[Linux-cluster] Testing cluster failover

[Enrique Sanchez]

On Thu, Jan 13, 2011 at 11:39 PM, KC LO <kclo2000 at gmail.com> wrote:
> Dear Friends,
>
> Thanks for your advise on my new cluster setup.  I am going into failover
> testing of the cluster.
>
> My configuraiton involves 3 active member and 1 standby servers and running
> Redhat 5.5 with Cluster installed.
> In my active server, when I type "reboot", it can successfully relocate the
> service to the standby node.
> However, If I type "init 0" to simulate server failure, it can't relocate to
> the standby node.  Is it the correct behaviour?
> Anything that I should check.  Any advice?
>

there are differences in behavior between an abnormal cluster exit and
a normal one, reboot makes it abnormal while "init 0" first calls all
services running on your run level with an "stop" option, making it
normal.

You need to configure the cluster to force a take-over even on these situations.

regards,
enrique.

-- 
Enrique Sanchez Vela
------------------------------------------