VCS 5.0 with ASM - Help needed

[rhulber2]

Hi,

I am setting up a 5 node cluster, 4 servers and 1 spare Redhat 5u4.

We are using ASM and are having problems when failing to the spare server.

All 4 nodes have 1 service group with the following:-

ASMdg

ASMinst

Diskgroup

IP

NIC

Oracle

Volume

When we fail node 1 to the spare the service group fails over correctly.

Then one of the other nodes in the cluster will complain about a ASM concurrent violation error.

I think the issue could be that all the ASMinst have the sam sid (+ASM) and use the same Pfile.

So for this to work should do I need have a different ASMinst for each service group?

For example should I call them +ASM1, +ASM2 etc?

Also the ASMdg Envfile, on all 4 nodes we use DG1 and DG2, should we change this so that node 1 uses DG1 and DG2 node 2 uses DG3 and DG4 etc.

Thanks

Roy

[Michael_Auria]

The following technote in our knowledge-base should help you track down the source of the issue.

How to tell which resource is causing the "Concurrency Violation" in a VERITAS Cluster Server (VCS) cluster

http://www.symantec.com/docs/TECH9074

Good luck ...

[vcsnewbie]

Refer to Page 146 of the following Oracle agent guide to configure ASM correctly:

https://sort.symantec.com/public/documents/sf/5.1/solaris/pdf/vcs_oracle_agent.pdf
 

[Jawahar_Mohan]

Hello Roy,

I think this is configuration issue. Please ref to the Oracle Agent Installation & Config Guide sample Oracle ASM Configuratins.

Since you have multiple Oracle instances, the Oracle instances can share a single ASM instance. So, instead of having ASMinst for each group configure it as a parallel sevice group and create dependency for each oracle service group to this service group using Online Local Firm dependency. Note that each database will have its own ASM diskgroup, VxVM diskgroup and a virtual IP in its own service group so they can failover independetly. If this doesn't help, please run VRTSexplorer and contact Symantec Technical support.

Thanks,

Jawahar Mohan

[rhulber2 · Accepted solution]

Hi,

Many thanks for your ideas and suggestions.

Managed to get this working by removing the ASMdg and making the ASMInst a parallel resource across all 5 nodes.

I now use a custom oracle start/shut scripts to shutdown the database and the ASM disk groups.

Failover testing has taken place and fails without any errors.

Would of been a lot easier if the DBA had created the ASM with individual SID's (all were +ASM) as well as individual disk groups (all were DG1 and DG2).

Thanks

Roy