HP Serviceguard for Linux Version A.11.16 Release Notes, Third Edition, August 2006

Serviceguard for Linux Version A.11.16 Release Notes
Known Problems and Workarounds
Chapter 162
cmapplyconf : Unable to apply the configuration
• What is the workaround? Make certain that the ip addresses for all
interfaces are present in the /etc/hosts file. As documented for
A.11.16.01.
JAGaf45963: Cannot cmapplyconf from root from
node in cluster
• What is the problem? The resource monitor daemon, cmresmond,
opens file descriptors to agents that read the disk to determine if the
disk to be monitored is alive and working. If a “cmresmond
--reload” is issued at the same time as the file descriptor is open, a
file descriptor will be leaked. Over time, this can leak enough to hit
the open file descriptor limit. Once the file descriptor limit is
reached, unpredictable things can happen with I/O and a coredump
can occur.
• What is the workaround? Use “cmresmond --restart” instead of
“cmresmond --reload”. This will stop the resource monitor and
start it again on a different PID thus clearing any file descriptors
that were open by cmresmond.
JAGaf46362: cmviewcl shows packages as
down/unowned when they should be unknown
• What is the problem? In situations when the cluster coordinator's
cmcld died and before the TOC of the node, cmviewcl will show a
package that is actually running as “down” and “unowned”. It would
be more appropriate to display “unknown” for the status and
ownership.
• What is the workaround? There is no workaround.