[ovirt-users] VM has been paused due to NO STORAGE SPACE ERROR ?!?!?!?!

[Nicolas Ecarnot]

Le 16/03/2018 à 15:48, Alex Crow a écrit :
> On 16/03/18 13:46, Nicolas Ecarnot wrote:
>> Le 16/03/2018 à 13:28, Karli Sjöberg a écrit :
>>>
>>>
>>> Den 16 mars 2018 12:26 skrev Enrico Becchetti 
>>> <enrico.becchetti at pg.infn.it>:
>>>
>>>        Dear All,
>>>     Does someone had seen that error ?
>>
>> Yes, I experienced it dozens of times on 3.6 (my 4.2 setup has 
>> insufficient workload to trigger such event).
>> And in every case, there was no actual lack of space.
>>
>>>     Enrico Becchetti Servizio di Calcolo e Reti
>>> I think I remember something to do with thin provisioning and not 
>>> being able to grow fast enough, so out of space. Are the VM's disk 
>>> thick or thin?
>>
>> All our storage domains are thin-prov. and served by iSCSI (Equallogic 
>> PS6xxx and 4xxx).
>>
>> Enrico, do you know if a bug has been filed about this?
>>
> Did the VM remain paused? In my experience the VM just gets temporarily 
> paused while the storage is expanded. RH confirmed to me in a ticket 
> that this is expected behaviour.

AFAIR, most of them went back up and running by themselves (we had to 
manually some of them from times to times).
The storage side weakness is an interesting trail to follow.
We also experienced this behavior when migrating lots of VMs at once, 
yet using a dedicated storage network.

Being on this mailing list since long, I remember we already discussed 
several times about how some users feel how oVirt can appear sensitive 
to storage latencies. On my side, the site where most of our workload 
resides is still in 3.6, so I can not yet witness the efforts oVirt devs 
have made to cope with this in 4.2 but I'm sure they did.

-- 
Nicolas ECARNOT