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 150
3. Now, when you are able to take the entire cluster down, run
cmhaltcl -f on one of the nodes that is still running in the
Serviceguard cluster.
4. After the cluster has successfully halted, start the Serviceguard
cluster on the nodes that were upgraded to CVM 3.5. To do this,
you must not run the normal cmruncl command to start up the
Serviceguard cluster, since that will attempt to start all cluster
nodes. Instead, run cmviewcl -n node1 -n node2 -n noden,
where the nodes specified are the nodes that were upgraded to
CVM 3.5.
5. Now perform the CVM upgrade from 3.2 to 3.5 on the remaining
nodes in the cluster.
6. After the upgrade to CVM 3.5 has been completed on each node,
the node can rejoin the running Serviceguard cluster by running
the command, cmrunnode on that node.
For example, in a 4 node cluster, you could take down 2 nodes in the
cluster and upgrade CVM to 3.5, while the other 2 nodes continue
running the cluster with CVM 3.2. Then halt the entire cluster, and
start up the cluster on the 2 nodes that were upgraded to CVM 3.5.
Now upgrade the remaining CVM 3.2 nodes to CVM 3.5; once the
upgrade is complete, they can be added back into the cluster.
JAGaf91413: cmmodnet does not check SUBNET mask
when adding an IP
• What is the problem? If the subnet mask of a network interface has
been changed from the value originally used when the cluster
configuration was last applied, any package relocatable IP addresses
added to that network interface (when a package starts) will still use
the original subnet mask, resulting in an inconsistent network
configuration
• What is the workaround? Check error messages for cmapplyconf.
Error message will prompt the user to fix the subnet mask problem