On Mon, Oct 21, 2013 at 07:49:58AM -0400, Tom Parker wrote:
> Thanks for the feedback.  Dejan, I have some SLES nodes that are running
> around 30 pretty heavy VMs and I found that while I never go to 5s that
> the time it would take to reboot was not a constant.

As mentioned elsewhere, it is actually 5s _plus_ however long it
takes to run 5 times the Xen status procedure.

>   I have a feeling
> that this bug in xen-list

xen-list is coming from SUSE actually. I don't know if it's
upstream.

> may take a while to be fixed upstream and
> trickle down into the released xen packages so we may be using this fix
> for a while.

Yes, the workaround is probably going to stay around for some
time.

> The full longdesc now reads:
> 
> <longdesc lang="en">
> When the guest is rebooting, there is a short interval where the guest
> completely disappears from "xm list", which, in turn, will cause the monitor
> operation to return a "not running" status.
> 
> If a monitor status returns "not running", then test status
> again for wait_for_reboot seconds (perhaps it'll show up).
> 
> NOTE: This timer increases the amount of time the cluster will
> wait before declaring a VM dead and recovering it.
> </longdesc>

I sincerely hope that this won't be necessary.

On a side note: What would you set this wait_for_reboot
parameter to? Which value would make you feel safe? Which value
would really keep the data safe?

Cheers,

Dejan

> 
> Tom
> 
> On 10/21/2013 03:28 AM, Ulrich Windl wrote:
> > "When the guest is rebooting, there is a short interval where the guest
> > completely disappears from "xm list", which, in turn, will cause the monitor
> > operation to return a "not running" status. If the guest cannot be found , 
> > this
> > value will cause some extra delay in the monitor operation to work around 
> > the
> > problem."
> 
> _______________________________________________
> Linux-HA mailing list
> Linux-HA@lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to