[Linux-HA] failback of v2 resource group

[Andrew Beekhof]

On 8/23/06, ha at ew.nsci.us <ha at ew.nsci.us> wrote:
>
> g'morning! (or whatever your timezone is)
>
> We are running heartbeat v2.0.5 and last night a resource group (postfix)
> failed over to our other server due to timeouts because the system was
> under high load (yay! heartbeat works!).  Now that it is ready for normal
> operation, how can I failback the postfix resource group?  I've tried
> crm_resource -U -r postfix (though this doesn't look like the right
> command).

"with power and flexibility comes complexity"

So we have the classic "it depends" answer.

It depends on why the resource moved to the other node.
there are basically three cases (two of which apply to you here)
1) You asked it to move by using crm_resource -M
2) It failed to start
3) You configured resource_failure_stickiness and a monitor failed one
too many times

The corresponding method of moving the resource back is then:
1) Another crm_resource -M (or crm_resource -U depending on your
stickiness values)
2) Verifying the reason for failure is rectified and using
crm_resource -C -r rsc_name -H original_node
3) Reset the failure count with: crm_failcount -D -r rsc_name -U original_node

Hope that clears things up.