crash dump

[tanislavm]

hi,

i wish to force a crash dump on a vcs node.how should i perform it?

should i failover first the service group or no?

then i force the crash dump on this node.

if i do not failover the service group,then:

-if the node where i forced the crash dump will shutdown,then if i like to start the service group on other node-i should import forcefully the disk group and maybe other cleanup things for the application to start.

-if the node where i forced the crash dump reboots,and join to the vcs,then the service group will be brought up on this node automatically.

please comment.

tnx so much.

[Gaurav_S · Accepted solution]

Hello,

What is the reason of forcing a crash dump ?

Usually crash dump is performed to get current state of memory pages ... if there is a problem existing, then you take a crash dump so that same can be analyzed for the problem.

If you switch service groups, & if problem doesn't exist on server, then crash dump will be of no use ...

You have options in solaris like Savecore -L to take a live core (not a preferred choice though) .. otherwise you can use methods like sync or halt -d to generate dumps ..

G

[tanislavm]

hi,

so the answer would be to take the crash dump on the node as it is,i mean the service group will remains on the node.tnx so much.

[Gaurav_S]

If you are taking the crash dump when there is a problem on the server & if the suspect is VCS related components.. then yes ..

G

[tanislavm]

hi,

tnx so much for your so valuable answers as usual.