alert on stand-by server failed in a MC/SG cluster

[Gary Yu]

Dear all,

in a 2 servers MC/SG cluster running one package, normally the standby server is just idle, I know that both severs check health of each other by heartbeat, in case of the standby server dies, the primary server still runs healthy, MC/SG will automatically reform itself into a one node cluster (am I rignt ?), but how can we catch and being alerted on this specific situation from MC/SG? I mean, we don't want this happened silently, so that miss the chance to fix the standby server promptly.

thanks,
Gary

[melvyn burnard]

why not take a look at Serviceguard Manager?
It is free!!, and the client can run on HP-UX, Linux, or Windows

See here:

http://software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=B8325BA

[Geoff Wild]

What you really want is to monitor your cluster with OVO - that way you will be notified on any issues...

Rgds...Geoff

[Bharat Katkar · Accepted solution]

Gary,
That is a very important things you have noticed and only way out is to monitor the situation in any possible way. Also in addition to it what will you do if your standby NIC is bad or what if packages don't switch automatically when failover occurs. As suggested earlier you can go in for SG manager or configure it to send email notifications but again monitoring doesn't go away. Also you should follow routine failover test if the setup is really critical.

Our daily activities starts with monitoring the entire setup and running cmviewcl was one of them. Looking after LED indications of the server, hardware status of both the server, monitoring syslog files, etc..
Hope that helps.

Regards,