How to Check VCS Failover and Primary node.

[99cassano]

Hi,

I am a newbie to VCS and is learning by setting up trial version in two VMs.

How do i check in command that the VCS has been failed over? and after failover, which one is the Primary node?

Thanks

[mikebounds]

Use "hastatus -sum"

This should show the service group as faulted on one node and should so ONLINE on the other node.

Mike

[99cassano]

Thanks Mike.

One additional question here. If i don't want to create fault in one node to test the fault over, do i invoke hastop -local in one node? However, in this case if you do hastatus -sum after failover, both nodes will still be showing ONLINE. So where I should look for the current primary node is? 

[mikebounds]

To test failover, you can just use "hagrp -switch grp_name -to sysname" and "hastatus -sum" should show the group is now online on the other system.

Can you provide output of "hastatus -sum" after "hastop -local" as it should show one system is NOT in running state.  If the group was online on the node you ran "hastop -local", then "hastatus -sum" should show group is OFFLINE on both nodes (as hastop -local will OFFLINE, not FAILOVER the groups), but if group was online on the other node, then it should remain ONLINE.  If you want groups to failover when you run "hastop -local", you need to add the "-evacuate" flag, so run "hastop -local -evaculate"

Mike

[99cassano]

Hi Mike,

Thanks. so I should use "hastop -local -evacuater" when I do VCS failover test? Is this how the real world is doing when they have to do to test disater scenario"? I didn't setup any service group in my test environment yet so when I do hastop -local, the Cluster Service will be online on the failover node automatically. So, if I define the service group resoures for an application later, will hasstop -local -evacuate safetly failover everything to the second node?

Thanks 

[mikebounds · Accepted solution]

See answers below:
 

so I should use "hastop -local -evacuater" when I do VCS failover test? Is this how the real world is doing when they have to do to test disater scenario"?

Normally you test by downing a node (like power down system board or pull power cable, or for a VM you can just power VM off)

I didn't setup any service group in my test environment yet so when I do hastop -local, the Cluster Service will be online on the failover node automatically.

This is because ClusterService isa special service group which VCS always tries to keep online

So, if I define the service group resoures for an application later, will hasstop -local -evacuate safetly failover everything to the second node?

Yes

Mike