cancel
Showing results for 
Search instead for 
Did you mean: 

ClusterFailOverPolicy is set to Auto

Marius_Gordon
Level 4
Certified

Advice needed please.

We have a global cluster. 1 node per cluster.

The ClusterFailOverPolicy is set to Auto

However when I reboot node 1 in cluster 1 it does not failover to the DR node automatically.

Is this by design? Is there a way to set the cluster to failover when rebooted?

5 REPLIES 5

Mouse
Moderator
Moderator
Partner    VIP    Accredited Certified

The engine_A log will tell you more why it was not triggered but I'd expect that a graceful reboot is not considered as a failure, it's rather an arbitrary action 

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Hi,

The point is that YOU rebooted the cluster so YOU could have performed the failover. VCS only reacts to unexpected occurences i.e. faults. 

Thanks for the feedback.

The reboot was unexpected. It was not a healthy reboot.

Mouse
Moderator
Moderator
Partner    VIP    Accredited Certified

You need to check the engine_A.log. The reboot might be unexpected to you, however as long as it is performed through a standard procedure, such as init, shutdown or reboot (there are other options, too), VCS will treat it as graceful. An example of unexpected reboot would be press of the reset button, Stop-A or its equivalent in Solaris, pulling power cables out, etc.

Walter_Heim
Level 3

Hi Marius,

I would guess that your node is rebooting before the timeout is reached which would cause a cluster to cluster failover.

The typical time for cluster to cluser failover is measured in minutes.  Cluster to cluster failover will happen if given enough time with the original cluster being offline/unavailable.  I think I've seen it take 6-11 minutes or longer depending an multiple variable settings in each cluster..  This is by design so that a false site failover is not preformed which might cause a replication takeover to be done.

There are a few things that could be altered to bring down the cluster to cluster failover times.  But I don't think you will be able to get much below 5 or 6 minutes.  If you original node is rebooting faster than that, there might not be much you can do except initiate manual move of the service groups prior to rebooting the node.

Thank you,

Wally