAN data rate might degrade when fallback bridging or multicast routing is enabled. The amount of degradation depends on the processor loading. Typically, the switch can egress SPAN up to 40,000 packets per second (64-byte packets). As long as the total traffic being monitored is below this limit, there is no degradation. However, if the traffic being monitored exceeds the limit, only a portion of the source stream is spanned. When this occurs, the following console message appears: Decreased egress SPAN rate. In all cases, normal traffic is not affected; the degradation limits only how much of the original source stream can be egress spanned. If fallback bridging and multicast routing are disabled, egress SPAN is not degraded. There is no workaround. If possible, disable fallback bridging and multicast routing. If possible, use ingress SPAN to observe the same traffic. (CSCeb01216)
• On Catalyst 3750 switches running Cisco IOS Release 12.1(14)EA1 and later and on Catalyst 3560 switches running Cisco IOS release 12.1(19)EA1 or later, some IGMP report and query packets with IP options might not be ingress-spanned. Packets that are susceptible to this problem are IGMP packets containing 4 bytes of IP options (IP header length of 24). An example of such packets would be IGMP reports and queries having the router alert IP option. Ingress-spanning of such packets is not accurate and can vary with the traffic rate. Typically, very few or none of these packets are spanned. There is no workaround. (CSCeb23352)
• Cisco Discovery Protocol (CDP), VLAN Trunking Protocol (VTP), and Port Aggregation Protocol (PAgP) packets received from a SPAN source are not sent to the destination interfaces of a local SPAN session. The workaround is to use the monitor session session_number destination { interface interface-id encapsulation replicate} global configuration command for local SPAN. (CSCed24036)

Stacking (Catalyst 3750 switch stack only)

These are the Catalyst 3750 switch stack limitations:

• If the stack master is immediately reloaded after adding multiple VLANs, the new stack master might fail. The workaround is to wait a few minutes after adding VLANs before reloading the stack master. (CSCea26207)
• If the console speed is changed on a stack, the configuration file is updated, but the baud rate is not. When the switch is reloaded, meaningless characters might appear on the console during bootup before the configuration file is parsed and the console speed is set to the correct value. If manual boot is enabled or the startup configuration is deleted after you change the console speed, you cannot access the console after the switch reboots. There is no workaround. (CSCec36644)
• If a switch is forwarding traffic from a Gigabit ingress interface to a 100 Mbps egress interface, the ingress interface might drop more packets due to oversubscription if the egress interface is on a Fast Ethernet switch (such as a Catalyst 3750-24TS or 3750-48TS switch) than if it is on a Gigabit Ethernet switch (such as a Catalyst 3750G-24T or 3750G-24TS switch). There is no workaround. (CSCed00328)
• If a stack member is removed from a stack and either the configuration is not saved or another switch is added to the stack at the same time, the configuration of the first member switch might be lost. The workaround is to save the stack configuration before removing or replacing any switch in the stack. (CSCed15939)
• When the switchport and no switchport interface configuration commands are entered more than 20,000 times on a port of a Catalyst 3750 switch, all available memory is used, and the switch halts.

There is no workaround. (CSCed54150)

• In a private-VLAN domain, only the default private-VLAN IP gateways have sticky ARP enabled. The intermediate Layer 2 switches that have private VLAN enabled disable sticky ARP. When a stack master switch-over occurs on one of the Catalyst 3750 default IP gateways, the message IP-3-STCKYARPOVR appears on the consoles of other default IP gateways. Because sticky ARP is not disabled, the MAC address update caused by the stack master switch-over cannot complete.

The workaround is to complete the MAC address update by entering the clear arp privileged EXEC command. (CSCed62409)

• When a Catalyst 3750 switch is being reloaded in a switch stack, packet loss might occur for up to 1 minute while the Cisco Express Forwarding (CEF) table is downloaded to the switch. This only impacts traffic that will be routed through the switch that is being reloaded. There is no workaround. (CSCed70894)
• Inconsistent private-VLAN configuration can occur on a switch stack if a new stack master is running the SMI and the old stack master was running the EMI.

Private VLAN is enabled or disabled on a switch stack, depending on whether or not the stack master is running the EMI or the SMI:

– If the stack master is running the EMI, all stack members have private VLAN enabled.

– If the stack master is running SMI, all stack members have private VLAN disabled.

This occurs after a master-switchover (MSO) when the previous stack master was running the EMI and the new stack master is running the SMI. The stack members are configured with private VLAN, but any new switch that joins the stack will have private VLAN disabled.

These are the workarounds. Only one of these is necessary:

– Reload the stack after an EMI to SMI MSO (or the reverse).

– Before an EMI-to-SMI MSO, delete the private-VLAN configuration from the existing stack master. (CSCee06802)

• Port configuration information is lost when changing from switchport to no switchport modes on Catalyst 3750 switches.

This is the expected behavior of the offline configuration (provisioning) feature. There is no workaround. (CSCee12431)

Trunking

These are the trunking limitations:

• The switch treats frames received with mixed encapsulation (802.1Q and Inter-Switch Link [ISL]) as frames with FCS errors, increments the error counters, and causes the LED to blink amber. This happens when an ISL-unaware device receives an ISL-encapsulated packet and forwards the frame to an 802.1Q trunk interface. There is no workaround. (CSCdz33708)
• IP traffic with IP options set is sometimes leaked on a trunk port. For example, a trunk port is a member of an IP multicast group in VLAN X but is not a member in VLAN Y. If VLAN Y is the output interface for the multicast route entry assigned to the multicast group and an interface in VLAN Y belongs to the same multicast group, the IP-option traffic received on an input VLAN interface other than one in VLAN Y is sent on the trunk port in VLAN Y because the trunk port is forwarding in VLAN Y, even though the port has no group membership in VLAN Y. There is no workaround. (CSCdz42909).
• If a Catalyst 3750 switch stack is connected to a designated bridge and the root port of the switch stack is on a different switch than the alternate root port, changing the port priority of the designated ports on the designated bridge has no effect on the root port selection for the Catalyst 3750 switch stack. There is no workaround. (CSCea40988)
• For trunk ports or access ports configured with 802.1Q tagging, inconsistent statistics might appear in the show interfaces counters privileged EXEC command output. Valid 802.1Q frames of 64 to 66 bytes are correctly forwarded even though the port LED blinks amber, and the frames are not counted on the interface statistics. There is no workaround. (CSCec35100).

VLAN

This is the VLAN limitation:

If the number of VLANs times the number of trunk ports exceeds the recommended limit of 13,000, the switch can fail. The workaround is to reduce the number of VLANs or trunks. (CSCeb31087)

Cluster Limitations and Restrictions

These limitations apply to the Catalyst 3750, 3560, and 2970 switches:

• When there is a transition from the cluster active command switch to the standby command switch, Catalyst 1900, 2820, and 2900 4-MB switches that are cluster members might lose their cluster configuration. You must manually add these switches back to the cluster.
  (CSCds32517 and CSCds55711)
• When a Catalyst 2900 XL or 3500 XL cluster command switch is connected to a Catalyst 3550 or to a 3750 switch, the command switch does not find any cluster candidates beyond the Catalyst 3550 or the 3750 switch if it is not a member of the cluster. You must add the Catalyst 3550 or the 3750 switch to the cluster. You can then see any cluster candidates connected to it. (CSCdt09918)
• If both the active command switch and the standby command switch fail at the same time, the cluster is not automatically recreated. Even if there is a third passive command switch, it might not recreate all cluster members because it might not have all the latest cluster configuration information. You must manually recreate the cluster if both the active and standby command switches simultaneously fail. (CSCdt43501)

CMS Limitations and Restrictions

These limitations apply to the Catalyst 3750, 3560, and 2970 switches:

• CMS performance degrades if the Topology View is open for several hours on a Solaris machine. The cause might be a memory leak.The workaround is to close the browser, reopen it, and launch CMS again. (CSCds29230)
• If you are printing a Topology View or Front Panel View that contains many devices and are running Solaris 2.6 with JDK1.2.2, you might get an Out of Memory error message. The workaround is to close the browser, re-open it, and launch CMS again. Before you perform any other task, open the view that you want to print, and click Print in the CMS menu. (CSCds80920)
• A red border appears around the text-entering area of some CMS dialogs. The color of the border changes to green when text is entered. This is only a cosmetic error. The colored border does not prevent you from entering text. (CSCdv82352)
• You cannot switch modes (for example, from Guide Mode to Expert Mode) for an open CMS window. The workaround is to close the open window, select the mode that you want, and then reopen the CMS window. For the mode change to take effect on any other CMS window that is open, you need to close that window and then reopen it after you select the new mode. (CSCdw87550)
• If you open a window in which you can enter text, open another window, and return to the first window, right-clicking in the text field might make the cursor in this field disappear. You can still enter text in the field. (CSCdy44189)
• CMS fails when a switch is running the cryptographic software image and the vty lines have been configured to use only SSH using the transport input ssh and line vty 0 15 global configuration commands. The workaround is to allow SSH and Telnet access through the vty lines by using the transport input ssh telnet and line vty 0 15 global configuration command. (CSCdz01037)
• When you add a new member with a username and password that is different from the existing cluster member usernames and passwords, CMS produces an exception error because of an authentication failure. The workaround is to add the new member without any username and password. When the new member is added to the cluster, remove the existing username and password from the Username and Password fields, enter a new username and password, and then apply it to all cluster members. (CSCdz07957)
• When the Link Graphs application has run for hours displaying packet drop and error information, sometimes the X-axis crosses the Y-axis at a negative y value instead of at y = 0. This condition occurs with all supported operating systems, browsers, and Java plug-ins. There is no workaround. (CSCdz32584)
• After you click Apply or Refresh in the Simple Network Management Protocol (SNMP) window, the window size changes. (CSCdz75666, CSCdz84255)
• When you enable log scaling for Link Graphs, the Y-axis scale becomes illegible. There is no workaround. (CSCdz81086)
• The CMS window does not return to full size after resizing the NE or IE when using Netscape version 6.xx on Solaris and Linux. This is a Netscape browser problem. There is no workaround. (CSCea01179)
• CMS sometimes halts after you click Apply when using Netscape 4.7 on the Japanese version of Windows 98 or Windows ME. The workaround is to use Microsoft Internet Explorer or Netscape 6.0 or later. (CSCea27408)
• Changing the password or current authentication while CMS is running causes HTTP requests to fail. The workaround is to close all browser sessions and then relaunch CMS. (CSCeb33995)
• The CMS plug-in is not supported in Netscape 4.7x. The workaround is to use a supported browser, such as Netscape 7.1 or Internet Explorer 5.5 or 6.0. (CSCed21655)
• When TACACS authentication is only enabled on a command switch, member switches cannot be configured. The workaround is to enable TACACS authentication on the member switches. (CSCed27723)
• If an ACL is deleted from a device, all QoS classes that use this ACL for traffic classification become unusable (only on Catalyst 2970 and 3750 switches). The modification of these classes to use any other traffic classification (match statement) fails. The workaround is to delete the QoS class that uses the undefined ACL and then recreate it with the intended traffic classification (match statement). (CSCed40866)
• Host names and Domain Name System (DNS) server names that contain commas on a cluster command switch, member switch, or candidate switch can cause CMS to behave unexpectedly. You can avoid this instability in the interface by not using commas in host names or DNS names. Do not enter commas when also entering multiple DNS names in the IP Configuration tab of the IP Management window in CMS.
• Access control entries (ACEs) that contain the host keyword precede all other ACEs in standard ACLs. You can reposition the ACEs in a standard ACL with one restriction: No ACE with the any keyword or a wildcard mask can precede an ACE with the host keyword.

Switch Stack Notes

These notes apply to switch stacks:

• Always power off a switch before adding or removing it from a switch stack.
• The Catalyst 3560 and 2970 switches do not support switch stacking. However, the show processes privileged EXEC command still lists stack-related processes. This occurs because these switches share common code with other switches that do support stacking.

Cisco IOS Notes

These notes apply to Cisco IOS software:

• The 802.1x feature in Cisco IOS Release 12.1(14)EA1 and later is not fully backward-compatible with the same feature in Cisco IOS Release 12.1(11)AX. If you are upgrading a Catalyst 3750 or a 2970 switch running Cisco IOS Release 12.1(11)AX that has 802.1x configured, you must re-enable 802.1x after the upgrade by using the dot1x system-auth-control global configuration command. This global command does not exist in Cisco IOS Release 12.1(11)AX. Failure to re-enable 802.1x weakens security because some hosts can then access the network without authentication.
• The behavior of the no logging on global configuration command changed in Cisco IOS Release 12.2(18)SE and later. In Cisco IOS Release 12.1(19)EA and earlier, both of these command pairs disabled logging to the console:

– the no logging on and then the no logging console global configuration commands

– the logging on and then the no logging console global configuration commands

In Cisco IOS Release 12.2(18)SE and later, you can only use the logging on and then the no logging console global configuration commands to disable logging to the console. (CSCec71490)

CMS Notes

These notes apply to CMS configuration:

• If you use CMS on Windows 2000, it might not apply configuration changes if the enable password is changed from the CLI during your CMS session. You have to restart CMS and enter the new password when prompted. Platforms other than Windows 2000 prompt you for the new enable password when it is changed.
• CMS does not display QoS classes that are created through the CLI if these classes have multiple match statements. When using CMS, you cannot create classes that match more than one match statement. CMS does not display policies that have such classes.
• If you use Internet Explorer Version 5.5 and select a URL with a nonstandard port at the end of the address (for example, www. cisco.com :84), you must enter http:// as the URL prefix. Otherwise, you cannot launch CMS.
• Within an ACL, you can change the sequence of ACEs that have the host keyword. However, because such ACEs are independent of each other, the change has no effect on the way the ACL filters traffic.
• If you have a proxy server configured on your web browser, CMS can run slowly and take 2 to 3 minutes to process each command that is entered.
• If you use the Netscape browser to view the CMS GUI and you resize the browser window while CMS is initializing, CMS does not resize to fit the window.

The workaround is to resize the browser window again when CMS is not busy.

• In the Front Panel view or Topology view, CMS does not display error messages in read-only mode for these switches:

– Catalyst 2900 XL or Catalyst 3500 XL member switches running Cisco IOS Release 12.0(5)WC2 or earlier

– Catalyst 2950 member switches running Cisco IOS Release 12.0(5)WC2 or earlier

– Catalyst 3550 member switches running Cisco IOS Release 12.1(6)EA1 or earlier

In the Front Panel view, if the switch is running one of the software releases listed previously, the device LEDs do not appear. In Topology view, if the member is an LRE switch, the CPE devices that are connected to the switch do not appear. The Bandwidth and Link graphs also do not appear in these views.

Open Cisco IOS Caveats

Unless otherwise noted, these severity 3 Cisco IOS configuration caveats apply to the Catalyst 3750, 3560, and 2970 switches:

• CSCeb42949 (Catalyst 3750 switches)

A Catalyst 3750 switch does not work with the User Registration Tool (URT). The PC attempting to connect to the network can log in successfully, but it is not allowed to pass traffic after the port is moved to the user VLAN. The MAC address for that device shows BLOCKED.

There is no workaround.

• CSCec75279 (Catalyst 3750 switches)

Sometimes when sticky secure addresses are configured and some sticky addresses have been learned, and then switches are added to a switch stack, the number of sticky addresses shown is incorrect.

The workaround is to disable sticky addresses by using the no switchport port-security mac-address sticky interface configuration command before adding new stack members.

• CSCec84254

Some Catalyst 2950, 3550, and 3750 switches do not link up with some media converters running at 100 Mbps. This affects the 10/100BASE-T interfaces on these Catalyst 2950 switches:

– Catalyst 2950-12

– Catalyst 2950-24

– Catalyst 2950C-24

– Catalyst 2950SX-24

– Catalyst 2950SX-48-SI

– Catalyst 2950T-24

– Catalyst 2950T-48-SI

This affects the 10/100BASE-T interfaces on these Catalyst 3550 switches:

– Catalyst 3550-24-DC-SMI

– Catalyst 3550-24 running the SMI or EMI

– Catalyst 3550-48 running the SMI or EMI

This affects the 10/100BASE-T interfaces on these Catalyst 3750 switches:

– Catalyst 3750-24TS running the SMI or EMI

– Catalyst 3750-48TS running the SMI or EMI

– This problem does not affect any other Catalyst switches. This problem occurs only when the listed switch is running one of these software releases:

– Cisco IOS Release 12.1(13)EA1, Cisco IOS Release 12.1(13)EA1a, Cisco IOS
Release 12.1(13)EA1b, or Cisco IOS Release 12.1(13)EA1c

– Cisco IOS Release 12.1(14)EA1, Cisco IOS Release 12.1(14)EA1a, or Cisco IOS
Release 12.1(14)EA1b

– Cisco IOS Release 12.1(19)EA1, Cisco IOS Release 12.1(19)EA1a, Cisco IOS Release 12.1(19)EA1b, Cisco IOS Release 12.1(19)EA1c, or Cisco IOS Release 12.1(19)EA1d.

The workaround is to use Cisco IOS Release 12.1(12c)EA1 or earlier.

• CSCec88895 (Catalyst 3750 switches)

There is a discrepancy between the output of the show controllers ethernet-controller tengigabitethernet1/0/1 and the show interfaces tengigabitethernet1/0/1 privileged EXEC commands on a 10-Gigabit Ethernet interface.

The workaround for 10-Gigabit Ethernet interfaces is to use the show interface privileged EXEC command for the byte count and the number of pause frames received. Use the show controllers ethernet-controller privileged EXEC command for the frame count and the FCS and CRC error-frame count.

• CSCed11059 (Catalyst 3750 switches)

When cross-stack UplinkFast (CSUF) is configured and you quickly enter the shutdown interface configuration command followed by the no shutdown interface configuration commands, one of the uplink ports might cause the uplink ports to be blocked in some VLANs for twice the forward-delay time, causing CSUF to not work for some VLANs. The new root port begins forwarding on its own.

There is no workaround. The new root port will go to a forwarding state on its own after twice the forward delay value on the VLANs.

• CSCed12889 (Catalyst 3750 switches)

When redundant uplinks are from the same stack member in a switch stack and UplinkFast is configured, dummy multicast packets are not sent.

The workaround is to not have redundant uplinks from the same stack member. Provide uplink connectivity from ports across the switch stack rather than from one switch in the stack.

• CSCed37222 (Catalyst 3750 switches)

When port security is enabled and some sticky addresses are known, and a stack master switchover happens before all of the addresses have been propagated through the switch stack, sometimes sticky addresses are lost.

There is no workaround.

• CSCed46277 (Catalyst 3750 switches)

After the stack master fails and another is elected, switch ports on the new stack master lose the hardware configuration of per-user ACLs even though the Cisco IOS software shows the ACL as installed. This problem does not affect stack members.

There is no workaround.

• CSCed53633 (Catalyst 3750 switches)

A spanning-tree loop might occur if all of these conditions are true:

– Port security is enabled with the violation mode set to protected.

– The maximum number of secure addresses is less than the number of switches connected to the port.

– There is a physical loop in the network through a switch whose MAC address has not been secured, and its BPDUs cause a secure violation.

The workaround is to change any one of the listed conditions.

• CSCed63521 (Catalyst 3750 switches)

Some ports of a stack member might not be able to communicate with other ports of other stack members even though the switch MAC address table and CEF table from the stack master are correct. Some ports in the affected switch might be able to communicate by using stale CEF entries. The Catalyst 3750 stack member switch fails to download CEF tables from the stack master.

There is no workaround.

• CSCed65309 (Catalyst 3750 or 3560 switches)

Some invalid ARP packets are not dropped on dynamic ARP inspection-enabled VLANs. Dynamic ARP inspection does not verify that certain ARP fields are valid and does not drop ARP packets with invalid values for those fields. The fields are hardware size, protocol size, and operation type. These packets also are not dropped by the switch on nondynamic ARP-enabled VLANs.

There is no workaround.

• CSCed65410 (Catalyst 3750 or 3560 switches)

If dynamic ARP inspection is enabled on an internal VLAN used by a routed port, ARP traffic on the routed port is affected by the dynamic ARP inspection processing. For example, ARP packets will be rate-limited.

The workaround is to not enable dynamic ARP inspection on internal VLANs.

• CSCed71422 (Catalyst 3750 or 3560 switches)

A CPUHOG message sometimes appears when you configure a private VLAN. Port security must be enabled on one or more ports affected by the private VLAN configuration.

There is no workaround.

• CSCed74349 (Catalyst 3750 switches)

An EtherChannel is not properly error-disabled if these conditions are true:

– The channel is carrying a VLAN that is enabled for dynamic ARP inspection.

– The channel is configured with a rate limit for dynamic ARP inspection.

– At least one of the ports in the channel is on a stack member.

– ARP packets are received on a port in the channel on a stack member at a higher rate than the configured rate limit for the channel.

Under these circumstances, a system message states that the rate limit was exceeded on the channel, but the channel will not be error-disabled.

The workaround is to use physical ports on the stack master for any EtherChannel that carries dynamic ARP inspection VLANs and has rate limits.

• CSCed87243

If the VTP password is configured but the VTP domain name is not configured and if the switch reloads twice, the switch does not retain the VLAN information.

Use of these workarounds:

– Delete the vlan.dat file, which deletes the VTP password.

– Delete the VTP password by using the no vtp password global configuration command.

– Assign a VTP domain name.

• CSCed91730 (Catalyst 3750 switches)

When a secondary VLAN is associated and then quickly disassociated, sometimes the MAC address tables across the switch stack become unsynchronized. This is a rare condition that happens when Port Fast is enabled on the host ports and traffic is continuously received on that port.

The workaround is to clear the MAC address table by using the clear mac address-table dynamic privileged EXEC command.

• CSCed94657 (Catalyst 3750 or 3560 switches)

If a secondary VLAN that was mapped to a promiscuous port is disassociated from the primary VLAN, the LED on the port turns from green to amber. This also occurs if the secondary VLAN is deleted.

The workaround is to remove the secondary VLAN from the mapping of the promiscuous port.

• CSCed95822 (Catalyst 3750 switches)

Dynamic ARP inspection log entries might be lost after a switch failure. Any log entries that are still in the log buffer (have not been output as a system message) on a switch that fails will be lost.

When you enter the show ip arp inspection log privileged EXEC command, the log entries from all switches in the stack are moved to the switch on which the command was entered.

There is no workaround.

• CSCee07107 (Catalyst 3750 switches)

ARP and reverse ARP (RARP) packets are not properly filtered by a configured VLAN map. If you enable a VLAN for dynamic ARP inspection and a VLAN map is applied to the VLAN, ARP and RARP packets received in that VLAN on stack member ports that should be dropped by the VLAN map are not dropped.

There is no workaround.

• CSCee08756

Configuring multiple ports to a static address in a private VLAN is not supported in this release. If you add more than one port to a static address in a private VLAN, the traffic destined to that static address from a host (secondary VLAN) port to promiscuous port might be dropped.

The workaround is to not configure multiple ports to a static address in a private VLAN. You can use the shutdown and no shutdown interface configuration commands on a promiscuous port to resume the flow of traffic.

• CSCee11197 (Catalyst 3750 switches)

You can only enter values ranging from 1 to 1023 when configuring the VLAN for an access port from SNMP by using the vlanPortVlan object of the CISCO-STACK-MIB.

These are the workarounds:

– Use the interface vlan global configuration command to configure the VLAN for the access port.

– From SNMP, use the vmVlan object of the CISCO-VLAN-MEMBERSHIP-MIB.

You can use both of these workarounds to enter a value ranging from 1 to 4095.

• CSCee11587 (Catalyst 3750 switches)

These CISCO-STACK-MIB objects always return the invalid value of zero:

– vlanPortSwitchLevel

– vlanPortIslAdminStatus

– vlanPortIslOperStatus

– vlanPortAdminStatus

There is no workaround.

• CSCee14018 (Catalyst 3750 or 3560 switches)

Port ACLs are not applied to IGMP control packets with IP options.

There is no workaround.

• CSCee14293 (Catalyst 3750 or 3560 switches)

After a multicast group exceeds the maximum number that a private VLAN can support, the required ternary content addressable memory (TCAM) entries cannot present for the last group, and the forwarding behavior for that multicast group is incorrect.

For a private VLAN multicast group, each group needs 3 TCAM entries (one SFT entry and 2 LFT entries) when IP multicast routing is enabled on the private VLAN primary VLAN. (For a regular VLAN, only 1 SFT TCAM entry is required, and approximately1000 groups can be supported. For the private VLAN group, only one third of the regular groups can be supported.

There is no workaround.

• CSCee14673 (Catalyst 3750 switches)

A Catalyst 3750 switch running Cisco IOS Release 12.1(19)EA1a might continuously show this message:

There is no workaround.

• CSCee21251

Auto-RP discovery packets (addressed to 224.0.1.40) received from a mapping agent are not forwarded from the receiving switch when the receiving port is a routed port. This problem does not occur if these packets are received on a port on a VLAN interface.

There is no workaround.

• CSCee21842 (Catalyst 3750 or 3560 switches)

When static MAC addresses are configured on two isolated ports and then the isolated VLAN is changed to a community VLAN, Layer 2 traffic is blocked between the two private VLAN community ports.

The workaround is to remove the static MAC address by using no mac address-table static mac-addr vlan vlan-id [ interface interface-id ] global configuration command and then to reconfigure that static address by using the mac address-table static mac-addr vlan vlan-id interface interface-id global configuration command.

• CSCee22376

When an SNMP version 3 user is configured with the encrypted option and password, the switch reloads when the MIB object usmUserAuthKeyChange is set.

The workaround is to configure a user without the encrypted option. (For example, snmp-server user username groupname v3 auth md5 password.)

• CSCee30022

If you try to add an aggregate policer to a policy map, this message appears:

and the aggregate policer is not added.

The workaround is to delete the policy map by using the no policy-map policy-map-name global configuration command, recreate it with the desired configuration, and then re-attach it to the interfaces by using the service-policy input policy-map-name interface configuration command.

• CSCee30090

If you modify a policer, this message appears:

If you then attempt to remove an aggregate policer, the removal of the policy map fails, and this message appears:

The workaround is to delete the policy map by using the no policy-map policy-map-name global configuration command, recreate it with the desired configuration, and then re-attach it to the interfaces by using the service-policy input policy-map-name interface configuration command.

• CSCee30129

When you add an aggregate policer to a policy-map class, the aggregate policer is also added to another policy class within the same policy.

The workaround is to delete the policy map by using the no policy-map policy-map-name global configuration command, recreate it with the desired configuration, and then re-attach it to the interfaces by using the service-policy input policy-map-name interface configuration command.

• CSCee53804 (Catalyst 3750 or 3560 switches)

When enabled, DHCP snooping does not work with secondary VLANs of a private VLAN. DHCP discover messages from the private-VLAN hosts are not broadcast, and private-VLAN hosts cannot communicate with the DHCP server.

There is no workaround.

• CSCin68965 (Catalyst 3750 or 3560 switches)

When two ports of a Cisco IP Phone are connected to a switch and the higher voice VLAN ID (VVID) is configured on the switch port to which port P3 of the Cisco IP Phone is connected, the phone displays configuring IP and halts.

These are the workarounds. Only one of these is necessary:

– Configure the higher VVID on port P1 of the Cisco IP phone.

– Connect only one port of the Cisco IP Phone to the switch.

Open CMS Caveats

Unless otherwise noted, these severity 3 CMS caveats apply to the Catalyst 3750, 3560, and 2970 switches:

• CSCec61919

When a switch cluster has only one member switch and that member switch is down, CMS does not display the Remove From Cluster option.

There is no workaround.

• CSCed39693

When there are Catalyst 2950 and 2955 devices in a cluster, if you launch the QoS Queue Window to configure the devices and then try to view the settings for other devices by using the device selection menu, CMS halts after 20 to 30 selections.

The workaround is to close and then restart CMS.

• CSCed87031

When an Open Shortest Path First (OSPF) summary address is added for a 10.x.x.x network, a Windows exception error sometimes occurs.

The workaround is to add the address by using the router ospf <process-id>, area <area-id>, and range <address> <mask> configuration commands.

• CSCed88494

When you change the Spanning Tree Protocol (STP) mode from Rapid PVST+ to PVST+, a Java OutOfBoundsException error sometimes appears.

There is no workaround. The new STP mode is still configured even if the error message appears.

• CSCee06206

When a Catalyst 3750 stack member leaves or joins the switch stack, the entire stack disappears from the Topology View. Only the stack member that has left the stack should disappear from the Topology view.

There is no workaround.

• CSCee06244

When you select a remote device from the VLAN menu, the displayed table sometimes does not show all the connected links between the device selected in the Host Name and the Remote Device lists. This can also occur when you add a new device to a cluster and then open VLAN menu.

This is the workaround:

1. Click Refresh on the CMS toolbar two or three times, or select View > Refresh two or three times.

2. Click Refresh in VLAN Window.

• CSCee11710

The Telnet link on the TOOLS page (select TOOLS from the switch home page) does not work on Solaris systems.

There is no workaround.

• CSCee15761

The Device Manager Launch button does not work for Catalyst 1900 and 2820 switches.

The workaround is to launch Device Manager for these devices outside of CMS by opening a new browser and manually entering the URL for the switch.

• CSCee26671

When you click Refresh in the Stack Settings dialog, the latest information switch cluster does not appear.

The workaround is to close and then to reopen the Stack Settings dialog.

• CSCee25870

A Java exception error occurs when CMS is in read-only mode and you launch the Port Settings dialog. This only occurs on Catalyst 2900 XL, 3500 XL, and 2950 LRE switches.

The workaround is to open the Port Settings dialog with CMS in read-write mode.

• CSCee26637

When you open the Port Settings dialog for a Power-over-Ethernet (PoE) switch that is a member of a switch stack and the stack master is not a PoE switch, a Java exception error occurs.

The workaround is to configure the PoE switch as the stack master.

Cisco IOS Caveats Resolved in Cisco IOS Release 12.2(20)SE

Unless otherwise noted, these caveats were resolved in this release for the Catalyst 3750, 3560, and 2970 switches:

• CSCdz30046 (Catalyst 3560 switches)

When multicast VLAN registration (MVR) groups are added or deleted, the receiver port that joined the groups after the addition no longer receives traffic after the group is deleted. MVR data traffic to the group is no longer sent to the receiver port immediately after the no mvr group ip-address global configuration command is entered.

• CSCeb67510

When both the sharing and shaping weights are enabled, the receiving rates now follow the shared bandwidth weight if the priority queue is enabled on the egress queue.

• CSCec07637 (Catalyst 3560 switches)

When an ACL that denies packets is configured on an ingress or egress interface, the CPU usage is no longer as high as 70 percent when these packets are forwarded to the CPU to determine if an ICMP-unreachable packet should be generated.

• CSCec11048 (Catalyst 3560 switches)

When a configured secure MAC address exists on an interface, you can now change it to a sticky MAC address. Alternatively, if a sticky MAC address exists on an interface, you can now change it to a secure MAC address.

• CSCec12147 (Catalyst 3560 switches)

When the CISCO-STP-EXTENSIONS-MIB is polled, unknown indexes are no longer returned for some MIB objects.

• CSCec16481 (Catalyst 3750 or 3560 switches)

A Cisco device running Internetwork Operating System (IOS) and enabled for the Open Shortest Path First (OSPF) Protocol is vulnerable to a Denial of Service (DoS) attack from a malformed OSPF packet. The OSPF protocol is not enabled by default.

The vulnerability is only present in IOS release trains based on 12.0S, 12.2, and 12.3. Releases based on 12.0, 12.1 mainlines and all IOS images prior to 12.0 are not affected. Refer to the Security Advisory for a complete list of affected release trains.

Further details and the workarounds to mitigate the effects are explained in the Security Advisory which is available at the following URL:

http://www.cisco.com/warp/public/707/cisco-sa-20040818-ospf.shtml

• CSCec21040 (Catalyst 3560 switches)

When an 802.1x-enabled port is authenticated with a RADIUS-assigned VLAN, if the port is shut down or the link is removed, a traceback message no longer appears.

• CSCec21320 (Catalyst 3560 switches)

After a link is up, a switch sends three Extensible Authentication Protocol (EAP) Request/Identity messages to the client. There is a 30-second gap between messages. However, PCs that are running Windows XP or Windows 2000 drop the first message so that the second message that the client receives appears to be the first, which is at least 30 seconds after the link is up. Therefore, a user does not see a password window until at least 30 seconds after the link is up.

• CSCec22431 (Catalyst 3560 switches)

Telnet and ping traffic is no longer disrupted during SNMP polling of the VlanTrunkPortTable table in the CISCO-VTP-MIB.

• CSCec22572 (Catalyst 3560 switches)

When per-user access control lists (ACLs) are downloaded from a RADIUS server after successful 802.1x authentication, disabling 802.1x now removes the attached per-user ACLs from the interface.

• CSCec27421 (Catalyst 3560 switches)

If QoS is enabled and the trust state is not configured on an ingress interface, now only the mapping of the class of service (CoS) value of 0 to the ingress or egress queues takes effect when you enter the mls qos srr-queue input cos-map or the mls qos srr-queue output cos-map global configuration command. Other CoS values DSCP values to queue mapping have no effect on traffic from that interface.

• CSCec29970 (Catalyst 3560 switches)

If you change the input priority queue for queue 2 by using the mls qos srr-queue input priority-queue 2 bandwidth global configuration command, the configurations that are generated no longer contain an extra input keyword such as mls qos srr-queue input priority-queue input 2 bandwidth. In previous releases, the extra keyword caused an error message if the command was saved and the switch was reloaded.

• CSCec31436 (Catalyst 3560 switches)

When there are many configured secure and sticky MAC addresses on a port, addresses are no longer dropped and removed from the configuration when the switch restarts.

• CSCec32453 (Catalyst 3560 switches)

When you configure a unicast MAC address filter that matches a Windows XP 802.1x client MAC address, the Windows XP 802.1x client now no longer repeatedly tries to re-authenticate itself.

• CSCec35148 (Catalyst 3560 switches)

Processor memory no longer leaks if you change the policy-based routing (PBR) configuration.

• CSCec89120 (Catalyst 3750 or 3560 switches)

The command switch now discovers candidates more than one CDP hop beyond its routed port.

• CSCed04063

When the kerberos clients mandatory global configuration command is entered on a switch and the switch is connected to a host that does not support Kerberos through a Telnet session, the switch no longer halts when the you press the Enter key.

• CSCed18488 (Catalyst 3750 or 3560 switches)

When (*,G) and (S,G) entries are created in a multicast routing table on a remote port by Protocol-Independent Multicast-Sparse Mode (PIM-SM) registering, the RPF leak flag is now set for hardware entry for the group.

• CSCed30095 (Catalyst 3750 switches)

A topology change on a member switch no longer causes fast-aging of the dynamically learned addresses. In previous releases, this occurred in per-VLAN spanning-tree (PVST) mode when a topology change notification (TCN BPDU) was generated and propagated from a member switch but was not sent from the root port on the master.

• CSCed33792 (Catalyst 3750 switches)

Members of a switch stack no longer fail after the debug all privileged EXEC command is entered.

• CSCed34921 (Catalyst 3750 switches)

Changing the LACP system-priority, either locally or on the neighbor switch, no longer creates assert failure and traceback error messages for the ports in the EtherChannel if there is a Layer-3 (routed port) Link Aggregation Control Protocol (LACP) EtherChannel on the s tack master,

• CSCed54175 (Catalyst 3750 or 3560 switches)

The switch now accepts duplicate remark statements in named ACLs.

• CSCee02006 (Catalyst 3750 switches)

A Catalyst 3750 stack member switch no longer reloads or displays a message similar to this:

A MAC address is now correctly learned on a secure port, ages out, and is then learned on another secure port on a different stack member switch.

Cisco CMS Caveats Resolved in Cisco IOS Release 12.2(20)SE

Unless otherwise noted, these caveats were resolved in this release for the Catalyst 3750, 3560, and 2970 switches:

• CSCec18805

In the IP Multicast Wizard, multicast-enabled member devices are now correctly listed in the Enabled Multicast list box instead of in the Current Candidate list box.

• CSCed34582

The Front Panel View now correctly displays the port LEDs.

Corrections to the Catalyst 3750, 3560, and 2970 Switch Software Configuration Guides

In printed copies of the software configuration guides, the URL listed in the “Privilege Levels” section of the “Getting Started with CMS” chapter is incorrect. The section lists this URL:

http://ip_address/level/13

This is the correct URL (the closing “/” is required):

http://ip_address/level/13/

In printed copies of the software configuration guides, in the “Classifying Traffic by Using ACLs” section of the “Configuring QoS” chapter, this information in Step 3 to create a Layer 2 MAC ACL is incorrect:

• For src-MAC-addr, enter the MAC address of the host from which the packet is being sent. You specify this by using the hexadecimal format (H.H.H), by using the any keyword as an abbreviation for source 0.0.0, source-wildcard 255.255.255, or by using the host keyword for source 0.0.0.
• For dst-MAC-addr, enter the MAC address of the host to which the packet is being sent. You specify this by using the hexadecimal format (H.H.H), by using the any keyword as an abbreviation for source 0.0.0, source-wildcard 255.255.255, or by using the host keyword for source 0.0.0.

This is the correct information:

• For src-MAC-addr, enter the MAC address of the host from which the packet is being sent. You specify this by using the hexadecimal format (H.H.H), by using the any keyword as an abbreviation for source 0.0.0, source-wildcard ffff.ffff.ffff, or by using the host keyword for source 0.0.0.
• For dst-MAC-addr, enter the MAC address of the host to which the packet is being sent. You specify this by using the hexadecimal format (H.H.H), by using the any keyword as an abbreviation for source 0.0.0, source-wildcard ffff.ffff.ffff, or by using the host keyword for source 0.0.0.

Additions to the Catalyst 3750 Switch Software Configuration Guide

The next sections provide updated information for the “Managing Switch Stacks” chapter.

Major Version Number Incompatibility Among Switches

---

Note The information in the “Major Incompatibility Between Switches” section was retitled and should be replaced with this information.

Switches with different Cisco IOS software versions likely have different stack protocol versions. Switches with different major version numbers are incompatible and cannot exist in the same switch stack.

Minor Version Number Incompatibility Among Switches

---

Note The information in the “Minor Incompatibility Between Switches” section was retitled and should be replaced with this information.

Switches with the same major version number but with a different minor version number as the stack master are considered partially compatible. When connected to a switch stack, a partially compatible switch enters version-mismatch (VM) mode and cannot join the stack as a fully functioning member. The software detects the mismatched software and tries to upgrade (or downgrade) the switch in VM mode with the switch stack image or with a tar file image from the switch stack flash memory. The software uses the automatic upgrade (auto-upgrade) and the automatic advise (auto-advise) features. For more information, see the “Understanding Auto-Upgrade and Auto-Advise” section.

To see if there are switches in VM mode, use the show switch user EXEC command. The port LEDs on switches in VM mode will also stay off. Pressing the Mode button does not change the LED mode.

Understanding Auto-Upgrade and Auto-Advise

---

Note This is a new section, not previously in the “Managing Switch Stacks” chapter.

When the software detects mismatched software and tries to upgrade the switch in VM mode, two software processes are involved:

• Automatic upgrade (auto-upgrade)—a process that automatically copies (auto-copy) the software image running on any stack member or copies a tar file from the switch stack flash memory to the switch in VM mode to upgrade (auto-upgrade) it. By default, auto-upgrade is enabled (the boot auto-copy-sw global configuration command is enabled). You can disable auto-upgrade by using the no boot auto-copy-sw global configuration command on the stack’ active switch. You can check the status of auto-upgrade by using the show boot privileged EXEC command and by checking the Auto upgrade line in the display.

Auto-upgrade occurs if it is enabled, if there is enough flash memory in the switch in VM mode, and if:

– The software image running on the switch stack is suitable for the switch in VM mode, or

– There is a tar file from the switch stack that is suitable for the switch in VM mode. A switch in VM mode might not run all released software. For example, new switch hardware is not recognized in earlier versions of software.

The auto-upgrade and the auto-copy processes wait for a few minutes before starting.

When the auto-upgrade process is complete, the switch that was in VM mode reloads and joins the stack as a fully functioning member. If you have both StackWise cables connected during the reload, network downtime does not occur because the switch stack operates on two rings.

---

Note Auto-upgrade performs the upgrade only when the two images are the same type. For example, it does not automatically upgrade a switch in VM mode from EMI to SMI (or the reverse) or from cryptographic to noncryptographic (or the reverse).

• Automatic advise (auto-advise)— when the auto-upgrade process cannot find appropriate stack member software to copy to the switch in VM mode, the auto-advise process tells you the command (archive copy-sw or archive download-sw privileged EXEC command) and the image name (tar filename) needed to manually upgrade the switch stack or the switch in VM mode. The recommended image can be the running switch stack image or a tar file in any flash file system in the switch stack (including the switch in VM mode). Auto-advise cannot be disabled, and there is no command to check its status.

The auto-advise software does not give suggestions when the switch stack software and the software of the switch in VM mode do not contain the same feature sets. For example, if the switch stack is running the SMI and you add a switch that is running the EMI, the auto-advise software does not provide a recommendation. The same events occur when cryptographic and noncryptographic images are running.

Auto-Upgrade and Auto-Advise Example Mess1ages

---

Note This is a new section, not previously in the “Managing Switch Stacks” chapter.

When you add a switch that has a different minor version number to the switch stack, the software displays messages in sequence (assuming that there are no other system messages generated by the switch).

This example shows that the switch stack detected a new switch that is running a different minor version number than the switch stack. Auto-copy launches, finds suitable software to copy from a stack member to the switch in VM mode, upgrades the switch in VM mode, and then reloads it:

This example shows that the switch stack detected a new switch that is running a different minor version number than the switch stack. Auto-copy launches but cannot find software in the switch stack to copy to the switch in VM mode to make it compatible with the switch stack. The auto-advise process launches and recommends that you download a tar file from the network to the switch in VM mode:

For information about using the archive download-sw privileged EXEC command, refer to the “Working with Software Images” section in Appendix B, “Working with the Cisco IOS File System, Configuration Files, and Software Images.”

---

Note Auto-advise and auto-copy identify which images are running by examining the info file and by searching the directory structure on the switch stack. If you download your image by using the copy tftp: command instead of by using the archive download-sw privileged EXEC command, the correct directory structure is not properly created. For more information about the info file, see the “tar File Format of Images on a Server or Cisco.com” section in Appendix B, “Working with the Cisco IOS File System, Configuration Files, and Software Images.”

Additions to the Catalyst 3750 Switch Command Reference

The display for the show controllers ethernet-controller command was enhanced to show the XENPAK module serial EEPROM contents. For information about the EEPROM map and the field descriptions for the display, refer to the XENPAK multisource agreement (MSA) at these URLs:

http://www.xenpak.org/MSA/XENPAK_MSA_R2.1.pdf

http://www.xenpak.org/MSA/XENPAK_MSA_R3.0.pdf

To determine which version of the XENPAK documentation to read, check the XENPAK MSA Version supported field in the display. Version 2.1 is 15 hexadecimal, and Version 3.0 is 1e hexadecimal.

This is an example of output from the show controllers ethernet-controller tengigabitethernet1/0/1 phy command for the 10-Gigabit Ethernet interface:

Correction to the Catalyst 3750, 3560, and 2970 Switch Hardware Installation Guides

This is a new step for the “Configuring the Switch Settings” section in the “Using Express Setup” chapter:

Step 2

Enter a VLAN ID in the Management Interface (VLAN ID) field. This is the management interface through which you manage the switch and to which you assign IP information. The Management Interface field displays 1 by default. The VLAN ID range for this field is 1 to 1001.

Cisco.com

You can access the most current Cisco documentation at this URL:

http://www.cisco.com/univercd/home/home.htm

You can access the Cisco website at this URL:

http://www.cisco.com

You can access international Cisco websites at this URL:

http://www.cisco.com/public/countries_languages.shtml

Ordering Documentation

You can find instructions for ordering documentation at this URL:

http://www.cisco.com/univercd/cc/td/doc/es_inpck/pdi.htm

You can order Cisco documentation in these ways:

• Registered Cisco.com users (Cisco direct customers) can order Cisco product documentation from the Ordering tool:

http://www.cisco.com/en/US/partner/ordering/index.shtml

• Nonregistered Cisco.com users can order documentation through a local account representative by calling Cisco Systems Corporate Headquarters (California, USA) at 408 526-7208 or, elsewhere in North America, by calling 800 553-NETS (6387).

Cisco Technical Support Website

The Cisco Technical Support Website provides online documents and tools for troubleshooting and resolving technical issues with Cisco products and technologies. The website is available 24 hours a day, 365 days a year at this URL:

http://www.cisco.com/techsupport

Access to all tools on the Cisco Technical Support Website requires a Cisco.com user ID and password. If you have a valid service contract but do not have a user ID or password, you can register at this URL:

http://tools.cisco.com/RPF/register/register.do

Submitting a Service Request

Using the online TAC Service Request Tool is the fastest way to open S3 and S4 service requests. (S3 and S4 service requests are those in which your network is minimally impaired or for which you require product information.) After you describe your situation, the TAC Service Request Tool automatically provides recommended solutions. If your issue is not resolved using the recommended resources, your service request will be assigned to a Cisco TAC engineer. The TAC Service Request Tool is located at this URL:

http://www.cisco.com/techsupport/servicerequest

For S1 or S2 service requests or if you do not have Internet access, contact the Cisco TAC by telephone. (S1 or S2 service requests are those in which your production network is down or severely degraded.) Cisco TAC engineers are assigned immediately to S1 and S2 service requests to help keep your business operations running smoothly.

To open a service request by telephone, use one of the following numbers:

Asia-Pacific: +61 2 8446 7411 (Australia: 1 800 805 227)
EMEA: +32 2 704 55 55
USA: 1 800 553 2447

For a complete list of Cisco TAC contacts, go to this URL:

http://www.cisco.com/techsupport/contacts

Definitions of Service Request Severity

To ensure that all service requests are reported in a standard format, Cisco has established severity definitions.

Severity 1 (S1)—Your network is “down,” or there is a critical impact to your business operations. You and Cisco will commit all necessary resources around the clock to resolve the situation.

Severity 2 (S2)—Operation of an existing network is severely degraded, or significant aspects of your business operation are negatively affected by inadequate performance of Cisco products. You and Cisco will commit full-time resources during normal business hours to resolve the situation.

Severity 3 (S3)—Operational performance of your network is impaired, but most business operations remain functional. You and Cisco will commit resources during normal business hours to restore service to satisfactory levels.

Severity 4 (S4)—You require information or assistance with Cisco product capabilities, installation, or configuration. There is little or no effect on your business operations.

This document is to be used in conjunction with the documents listed in the “Related Documentation” section.

CCIP, CCSP, the Cisco Arrow logo, the Cisco Powered Network mark, Cisco Unity, Follow Me Browsing, FormShare, and StackWise are trademarks of Cisco Systems, Inc.; Changing the Way We Work, Live, Play, and Learn, and iQuick Study are service marks of Cisco Systems, Inc.; and Aironet, ASIST, BPX, Catalyst, CCDA, CCDP, CCIE, CCNA, CCNP, Cisco, the Cisco Certified Internetwork Expert logo, Cisco IOS, the Cisco IOS logo, Cisco Press, Cisco Systems, Cisco Systems Capital, the Cisco Systems logo, Empowering the Internet Generation, Enterprise/Solver, EtherChannel, EtherFast, EtherSwitch, Fast Step, GigaDrive, GigaStack, HomeLink, Internet Quotient, IOS, IP/TV, iQ Expertise, the iQ logo, iQ Net Readiness Scorecard, LightStream, Linksys, MeetingPlace, MGX, the Networkers logo, Networking Academy, Network Registrar, Packet, PIX, Post-Routing, Pre-Routing, ProConnect, RateMUX, Registrar, ScriptShare, SlideCast, SMARTnet, StrataView Plus, SwitchProbe, TeleRouter, The Fastest Way to Increase Your Internet Quotient, TransPath, and VCO are registered trademarks of Cisco Systems, Inc. and/or its affiliates in the United States and certain other countries.

All other trademarks mentioned in this document or Website are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (0403R)

Copyright © 2004 Cisco Systems, Inc. All rights reserved.