On Sun, Feb 01, 2015 at 03:39:55PM +0100, Joop wrote:
On 1-2-2015 10:21, Lior Vernia wrote:
>
> So the problem isn't grave (though I'll document those cases in the
> feature page), and the solution won't be trivial. Would you really want
> vdsm to track this data? And then engine would have to collect it before
> a VM is run or a vNIC is hot-plugged?...
>
Using Zabbix as a monitoring system instead of letting vdsm do it solves
most of not all of your problems. Zabbix does have a proxy daemon which
can buffer data when either hosts or vms go down and send that to the
main server as soon as the link is restored. Thins work OK as long as
the agent or proxy isn't down and should thus not be part of the cluster.
I'm not against shedding off (some of) vdsm monitoring chores, but I
don't see how this is fundamentally different from caching the
statistics in vdsm.