On Fri, Jul 25, 2008 at 12:25, Ivan <[EMAIL PROTECTED]> wrote:
> On Thu, 2008-07-03 at 11:18 +0200, Andrew Beekhof wrote:
>> i dont know anything about stop_delay, but if you specify the
>> start_delay for the operation
>>
>>   ie. <op name=start interval=0 timeout=60s start_delay=40s/> , and/or
>>        <op name=stop interval=0 timeout=60s start_delay=40s/>
>>
>> then everything will work as you expect (the 40s is automagically
>> added to the 60s timeout)
>
> I just realised that the monitor operation requires the same start delay
> as the "start" operation otherwise HA will complain obviously...

IIRC only if the start-delay of the start action is more than 5 minutes
otherwise the cluster will wait for the start action to complete
before sending off the monitor (at which point it's own start-delay
begins)

>
>> > Would these delays be considered by HA when it comes to migration?
>>
>> No, you need to specify it for every operation you want to delay the
>> execution of.
>
> That's part I'd like some advice about. I found Xen agent working with
> "migrate_from" and "migrate_to" operations?
>
> Which one would I need to use for instance if I wanted to move ~15VMs to
> the other node in a 2 node cluster by putting one node into standby?

Not sure... I don;t know much about the Xen RA.
Probably migrate_to.

>
> I want the cluster NOT to move all at the same time because some of them
> would crash due to the load but one by one they work always...
>
> TA,
> Ivan
>
>
>
>
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to