how do halt a cfs Cluster?

[gigiz]

Hi guys,

I have a cfs Cluster, with cfs fie system (shared), multinode package with dependency.

what is the exact procedure for halt the cfs cluster ?

[Suman_1978]

Hi,

Hope these links help.

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c01279533

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c02254740

[gigiz]

Mmm, this is for manage the star or stop of package (DG/MP).

I want know what is a commad to halt  a CFS CLUSTER.

I think that a command is cmhaltcl -f, but i want a opinion .

I have read in the " Sample Configurations with SGeRAC and Oracle RAC 10gR2"  document this : "When halting the Serviceguard cluster (“cmhaltcl –f”), the package dependencies ensure that the Oracle Clusterware packages are halted first before the cluster file systems are unmounted."

Any comment?

[Matti_Kurkela · Accepted solution]

First, shutdown all applications using the CFS filesystem(s).

Then halt the packages as per the instructions in Suman_AN's links.

If some package does not halt or reports errors, read the package logs to find more information, and fix things as necessary (usually, identify and then properly shutdown any user processes that have been accidentally left running, accessing the CFS filesystems). Make notes of all things you find you will need to do in this step.

Finally halt the cluster infrastructure processes on all nodes using "cmhaltcl -f".

In theory, if your cluster is 100% properly configured, then "cmhaltcl -f" should do all this automatically, using the package dependencies to find the correct order to halt the various things. But if you knew that for sure, you probably would not be asking here, would you?

When it's time to restart the cluster again, run "cmruncl", then check if all the necessary filesystem packages and applications are started automatically. If they are, you can be reasonably sure your cluster is properly configured; if not, make notes of all the things you need to start manually.

Your notes will be your high-level list of things to check/add/fix in the cluster configuration. Essentially, "Must find what should be changed so that ________ will happen automatically without errors at next cluster/package shutdown/restart."