Serviceguard Version A.11.16 Release Notes, 2nd Edition, September 2004

Serviceguard Version A.11.16 Release Notes, Second Edition
Known Problems and Workarounds
Chapter 1 49
JAGaf24444: Unable to receive device query message
• What is the problem? Serviceguard configuration commands
(cmquerycl, etc) may fail if you have no LVM volume groups
configured on one or more of your systems.
• What is the workaround? The work around is to create a volume
group and import it on all nodes that have no LVM volume groups.
There is a patch for Serviceguard A.11.14: PHSS_31015. Patch
numbers for A.11.15 and A.11.16 will be posted when testing is
complete.
JAGaf32484: Problems doing a rolling upgrade when
moving from CVM 3.2 to 3.5
• What is the problem? In a Serviceguard cluster with CVM 3.2, it is
not possible to perform a rolling upgrade from CVM 3.2 to CVM 3.5.
After a node is upgraded to CVM 3.5, when it attempts to rejoin the
CVM cluster, it will fail, possibly causing the node to be TOC'ed.
This problem exists regardless of whether the HP-UX operating
system revision was changed during the rolling upgrade. By "rolling
upgrade", we mean halting one node in the cluster, upgrading CVM
to 3.5 and then starting the node back in the cluster, then performing
the same steps for each node in the cluster. The cluster is never
completely halted. However, this issue does require the cluster to be
completely halted for at least some period of time, since it is not
possible to run the cluster with a mix of CVM 3.2 and 3.5.
• What is the workaround?
A workaround exists to do a semi-rolling upgrade to a cluster being
upgraded from CVM 3.2 to 3.5. It requires a small amount of
downtime though.
This workaround involves managing the nodes in the cluster so that
all nodes in the clusters are running only CVM 3.2 or only CVM 3.5,
but never a combination of the two. To do this:
1. Halt Serviceguard on one or more of the cluster nodes while the
other nodes remain in the cluster, still running CVM 3.2.
2. Perform the upgrade from CVM 3.2 to 3.5 on the nodes which are
not running in the Serviceguard cluster, but do not restart them
in the Serviceguard cluster when the CVM upgrade is complete.