Adding nodes to cluster

[Rafael Gutierrez]

Hi.

I'm trying to create a cluster configuration, I'm trying to add the first two nodes but when I push OK button to submit, SAM is showing the next error message:

"Verify nodes error: This node is at revision 11.14 of ServiceGuard, node ECARV20D is at 11.15.
Unable to make a configuration changes when a node in the cluster is at a different revision"

It's pretty strange 'cause the same revision (SG 11.15 from de same depot) was installed on two nodes.

Any idea ?, thanks.

[Sanjay_6]

Hi,

Do a

swlist |grep -i Service

on both the system and compare.

Hope this helps.

Regds

[Steven E. Protter]

all nodes of the service guard cluster must be the same version of SG to be stable.

Run the command in the previous thread and upgrade the server that doesn't have the right version of the software.

Do it now to avoid headaches later.

Though mixed operation is possible, the configuration node needs to be the latest version and binary incompatability is possible. Thats straight out of the SG manual.

SEP

[Rafael Gutierrez]

On both nodes (ECARV10D and ECARV20D):

# swlist | grep Service
B3935DA A.11.15.00 MC / ServiceGuard
T1859BA A.11.15.00 ServiceGuard Extension for RAC Bundle

It's the same revision on both nodes, and I've installed SG from de same depot.

[Sanjay_6 · Accepted solution]

Hi,

I would suggest you to do a swremove and remove the current SG product from both the nodes and reinstall from the depot.

Hope this helps.

Regds

[Scott Van Kalken]

If both nodes have the RAC bundle and full service guard - should be ok.

If only one of the nodes has the RAC bundle, then that's the problem.

The RAC bundle is essentially a "cut down" version of service guard specifically for RAC.

If they've both got the same software, I would look at the installation order of the software.

Although I've never experienced it, potentially if the installation order is different for both products on the two servers, you could have blatted some binaries.

[Bernhard Mueller]

Hi,

do have an "old" node with that SG version (11.14)? which you are about to replace by a new node with the same name?

If that is the case, take the old node down and delete the arp cache on your new SG nodes before running cmquerycl.

Regards,
Bernhard

[melvyn burnard]

to be sure of your versions, do:
what /usr/lbin/cmcld
and see what version that indicates on both nodes.

Also, ensure that the CORRECT node is responding, i.e. that your host->ip address lookups are working correctly.

[Kent Ostby]

Here is a document devoted specifically to this question.

http://www1.itrc.hp.com/service/cki/search.do?category=c0&mode=id&searchString=UXSGKBRC00006816&searchCrit=allwords&docType=EngineerNotes&search.x=26&search.y=7