User manual

26
Release Notes for Catalyst 6500 Series Content Switching Module Software Release 2.2(8)
78-12569-16
Caveats
• CSCdy06651
It is not possible to configure an inband health probe to permanently disable a real server. When you
configure the health retries count failed seconds command with a failed value of zero, the CSM
should permanently disable the real server after it has first detected a failure. Instead, the CSM
immediately re-enables the server for load balancing. This caveat occurs in CSM releases 2.2(1),
2.2(2a), and 2.2(3). It is fixed in release 2.2(4) and all other CSM versions.
Workaround: Set the failed value to a very large number, for such as 100, instead of setting it to
zero.
• CSCdx91813
When you enable fail-action purge on a server farm, and a real server in that server farm goes down,
it is possible for the CSM to suspend, continue to pass traffic on existing connections, and respond
to most CLI commands. However, no new connections can be established through the device.
When this caveat is triggered by a real server going down, and a pair of CSMs is running in fault
tolerant mode, it is possible for the standby CSM to become active for seconds or minutes before it
reverts once again to the standby state. At this point, the active CSM is not able to load balance
connections. This caveat occurs in CSM releases 2.2(1), 2.2(2), and 2.2(3). It is fixed in release
2.2(4) and all other CSM versions.
Workaround: Disable fail-action purge on all server farms. If disabling fail-action purge on all
server farms is unacceptable for your configuration, you must upgrade to release 2.2.(4) or a
subsequent release.
• CSCdx78343
The connection is closed with a failure message upon receiving an RST from the client. When the
client requests an HTTP connection, the CSM establishes the connection between the client and
server. Then, if the client sends an RST, the CSM closes the connection and counts it as a failure by
increasing the count on the total conn failurs counter. This counter shows the number of failed
connections to a server. After the CSM makes a load-balancing decision, the number on the counter
is increased only if the server resets the connection or the server does not respond to the sync request
(SYN).
Workaround: None
• CSCdx77012
When you initiate an FTP session and reset the primary CSM with the hw-module module 4 reset
command, the secondary CSM takes over that session properly. If that FTP session lasts until the
primary CSM comes back online, and a preempt as active variable for that session still remains in
the secondary CSM's connection table, as shown in the sh ip slb conns command, any of the
following new sessions initiated will not be taken over by the secondary CSM when the primary
CSM attempts the next reset. If the initial FTP closes before the primary CSM fault-tolerant state is
initiated, this symptom does not occur.
Workaround: None.
• CSCdx76713
The alias IP address configuration is not removed when you enter the no module csm x or the no
vlan x command. As a result, the CSM sends an error when the same alias IP address is configured
again. This caveat occurs only in CSM release 2.2(3).
Workaround: Remove the alias address using the command no alias ip.