Mounting filesystem readonly on failover node

[PBu]

I have been asked to activate a volume group readonly and mount the filesystem on it on a node in a ServiceGuard 11.20 cluster. The volume group is already mounted as rw on another node in the cluster. From the Managing ServiceGuard document this would not seem to be a problem.

My question is that the readonly node is the failover node for the packages on the readwrite node. I have not been able to find anything on package behaviour in the event of a package failover when the package tries to activate the volume group as readwrite when it already activated on the failover as readonly with a filesystem on the volume group mounted as readonly.

During a failover will the volume group activation fail causing the package to fail to start on the failover node?

I appreciate any comments if anyone has experience with this sort of thing.

The systems in the cluster are rx6600's running HP 11.31 and ServiceGuard 11.20

[Torsten.]

I'm not exactly sure what you want to do, but use

# vgchange -a r ...

for read-only.

[Patrick Wallek]

This does not sound like a good idea...

It depends on how you have the VG mounted now.  If you have the VG mounted on the primary cluster node as exclusive ('vgchange -a e' in the package control script) then you may get an error if you try to mount the same VG on another node.

I believe you are correct in your assertion that the package may fail to start since it will not be able to (re)activate the VG on the failover node as it would already be active (if you were successful in getting mounted read-only).  You could work around this issue by modifying your package control script and have deactivate the VG (just in case) before it activates in read-write mode for the package start.

[Emil Velez_2]

why not export the filesystem and mount it read only using the package virtual ip using nfs. You can use the nfs package or add nfs features to your existing package to do that