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

Serviceguard for Linux Version A.11.16 Release Notes
Patches and Fixes in this Version
Chapter 154
JAGaf78275: Kernel problem on 1 node resulted in catastrophic
SG cluster failure
Problem: Under rare circumstances, involving the timer loop thread or
the system clock on one node, all the nodes in a cluster could halt (TOC)
at the same time.
Resolution: cmcld now checks the timestamps of heartbeat messages to
ensure the clock is advancing, and aborts if it detects no advance over a
set period. This results in the failure of the node that has the clock
problem, not the entire cluster.
JAGaf83113, JAGaf97723: Too many “Unable to stat
/etc/cmcluster/cmclconfig, No such file or directory” messages fill
up syslog
Problem: This message was being written to syslog every time a UDP
probe was received by a system not configured as a Serviceguard cluster
node.
Resolution: The log level of the message has been changed so it is not
written to syslog unless cmclconfd logging is enabled.
JAGaf88042: Wrong vgcfgrestore syntax for Linux LVM2 in
Serviceguard package script
Problem: The vgcfgrestore call in the package control script was capa-
ble of handling only LVM Version 1, resulting in package startup failures
when LVM Version 2 was used.
Resolution: The script has been modified to call vgcfgrestore correctly
for LVM Version 2 as well as Version 1.
JAGaf93186: amap on hacl-cfg/udp port causes inetd looping,
cmclconfd problem
Problem: UDP messages were not marked as invalid even if there were
invalid values for length and offset fields in the message, causing
cmclconfd to exit without receiving the message; cmviewcl might also
spin indefinitely. In the cmclconfd case the message remained in the
inetd socket buffer causing inetd to spawn another cmclconfd process.
Resolution: Messages are marked invalid if the length or offset fields
contain invalid values. See also JAGaf30745.