[Users] "Volume Group does not exist". Blame device-mapper ?

[Nicolas Ecarnot]

Le 29/01/2014 13:36, Itamar Heim a écrit :
> On 01/29/2014 02:35 PM, Nicolas Ecarnot wrote:
>> Le 29/01/2014 13:29, Maor Lipchuk a écrit :
>>> Hi Nicolas,
>>>
>>> Can u please attach the VDSM logs of the problematic nodes and valid
>>> nodes, the engine log and also the sanlock log.
>>>
>>> You wrote that many nodes suddenly began to become
>>> unresponsive,
>>> Do you mean that the hosts switched to non-responsive status in the
>>> engine?
>>> I'm asking that because non-responsive status indicate that the engine
>>> could not communicate with the hosts, it could be related to sanlock
>>> since if the host encountered a problem to write to the master domain it
>>> causes sanlock to restart VDSM and make the hosts non responsive.
>
> non-resposneive for engine is if vdsm is up/responsive.
> run locally;
> # vdsClient -s 0 getVdsCaps
>
> to check vdsm is ok

When I find the time for it, I'll reproduce the crash and run this 
command and let you know.
I must admit this was scary.

-- 
Nicolas Ecarnot