System Administration
cancel
Showing results for 
Search instead for 
Did you mean: 

Re: cluster configuration with cluster suite

 
Highlighted
Frequent Advisor

cluster configuration with cluster suite

Hi all,
I have a redhat cluster suite linux5.5 done with redhat suite manager.When the Node1 restarts the shared disks are mounted automaticaly.But when the node2 restarts, it can not mount the shared disks automaticaly.
I need some assistance please
1 REPLY 1
Honored Contributor

Re: cluster configuration with cluster suite

The cluster configuration you've attached contains no information at all about your shared disks. So my answer is going to be basically wild guesses. If I knew more about your disks, I might be able to give you a better answer.

Do you get any error messages when booting node2? What do they say? Please try to copy the exact message(s).

Are you using GFS or GFS2?

-------------

If you are, check at least these things:

- Compare the /etc/fstab files on both nodes. Is the line for the GFS filesystem configured the same way on both nodes?

- If your shared disks uses GFS or GFS2, you should define it as a clusterfs resource in your cluster configuration. This tells the cluster suite that your service requires the cluster filesystem, and so the cluster suite won't try to move the service to a node that has problems with that filesystem.

-------------

If you aren't using GFS or GFS2 (or some other cluster filesystem) on your shared disks, then you cannot mount them on two nodes simultaneously without causing data corruption.

If you're using LVM and have installed the cluster using Conga, then the cluster installation has switched your LVM to cluster mode: in this case, all the volume groups you've created after installing the cluster will be configured to cluster mode by default. In cluster mode, LVM will not allow you to activate cluster volume groups unless the clvmd daemon is running and connected with the clvmd daemons of the other running nodes. It should also stop any attempts to mount a filesystem on a cluster VG that is already mounted on another node, unless GFS or GFS2 filesystem is used.

In this case, you should define the shared disks as filesystem (fs) resource(s) in your cluster configuration, and add them to your service definition. You should remove the shared filesystems from /etc/fstab from both nodes... or at the very least, configure them using the "noauto" mount option and fdisk pass number "0" so the boot procedure won't try to check nor mount them.

After this change, checking & mounting the shared disks will be the responsibility of the cluster suite. The shared disks will be mounted only on the node that is actually running the service.

MK
MK