Table Of Contents
Release Notes for the Catalyst 6500 Series and Cisco 7600 Series Firewall Services Module Software Release 3.1(x)
This document contains release information for FWSM releases 3.1(1) through 3.1(20).
This document includes the following sections:
•You must install maintenance software Release 2.1(2) before you upgrade to FWSM Release 3.1. See Upgrading the Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module from Release 2.x to Release 3.1 for detailed information about upgrading to 2.1(2).
•For traffic that passes through the control-plane path, such as packets that require Layer 7 inspection or management traffic, the FWSM sets the maximum number of out-of-order packets that can be queued for a TCP connection to 2 packets, which is not user-configurable. Other TCP normalization features that are supported on the PIX and ASA platforms are not enabled for FWSM. You can disable the limited TCP normalization support for the FWSM using the no control-point tcp-normalizer command.
Upgrading the Software
See Upgrading the Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module from Release 2.x to Release 3.1 for detailed information about upgrading to Release 3.1.
To upgrade between 3.1(x) maintenance releases, see the "Managing Software, Licenses, and Configurations" chapter in the Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module Configuration Guide.
Note Due to CSCse74946, hitless upgrades using failover between 3.1(1) and other 3.1(x) maintenance releases are not supported. Only 3.1(1) is affected.
Chassis System Requirements
The switch models that support the FWSM include the following platforms:
•Catalyst 6500 series switches, with the following required components:
–Supervisor engine with Cisco IOS software (known as supervisor IOS) or Catalyst operating system (OS). See Table 1 for supported supervisor engine and software releases.
–MSFC 2 with Cisco IOS software. See Table 1 for supported Cisco IOS releases.
•Cisco 7600 series routers, with the following required components:
–Supervisor engine with Cisco IOS software. See Table 1 for supported supervisor engine and software releases.
–MSFC 2 with Cisco IOS software. See Table 1 for supported Cisco IOS releases.
Note The FWSM does not support a direct connection to a switch WAN port because WAN ports do not use static VLANs. However, the WAN port can connect to the MSFC, which can connect to the FWSM.
Table 1 shows the supervisor engine version and software. Please also consult and check the switch software requirements.
Table 1 Support for FWSM 3.1
Supervisor Engines1 Cisco IOS
12.2(33)SRD6 (for the Cisco 7600 series router)
720-3C-1GE (No PISA integration, no Route Health Injection, no Virtual Switching system)
12.2(18)SXF and higher
12.2(18)SXF2 and higher
2, 720, 32
Cisco IOS Software Modularity
8.5(3) and higher
2. 720, 32
1 The FWSM does not support the supervisor 1 or 1A.
2 When you use Catalyst OS on the supervisor, you can use any of the supported Cisco IOS releases above on the MSFC. (When you use Cisco IOS software on the supervisor, you use the same release on the MSFC.)
The FWSM supports the following management methods:
•Cisco ASDM—Software Release 5.0F supports FWSM software Release 3.1 features. ASDM is a browser-based configuration tool that resides on the FWSM. The system administrator can configure multiple security contexts. If desired, individual context administrators can configure only their contexts.
•Command-line interface (CLI)—Access the CLI by sessioning from the switch or by connecting to the FWSM over the network using Telnet or SSH. The FWSM does not have its own external console port.
Table 2 lists the new features for FWSM software Release 3.1(1).
Software License Information
The FWSM supports the following licensed features:
•Multiple security contexts. The FWSM supports two virtual contexts plus one admin context for a total of three security contexts without a license. For more than three contexts, obtain one of the following licenses:
Limitations and Restrictions
See the following limitations and restrictions on the FWSM:
•Multiple context mode does not support dynamic routing protocols such as RIP and OSPF. Use static routing instead.
•Transparent firewall mode supports a maximum of eight interface pairs per context; however, when multiple bridge-group interfaces exist in a single context, inspection may not work properly. We recommend that you create a separate context for traffic that requires inspection.
•For transparent firewall mode, you must configure a management IP address per interface pair.
•The outbound connections (from a higher security interface to a lower security interface) from an interface that is shared between the contexts can only be classified and directed through the correct context if you configure a static translation for the destination IP address. This limitation makes cascading contexts unsupported, because configuring the static translations for all the outside hosts is not feasible.
•The CPU-intensive commands, such as copy running-config startup-config (the same as the write memory command), might affect system performance, including reducing the successful rate of inspection and AAA connections. When a CPU-intensive action completes, the FWSM might produce a burst of traffic to catch up. If you limit the resource rates for a context, the burst might unexpectedly reach the maximum rate. We recommend using these commands during low traffic periods. Other CPU-intensive actions include the show arp command, polling the FWSM with SNMP, loading a large configuration, and compiling a large access list.
•If you try to save a new configuration file with the write memory all command in the system execution space, and there is not enough space on the disk, then the error "writing disk: message" displays; the new configuration is not saved, and the FWSM removes the existing old configuration file from the disk.
Be sure to either:
–Free some space from the disk.
–Go to each context and issue the write memory command instead of saving all contexts from the system.
Open Caveats in Software Release 3.1
This section contains open caveats in the latest maintenance release.
If you are running an older release, and you need to determine the open caveats for your release, then add the caveats in this section to the resolved caveats from later releases. For example, if you are running Release 3.1(4), then you need to add the caveats in this section to the resolved caveats from 3.1(5) and later to determine the complete list of open caveats.
When multicast routing is enabled and multicast packets are forwarded by the FWSM, forwarding statistics shown with the show mfib command are incorrect.
After removing a secondary VLAN from a firewall VLAN group on the switch, and then adding the VLAN to another group, the first VLAN group cannot be added to the FWSM, and a warning message such as the following appears:Secondary vlan 339 can't be configured as secure for module 9. Command rejected.
Windows Messenger Version 5.0 or 5.1 does not sign on with Live Communication Server 2003, Live Communication Server 2005, or any other SIP application that multiple SIP messages within the same packet; the packets are dropped. Cisco IP Phones that run SIP are not affected by this caveat.
Workaround: Configure your SIP applications to send smaller SIP messages, or increase the MTU on the FWSM interface using the mtu command if it was previously configured with a smaller than default MTU. The default MTU is 1500 bytes.
With bidirectional PIM, if the router that is configured as the RP is directly connected to the FWSM, no joins are sent to the RP by the FWSM. The debug logs show the following error message: "NO RPF NEIGHBOR o send J/P." The show mroute and show mfib commands display correct flags and RPF neighbors.
Workaround: Do not make the directly-connected router the RP.
URL filtering with Websense causes high CPU in high traffic loads.
High CPU is seen when a client accesses an ISEE server (sPOP) and HTTP inspection is enabled.
Workaround: Disable the tcp normalizer using the no control-point tcp-normalizer command or disable HTTP inspection.
The FWSM is not forwarding all DNS requests from the outside interface to the inside interface when the inspect dns max-length command is used.
Workaround: Disable the inspect dns max-length command.
Connections that have their TCP state bypassed (using the set connection advanced-options tcp-state-bypass command) generate SYN Timeout syslog messages when they idle out. The TCP SYN packets do indeed pass through the FWSM, but the syslog message indicates the tear down reason as a SYN timeout.
For example:Teardown TCP connection 13223832 for outside:10.10.10.100/1304 to inside:192.168.1.100/1234 duration 2:02:53 bytes 7798136 SYN Timeout
Also the connection flags for a connection with its TCP state bypassed indicate one of the following groups of flags:bBs - (b)State bypass, (B)initial SYN from outside, (s)awaiting outside SYNbBS - (b)State bypass, (B)initial SYN from outside, (s)awaiting inside SYNbs - (b)State bypass, (s)awaiting outside SYNbS - (b)State bypass, (s)awaiting inside SYN
Because the FWSM is not tracking the state of the connection, flags indicating the direction of traffic and whether or not correct SYN packets were received, may be inconsistent and misleading.
Directed BOOTP messages are redirected to a DHCP server if DHCP Relay is enabled on the FWSM and DHCP Relay servers are configured.
The primary and secondary FWSMs might crash in Thread name: snmp. This is caused when there is no proper response from the NP due to high traffic. Also there is no snmp-server host command configured in the system but in the configuration, there is the snmp-server enable traps snmp authentication linkup linkdown coldstart command.
Workaround: Remove the snmp-server traps command.
In single mode, using FTP with inspect ftp enabled results in a 10% drop in connections per second handled by the FWSM. Once a connection is established, data traffic does not experience any drop.
Failover is not working on the FWSM in transparent mode. When connectivity is broken on one or two interfaces, The FWSM is not updating the MAC address with the updated path. Therefore, users are losing their connections.
In rare circumstances, traffic matching a static policy NAT statement may fail with a "no translation group found" syslog message even though it matches the policy access list.
Workaround: Try redefining the policy access list with a different access list name and applying that to the static.
Outside policy PAT in multiple context mode uses an inactive access list to create xlates after the memory partition of the context is changed using the allocate acl-partition command.
Workaround: Reconfigure the access list and policy PAT after changing the memory partition.
When outside policy PAT is configured and traffic is sent from outside to inside host, then xlates on a standby unit have incorrect flags of Identity (I) instead of portmap (r) and shows the xlate as NAT instead of PAT.
The FWSM in multiple context mode might experience a depletion in the 16384 byte blocks if multiple contexts are subjected to SNMP polling simultaneously. Once in this condition, the FWSM must be rebooted to recover.
To detect if the FWSM is in this state, enter show blocks and look for the line starting with "Slow Path". If the CNT column is at 0 and stays at 0, this issue might be the cause.
For example:FWSM# show blocksSIZE MAX LOW CNT4 1800 1790 180080 1000 976 983256 1600 1529 15861550 11575 10483 115402048 1384 1349 138316384 8192 2181 2182Additional Block pools for 16384 size blocksIP Stack 1024 1023 1024ARP Stack 512 510 512Slow Path 5500 0 0 <--- Problem hereNP-CP 1024 1017 1024Others 132 132 132
Additionally, the output of show blocks old | begin 16384 will show output relating to SNMP.
For example:FWSM# show blocks old | b 16384Class 8, size 16384Block allocd_by freed_by data size alloccnt dup_cnt oper location0x0a7f0aa0 0x00411557 0x00a30608 44 101 0 put udp_usr_input/ifc:65535/snmp0x0a7ec780 0x00411557 0x00a30608 39 123 0 put udp_usr_input/ifc:65535/snmp0x0a7e8460 0x00411557 0x00a30608 39 132 0 put udp_usr_input/ifc:65535/snmp0x0a7e4140 0x00411557 0x00a30608 39 128 0 put udp_usr_input/ifc:65535/snmp0x0a7dfe20 0x00411557 0x00a30608 39 85 0 put udp_usr_input/ifc:65535/snmp0x0a7dbb00 0x00411557 0x00a30608 44 100 0 put udp_usr_input/ifc:65535/snmp0x0a7d77e0 0x00411557 0x0041dcc5 39 123 0 put udp_usr_input/ifc:65535/snmp...
Workaround: Configure the SNMP management server to not query the following OIDs:
This section lists the resolved caveats for each maintenance release and includes the following topics:
Resolved Caveats in Software Release 3.1(20)
The following caveats were resolved in Release 3.1(20) and were not previously documented. If you are a registered Cisco.com user, you can view more information about the caveat using the Bug Toolkit at the following website:
•CSCtk61424 — OpenSSL Ciphersuite Downgrade and J-PAKE Issues
The device may be affected by an OpenSSL vulnerabilities described in CVE-2010-4180 and CVE-2010-4252.
Device configured with any feature that uses SSL.
The Cisco PSIRT has assigned this bug the following CVSS version 2 score. The Base and Temporal CVSS scores as of the time of evaluation are 5.1/3.8:
CVE IDs CVE-2010-4180 and CVE-2010-4252 have been assigned to document this issue.
Additional information on Cisco's security vulnerability policy can be found at the following URL:
•CSCtl21186 — Cmd authorization fails for certain commands on fallback to LOCAL db
Certain commands like 'show running-config', 'show interface' are allowed to be executed by users with lower privilege-level when fallback has occured.
1. Fallback to LOCAL is configured
2. All FWSM commands are assigned their default privilege levels in LOCAL db.
3. Users with lower privilege-level than 15 login into privileged-exec mode and execute 'show running-config' or 'show interface' commands, and some config commands.
The Cisco PSIRT has assigned this bug the following CVSS version 2
score. The Base and Temporal CVSS scores as of the time of evaluation are 6.0/5.0:
CVE ID CSCtl94142 has been assigned to document this issue.
Additional information on Cisco's security vulnerability policy can befound at the following URL:
•CSCtl84952 — SCCP inspection DoS vulnerability
A vulnerability exists in the Cisco Firewall Services Module (FWSM) for Cisco Catalyst 6500 Series Switches and Cisco 7600 Series Routers that may cause the Cisco FWSM to reload after processing a malformed Skinny Client Control Protocol (SCCP) message. Devices are affected when SCCP inspection is enabled.
Cisco has released free software updates that address this vulnerability.
This advisory is posted at
Note: Cisco ASA 5500 Series Adaptive Security Appliances are affected by the vulnerability described in this advisory. A separate Cisco Security Advisory has been published to disclose this and other vulnerabilities that affect the Cisco ASA 5500 Series Adaptive Security Appliances. The advisory is available at
Resolved Caveats in Software Release 3.1(19)
The following caveats were resolved in Release 3.1(19) and were not previously documented. If you are a registered Cisco.com user, you can view more information about the caveat using the Bug Toolkit at the following website:
Table 3 Resolved Caveats in Release 3.1(19)
FWSM 3.2 - deny-flow-max stuck when denied is not at 4096
term pager command affects all sessions and future sessions
Resolved Caveats in Software Release 3.1(18)
The following caveat was resolved in Release 3.1(18), and was not previously documented. If you are a registered Cisco.com user, view more information about the caveat using the Bug Toolkit at the following website:
•CSCte49110 : FWSM setting DF bit on reassembled skinny packet
Resolved Caveats in Software Release 3.1(17)
The caveats in Table 4 were resolved in Release 3.1(17), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(16)
In failover, the standby FWSM becomes unresponsive with Thread Name: fover_ifc_test.
The caveats in Table 5 were resolved in Release 3.1(16), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(15)
The FWSM might unexpectedly stop passing traffic and reload. The output of the show crash command shows a traceback in thread "doorbell_poll". The NP Hard Debu in the NP Hard Assert Info (included in the show crash output shows a crash in processor NP1 or NP2 at PC 0x3a1a.
The caveats in Table 6 were resolved in Release 3.1(15), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(14)
The caveats in Table 7 were resolved in Release 3.1(14), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(13)
The FWSM reloads when a Cisco ASA 5500 series adaptive security appliance is configured as its failover peer and placed on the respective failover control VLAN.
Workaround: Disallow the FWSM failover control VLAN on all trunks and access ports of the switch, or configure a failover key.
The caveats in Table 8 were resolved in Release 3.1(13), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(12)
If you have two dynamic policy NAT commands, and traffic matches the access list in one of the NAT commands; then you change the access list in the other NAT command so there is an overlapping ACE that also matches the same traffic; then no NAT entries are created for that traffic.
Workaround: Remove and reapply the unchanged NAT statement (the NAT statement that was formerly used to match the traffic). This change forces the other NAT pool (with the updated access list) to take effect.
If you have overlapping static commands that both match the same traffic, and you add an ACE using the line keyword to an access list being used by the higher priority static command, then any traffic that should use the higher priority static command now uses the lower priority static command.
Workaround: Remove and readd the static command after you alter the access list.
The FWSM may experience a failover event or stop responding completely after an extensive series of ICMP Echo Request packets is generated either to the FWSM or from the FWSM command line interface.
In multiple context mode with Failover, the secondary FWSM might crash after you commit configuration changes on the primary unit. After the crash, reloading the secondary FWSM causes it to enter Failover Off (pseudo-Standby) state. Both units have to be reloaded to re-establish the failover pair.
The Maximum Segment Size (MSS) option in the TCP header in the SYN ACK segment is passed unchanged when traversing the FWSM, regardless of what is configured with the sysopt connection tcpmss command. The MSS option on the initial TCP SYN segment is adjusted correctly. This occurs when the TCP options length is small (8 bytes or so).
The caveats in Table 9 were resolved in Release 3.1(12), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(11)
The syslog message 305005 (No translation group found for...) should be generated for packets dropped due to a missing outside NAT exemption rule, but it is not. When outside NAT is configured along with nat-control enabled, all traffic not included in the outside NAT configuration must be included in an outside NAT exemption rule. If not, it is the expected behavior that these packets are dropped.
The FWSM might unexpectedly stop passing traffic and reload. The output of the show crash command shows a traceback in thread "doorbell_poll". The NP Hard Debu in the NP Hard Assert Info (included in the show crash output shows a crash in processor NP1 or NP2 at PC 0x59c2.
The caveats in Table 10 were resolved in Release 3.1(11), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(10)
The FTP client/server does not close a connection in some cases when the server uses a multiline 221 closure sequence:221-You have transferred 0 bytes in 0 files.221-Total traffic for this session was 2551 bytes in 1 transfers.221-Thank you for using the FTP service on orbi.221 Goodbye.
instead of the classic sequence:221 Goodbye;
Workaround: Disable FTP inspection or disable the 221 multiline closure sequence.
The FWSM logs syslog message #305006 ("<...> translation creation failed") even when sufficient NAT and/or PAT resources are available. This message occurs when the FWSM has a high NAT or PAT xlate reuse rate.
Workaround: Increase the NAT and/or PAT pool or reload the FWSM to temporary clear the condition.
If you use an access list for static policy NAT and then insert an ACE in the access list; and the access list includes another ACE lower down (at a higher number) that can match the same traffic as the new ACE; then traffic that should match the new ACE because it is hit first instead matches the older ACE at the higher line number.
Workaround: Finalize the access list configuration before attaching it to the static policy NAT command.
The FWSM might crash in Thread Name: websns_rcv_udp when Websense filtering is configured.
The FWSM might crash when RIP is running. The crash shows: Thread Name: route_process (Old pc 0x00bbf8b6 ebp 0x0a5fe764)
After failover, the inspect ftp feature does not work; the data channel is not opened on the first FTP connection attempt. However, the connection does go through on the second try.
Workaround: Retry your FTP attempt, and the connection succeeds.
When configuring outside policy NAT in conjunction with outside NAT exemption, the policy NAT is never applied as configured. Even though the flow is excluded from the NAT exemption by configuring a deny ACE, a dynamic identity xlate is built for the outside source. All traffic is NAT exempted.
For example:global (inside) 5 10.10.10.50-10.10.10.60nat (outside) 0 access-list nonat outsidenat (outside) 5 access-list nat outsideaccess-list nonat extended deny ip host 192.168.49.57 host 172.16.10.1access-list nonat extended permit ip any anyaccess-list nat extended permit ip host 192.168.49.57 host 172.16.10.1
192.168.49.57 should be translated to the global pool, but it is not.
Workaround: The outside NAT exemption is only required when the nat-control command is enabled. If you disable NAT control (no nat-control) then you can remove the outside NAT exemption command.
The caveats in Table 11 were resolved in Release 3.1(10), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(9)
If an existing BVI interface is remove and then re-added, the interface status shown by the show interface bvi command is seen as "administratively down" with a protocol status of "up" instead of the actual "up" and "up" status. The show interface ip brief command shows the status as "administratively down" with a protocol status of "down" instead of the actual "up" and "up" status.
The functionality of the interface is not affected.
Workaround: Use a bridge group number other than one which was removed. The interface status shows correctly after you reload the FWSM.
The wrong TLV parameters are received by the FP when a TLV update has a wrong field (the function ID is out of the range). This situation causes the FP to assert and generate a crash (door_bell pool).
When the last batch of commands committed includes inspection rules, and the new rules caused memory exhaustion, then the new rules are not automatically removed from the configuration even though they exceed the rule limit causing other rules not to load correctly.
Workaround: Remove the last batch of inspection rules from the current configuration.
When both the client and the server agreed with the use of the TCP window scale option, then the FWSM:
a. Does override the MSS of the client (in the first SYN).
b. But does not override the MSS of the server (in the SYN,ACK).
Workaround: Disable TCP Window Scaling on either the server or on the client.
HTTP authentication with the ssl trust-point command is not working after you reload the FWSM. The CA certificate imported is not used after the reload. The following syslog message displays:%FWSM-3-717023: SSL failed to set device certificate for trustpoint <trustpoint>. Reason: No device certificate found.
Workaround: Perform the following steps:
a. Enter the no crypto ca trustpoint trustpoint command.
b. Reimport the CA certificate.
c. Enter the ssl trust-point trustpoint command.
The failover interfaces are not reported through SNMP. Snmpwalk shows all interfaces, except the failover ones.
Packets might be passed by the standby FWSM in a failover pair during the short period of time that the FWSM is syncing just after booting.
NAT exemption is not used for communication between same-security-level interfaces when you have other NAT types configured that match the traffic; NAT exemption is supposed to take priority over other NAT types.
Workaround: Define a policy NAT statement to exclude hosts you wish to exempt.
If you enable SNMP traps when upgrading from 2.3(4) to 3.1(8), then the FWSM might experience a software-forced reload.
URL-filtering-denied traffic is unsuccessfully closed; you can see the dropped traffic using the show asp drop command.
Workaround: Disable the TCP normalizer by entering the no control-point tcp-normalizer command.
The caveats in Table 12 were resolved in Release 3.1(9), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(8)
When the configured connection limit (set connection conn-max) is exceeded, the port number shown in system message 201011 is shown in network-byte-order, not host-byte-order. For example, the following system message has the port number as shown:%FWSM-3-201011: Connection limit exceeded 50/50 for inbound packet from x.x.x.x/260 to y.y.y.y/17664 on interface outside
The real port numbers in this example are 1025 and 69.
Workaround: Convert the port numbers using the following calculation:
a. Convert the system message port number to hexadecimal. For example:
260 is 0x0104 in hexadecimal.
17664 is 0x4500 in hexadecimal.
b. Exchange the hexadecimal byte pairs. For example:
0x0104 exchanged is 0x0401.
0x4500 exchanged is 0x0045.
c. Convert the exchanged hexadecimal number to decimal to get the true port number. For example:
0x0401 is 1025 in decimal.
0x0045 is 69 in decimal.
The show memory detail command indicates 399% or 400% for the used memory in the admin context:hostname# changeto context adminhostname/admin#hostname/admin# show mem detailUsed memory: 4294561916 bytes (400%)------------- ----------------Total memory: 1073741824 bytes (100%)Most used memory: - 36676 bytes (400%)
If you configure URL filtering for HTTPS, then HTTPS sessions are subject to URL filtering in both the outbound direction (high security to low security interface), which is expected, and the inbound direction (low security to high security interface), which is not expected. For HTTP and FTP, only outbound connections are filtered.
When traffic hits an ACE while swapping the ACE order, the access list logging stops. For example, after swapping the ACEs of the below access list:access-list vbug extended permit ip host 10.1.1.2 host 10.0.0.100 log interval 10access-list vbug extended deny ip host 10.1.1.5 host 10.0.0.100 log interval 10
To:access-list vbug extended deny ip host 10.1.1.5 host 10.0.0.100 log interval 10access-list vbug extended permit ip host 10.1.1.2 host 10.0.0.100 log interval 10
Logs for the permit ACE stop showing up on the console.
Workaround: Stop the traffic, remove the access list, reconfigure it, and reapply.
When traffic matches an ACE, a system log message is generated in the syslog even though logging has been disabled for this ACE. For example:hostname(config)# access-list outside_in line 16 extended deny tcp host 192.168.120.103 host 172.16.1.28 eq https log disableMar 09 2007 18:35:07 VFW1 : %FWSM-1-106100: access-list outside_in denied tcp outside/192.168.120.103(32365) -> DMZ2/172.16.1.28(443) hit-cnt 1 (first hit) [0x1a9ac098, 0x24cf570]
Free memory on an FWSM slowly decreases over time until no free memory is available, leading to an outage. H323 RAS inspection must be enabled and non-H323 traffic on UDP/1718 and UDP/1719 must be present. This traffic will be dropped by the inspection because it is not H323 RAS traffic.
Workaround: Disable H323 RAS inspection. If this breaks H323 functionality, continuously monitor memory consumption on the FWSM and reload the FWSM when a critical level is reached.
You can verify the drops by looking at the output of the show service-policy command.
When the ICMP inspection is not enabled, if the FWSM could not route the packet from a low security source host to a high security destination host, it sends an ICMP network unreachable error back to the source host with the real IP address of the destination tried, instead of the mapped address. Also, a traceroute from a low security interface to a high security interface returns the real IP address of the destination to the source host.
Workaround: Configure ICMP inspection by entering the following commands:policy-map global_policyclass inspection_defaultinspect icmp
You cannot delete counters of all access lists by using the clear access-list counters command.
Workaround: You can only delete counters of access lists individually using the clear access-list id counters command.
When using the config net command on the FWSM to copy a configuration from a TFTP server to the running configuration, requests with long file names (more than 56 characters) fail or produce unexpected results.
For example:config net 192.168.1.100:configurations/filename
where filename is longer than 56 characters.
Workaround: Use shorter configuration filenames.
If you reload the FWSM and you are prompted to save the configuration, then choosing the Save All option only saves the system configuration and not the security context configurations.
Workaround: Enter the write memory all command in the system execution space before you reload.
If you have the maximum of 5 hsi-group commands in an h225-map, and you remove one or more groups, then you cannot add a new hsi-group command or edit an existing one.
Workaround: You must remove the whole h225-map and create a new one.
Changing an interface name causes a memory leak on active and standby FWSMs.
In manual commit mode, inactive access lists remain active after they are committed.
Workaround: Use auto-commit mode.
A lot of packets are exchanged between the FWSM and a host in a matter of milliseconds if out-of-order packets arrive on the FWSM in some situations. This situation occurs when the TCP sequence number of a flow changes on either side of the FWSM due to a change in the data payload when NAT is configured.
Workaround: Do not configure NAT.
If you have a failover pair, and you have to replace or redeploy the standby unit, then when you start up the new unit, failover becomes enabled even if the failover VLAN is not yet assigned from the switch, or if the switch is slow in assigning the VLAN. Therefore, the failover link is not active between the two units.
This situation can cause a number of problems; for example, the new unit becomes active, so you have two active units on the network. Then, when the VLAN is finally assigned to the new unit, if the new unit is designated as the primary, it forces the secondary unit to become standby. At best, this causes an unnecessary failover event, and at worst the primary unit becomes active before you have finished configuring it; the good configuration on the secondary unit gets overwritten by the bad configuration on the primary.
Workaround: Always assign the failover VLAN to both units before you enable failover. Also, be sure to have a complete configuration on the primary unit before you deploy it, even if it is the standby unit.
If you configure the half-closed timeout using the set connection timeout half-closed command in a policy map, then the half-closed connections do not get cleared.
Workaround: Set the global timeout using the timeout half-closed command. This timeout affects all connections, and not just those specified in the class map.
Authentication fails for an HTTPS session if the destination configured in an access list for AAA is a name that you associated with the IP address using the name command.
Workaround: Do not use the name in the access list; instead use the actual IP address.
The caveats in Table 13 were resolved in Release 3.1(8), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(7)
Currently, captures from the FWSM can only be retrieved from the FWSM using HTTPS in single context mode. In multiple context mode, another method must be used to transfer the captures of the FWSM.
Workaround: Use TFTP to transfer the captures from the FWSM in the system execution space.
Note The FWSM uses the IP address of the admin context to transfer the files, but the command is only valid in the system execution space.
Malformed MGCP packets cause an unexpected reload of the FWSM when the MGCP inspection engine is enabled.
The FWSM can experience a problem processing the FTP PORT command for active FTP sessions (when FTP inspection is enabled). This can prevent FTP data streams from flowing properly through the FWSM and prevent file transfer.
Workaround: Disable the TCP normalizer using the no control-point tcp-normalizer command (introduced in version 3.1.4(9))
This command has side effects: some voice-related inspections do not work properly after you disable TCP normalization, including H.323 and Skinny.
The FWSM sends system log message 106101 ("The number of ACL log deny-flows has reached limit number") when the total amount of deny flows is far beneath the configured limit.
Workaround: Disable the log option in the access-list entries.
After running a script that issues show commands, there was a memory loss of approximately 155 KB. When running the same script for just show running-config commands, there was a loss of approximately 100 KB of memory that was not recovered.
A capture shows extra packets while capturing fragments. When the FWSM receives fragments, the accelerated path sends them to the session management path. The session management path collects these fragments, and after the final fragment is received, it reassembles them and sends the complete packet back again to the accelerated path for processing. Currently we capture this packet as well.
The number of xlate entries shown in the standby unit is greater than the system limitation, which is 256 K entries. This problem is seen in standby unit when FWSM is configured to process outside PAT.
In manual mode, access list logging does not work if you replace an ACE without logging enabled with an ACE that has logging enabled.
Workaround: Use auto mode, or remove the existing ACE and then reapply the ACE with logging.
A client can successfully Telnet to the virtual HTTP IP address and get authenticated. Similarly, a client can successfully HTTP to the virtual Telnet IP address and get authenticated.
The warning keyword is not configurable with the virtual http command.
The caveats in Table 14 were resolved in Release 3.1(7), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(6)
In the show conn command output, the connection counter for "in most used" is incorrect. This happens when the FWSM is configured with a url-server with a large number of TCP connections. For example, url-server (inside) host 188.8.131.52 pro tcp conn 50.
Under a high load of SunRPC traffic, the standby unit might experience a crash after failover or after configuration replication.
Capture does not work properly with fragmented packets.
The packet capture tool does not capture all packets from the TCP session initialization sequence. When capture is configured to capture all TCP packets, for example, using a permit tcp any any access list, the capture buffer shows only the first SYN and the last ACK packets of the TCP session initialization. The buffer does not have the SYN_ACK packet.
An access list of approximately 74 K ACEs with 151 K nodes takes more then 90 minutes to install and compile on the FWSM without any traffic.
The FWSM resets with a show crash traceback of thread name doorbell poll.
When IPv6 capture is defined for an interface, all IPv6 packets on that particular interface are captured, irrespective of the access list attached to the capture.
You cannot reduce the circular buffer size for the capture command, once configured.
Workaround: Remove the capture command, and reconfigure the capture again.
The OSPF neighbor goes down after changing the interface IPv6 setting. The FWSM sends OSPF hellos, but debugging does not show receiving OSPF hellos.
A failure when compiling rules can cause some previously compiled rules to be removed under some special conditions. The previously compiled rules might get deleted from the configuration when there is any subsequent failed compilation of rules after more than 255 times of previous access list rule compilation.
The current Compilation ID and number of time of rule compilation can be seen by entering the debug acl download command before updating the rules.
Workaround: Reapply the previously-configured rules, including the startup configuration rules.
When the FWSM receives more numbers of packets than what it can hold in the capture buffer, the FWSM stops capturing packets for the linear buffer, but for the circular buffer, the capture continues by flushing out the earlier captured packets. So for the circular buffer, the total capture count will keep incrementing, but the number of packets shown will be limited by the buffer size.
The FWSM does not print the following system log message when the wrong credentials are entered during AAA authentication for virtual SSH with ACS (RADIUS or TACACS+):%FWSM-6-611102: User authentication failed: Uname: user
When a client attempts a traceroute through the FWSM, the ICMP error inspection engine does not allow the TTL expired messages to return to the client.
Under rare circumstances, multicast packets processed by the FWSM might cause the FWSM to build incorrect translations in its translation table. This situation then results in the FWSM dropping TCP and UDP traffic sourced from the host that sent the multicast traffic.
All of the following conditions must be met:
–Multicast routing must be enabled on the FWSM.
–NAT control must be disabled on the FWSM.
–The xlate that is built is considered an inside xlate (show xlate detail shows the flag 'i')
Workaround: To clear the incorrect translation, enter the clear xlate local x command where x is the IP address of the host that sent the multicast traffic and can no longer communicate through the FWSM.
See the following example.
Sample incorrect translation (show xlate debug):NAT from outside:192.168.1.100 to inside:192.168.1.100 flags si idle 0:00:55 timeout 0:01:00 connections 1
This translation shows that the host 192.168.1.100 is off of the outside interface of the FWSM, but the flags for the xlate show "si" indicating that the FWSM considers this an inside xlate, which is incorrect (this should be an outside xlate).
To clear the translation, enter:hostname# clear xlate local 192.168.1.100
Then, to prevent the incorrect translation from forming again do one of the following:
–Disable multicast routing on the FWSM (or disable IGMP and PIM on a per-interface basis).
–Apply an inbound access list on the interface where the host resides that blocks all traffic destined to the particular multicast address.
Capture of packets fails for a VLAN when a context that shares that VLAN is removed.
For example, R1 and R3 contexts have a shared VLAN 200.
Ping from outside host A to inside host B.
When capture is enabled only in R1, inside1 and outside1 interfaces, both interfaces are able to capture packets.
Now, enable capture in R3 in the shared VLAN. For a shared VLAN (VLAN 200 shared between R1 and R3), if you define a capture on both contexts, packet capture will happen only in the context where the capture was configured last. As expected, you should be able to see packet capture in inside3 and not inside1.
If you remove R3, and go back to R1 and configure capture again, R1 inside1 should be able to capture packets now that the VLAN is not shared, but capture fails.
Workaround: Disable and reenable capture.
The caveats in Table 15 were resolved in Release 3.1(6), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(5)
In multiple context mode with each context assigned to a failover group (join-failover-group) for Active/Active failover, if you disable failover in the system configuration (no failover) at startup, then the FWSM drops traffic.
Workaround: Remove the join-failover-group command from each context in the system configuration, or enable failover.
If you commit a very large access list (approximately 74 K rules) either automatically or manually, the FWSM takes approximately 2 hours to commit it, and you see a fatal error message even though it successfully committed:**** FATAL ERROR: Access Rule Download Failed ****
The traffic does not pass through the FWSM before and after receiving the fatal error.
No error message is generated when an access list associated with the aaa authentication match command is removed; all aaa authentication match commands are removed when you delete the associated access list.
The caveats in Table 16 were resolved in Release 3.1(5), and were not previously documented. If you are a registered Cisco.com user, view more information about each caveat using the Bug Toolkit at the following website:
Resolved Caveats in Software Release 3.1(4)
If an HTTP request or response packet with an invalid minor version string passes through the FWSM, the FWSM fails to log or deny the packet.
In manual commit mode, if you repeatedly clear an access list, you cannot add additional ACEs to the access list and it remains empty. For example, if you enter clear configure access-list name, then add an ACE, then clear it again, then you cannot add another ACE to the access list. You see the following error:ERROR: Unable to add, access-list config limit reached
Now, if you commit the access list, it will be removed along with any associated access-group commands.
When a static command is configured with a network mask, and an inbound ICMP packet is sent to the network IP address, the FWSM builds a static translation instead of generating system log message 305006.
Communication between the inside and outside interfaces of a context fails after adding the below two commands to the system configuration that does not yet have any other failover commands:failover group ncontext xxxxjoin failover group n
Workaround: Configure the below minimum failover commands:failover lan unit primaryfailover lan interface name vlannumberfailover interface ip name ipaddress mask standby ipaddressfailover
The FWSM might crash unexpectedly in Thread Name: SNMP.
After adding and removing an ActiveX or Java filter for any port, the original filter stops working.
Workaround: Enter the clear configure filter command, and then reconfigure the filter.
The limit-resource all command cannot be configured. This can lead to one context hogging the CPU and causing connectivity problems during the implementation of changes.
If you upgraded to Release 3.1, then the snmp-server listen-port command in the startup configuration in multiple context mode might cause the FWSM to crash if traffic is present when the FWSM boots.
Workaround: Boot into the maintenance partition and remove the startup configuration. Remove the snmp-server listen-port command from the startup configuration before copying it back to the FWSM.
Using a large number (10 K) of time range ACEs may cause the system to become unstable and crash.
When a time range applied to an ACE with an object-group becomes inactive, the ACE is still active and traffic passes through.
Modifications to an access list sometimes leads to a misordering of ACEs in the access list, which leads to incorrect access list filtering. This situation only occurs if an ACE includes an object group.
The FWSM changes the UDP checksum on non-NAT interfaces when the RP router is set downstream with respect to the FWSM.
Workaround: Use the nat command instead of the static command, or move the RP router upstream.
The FWSM might experience a memory leak after running traffic for 72 hours. You see the following error message:ERROR: Failed to allocate memory for show Conn request
When enabling URL filtering in multiple context mode, URL lookup requests are sent to the filtering server. Under certain circumstances, the FWSM delays these filtering requests so that web access performance is diminished.
Workaround: Disable URL filtering to restore regular web access performance.
When the FWSM is configured with a chain limit using the fragment chain command and traffic is sent with large size data, the FWSM should show the message "Discard IP fragment set with more than n elements", but the logs shows an incorrect value.
If the FWSM in failover ends up in pseudo-standby mode, it uses its own MAC address with active IP addresses; in which case, there will be two units using the same IP addresses.
The FWSM reloads when a service policy configured for inspecting FTP, TFTP, and HTTP (but not ICMP) is applied to an interface containing a class map that matches a large access list containing 10 K ACEs.
The FWSM reloads only if a new service policy is applied in addition to the default service policy, global_policy, which is applied globally to all interfaces by default.
The FWSM might reload on the first attempt when a service policy is applied to an interface. Sometimes on the first attempt, it gives the below error message:hostname(config)# service-policy abc interface inside21ERROR: Unable to add, fixup config limit reachedERROR: cannot add policy to rule engine
Workaround: Remove the default global_policy before applying the new service policy that uses the large access list.
When you perform an snmpwalk on the FWSM, the ifOutOctets MIB shows incorrect numbers.
UDP packets with a source port equal to zero will be dropped by the FWSM when you specify the destination port in the interface access list.
For example, the following access list allows any UDP source port but specifies the destination port of 53; the system log message shows that the packet was dropped:hostname(config)# access-list inside extended permit udp any any eq 53%FWSM-4-106023: Deny udp src inside:x.x.x.x/0 dst outside:y.y.y.y/53 by access-group "inside" [0x0, 0x0]
Workaround: Remove the destination port number in the access list and restrict access based only on protocol and IP addresses.
If you change the logging level of system log message 106100 from 6 to a lower level using the access-list command log level argument, the FWSM does not recognize the new level; you can only view the message if you show messages for level 6 or 7 (informational or debugging).
Workaround: View level 6 or 7 logging messages.
Resolved Caveats in Software Release 3.1(3)
Packet loss is experienced when a large packet needs to be fragmented by the FWSM on its way out.
Workaround: Increase the MTU size on the outgoing interface of the FWSM. For example, if the packet size is 5 K, change the MTU size on the outgoing interface of the FWSM to a larger value, like 8 K. Use the mtu interface_name bytes command to change the MTU size.
Resolved Caveats in Software Release 3.1(2)
In multiple transparent mode, if you configure a syslog server with the no logging permit-hostdown command, the FWSM fails to drop the traffic when the syslog server is unreachable.
When using Active/Active failover, if you remove an interface from a security context that is shared with other contexts (or remove the security context with a shared interface), then traffic on that VLAN might be dropped, even though the VLAN is still in use in other contexts.
If you enter the clear configure static command, and the static command was configured with the interface keyword, then the real interface IP address can be pinged from the supervisor engine. After failover, however, the real IP address cannot be pinged from the supervisor engine.
Workaround: Reconfigure the IP address on the interface.
If you enter the show np 2 vlan 4096 command on the FWSM, it crashes.
Workaround: None. Other VLAN values do not crash the FWSM.
If you use SIP IP Address Privacy in conjunction with PAT, then the FWSM fails to allow media traffic. The two inside phones register with the outside proxy, but the FWSM drops media connections.
Workaround: Use dynamic NAT instead of PAT with SIP Address Privacy.
If you modify the SIP inspection configuration, RTP traffic cannot pass through the FWSM.
Workaround: Use the default inspection policy.
In multiple transparent mode, if you enter the show asp table mac-address-table command in the system execution space, the FWSM crashes.
Using an extended ping in the system execution space over an SSH connection to the admin context hangs the session and might crash the FWSM if the session is cleared.
Workaround: Use Telnet to the admin context instead of SSH; or connect to the system execution space from the switch using the session command. You can also reduce the SSH timeout so that if it hangs, you do not have to manually clear it.
The FWSM might traceback with "Thread Name: Checkheaps" and "assertion "0" failed: file "malloc.c:, line 4578".
This occurs when a protocol using a port reserved for CTIQBE or H323 sends packets greater than 8192 bytes in size, with H323 or CTIQBE inspection enabled on the FWSM. The correct behavior is to create a system log message to indicate when the proxy buffer limit is reached during the reassembly process. When this happens for H323 and CTIQBE, the connection goes from proxy to non-proxy mode.
Workaround: This could be normal in many cases. A packet capture would help narrow down the packets causing the reassembly limit to exceed.
In rare situations, an FWSM might crash and not complete the crash process. As a result, the FWSM might not reload during the crash process as normally expected.
Workaround: There is no workaround at this time. The only option is to reload the FWSM if accessible by session or ssh or alternatively reset the FWSM from the CLI of the switch.
The packet capture feature is only capturing ingress (inbound) packets, not egress (outbound) packets.
Using outside policy NAT, all outside traffic requires NAT to pass through the FWSM even with NAT control disabled.
If you modify a global command, and then enter the clear xlate command, no translations are reestablished and traffic does not flow through the FWSM.
Workaround: Remove the current global command and re-add it.
See the following sections for related documentation:
See the following related hardware documentation:
•Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module Installation Note
•Catalyst 6500 Series Switch Installation Guide
•Catalyst 6500 Series Switch Module Installation Guide
See the following related software documentation:
•Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module Configuration Guide
•Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module Command Reference
•Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module Logging Configuration and System Log Messages
•Upgrading the Catalyst 6500 Series Switch and Cisco 7600 Series Router Firewall Services Module from Release 2.x to Release 3.1
•Catalyst 6500 Series Cisco IOS Software Configuration Guide
•Catalyst 6500 Series Cisco IOS Command Reference
Obtaining Documentation and Submitting a Service Request
For information on obtaining documentation, submitting a service request, and gathering additional information, see the monthly What's New in Cisco Product Documentation, which also lists all new and revised Cisco technical documentation:
Subscribe to the What's New in Cisco Product Documentation as an RSS feed and set content to be delivered directly to your desktop using a reader application. The RSS feeds are a free service. Cisco currently supports RSS Version 2.0.
Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: www.cisco.com/go/trademarks. Third-party trademarks mentioned 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. (1110R)
Any Internet Protocol (IP) addresses used in this document are not intended to be actual addresses. Any examples, command display output, and figures included in the document are shown for illustrative purposes only. Any use of actual IP addresses in illustrative content is unintentional and coincidental.
© 2011 Cisco Systems, Inc. All rights reserved.