Re: [Xen-devel] Xen 4.8 + Linux 4.9 + Credit2 = can't bootup

[Anshul Makkar <anshul.makkar@xxxxxxxxxx>]

On 05/01/2017 08:39, Dario Faggioli wrote:
> On Thu, 2017-01-05 at 02:05 +0000, Andrew Cooper wrote:
>> On 05/01/2017 01:52, Konrad Rzeszutek Wilk wrote:
>>> It works just fine with credit1 (see further down the log)
>>> but if I try credit2 it ends up hanging during bootup.
>>>
>>> I am a going to naively assume it is due to how the vCPUs are
>>> exposed (Where they match the physical CPUs under credit1),
>>> but under credit2 they are different.
>>>
>>> The dom0_max_vcpus does not seem to have any affect. When I remove
>>> it
>>> things are still being problematic.
>>>
>>> Help!?
>>
>> This matches the symptoms seen by XenServer when trying to stress
>> 32vcpu
>> guests under Credit2.  Malcolm did find (based on interpreted iperf
>> throughput graphs) that Credit2 did seem to preferentially schedule
>> the
>> lower-number vcpus, rather than scheduling them evenly.
>>
> To be fair (and just for the records, since the cause seems actually to
> be something else), this was with an old version (at least two Xen
> releases ago, IIRC, certainly not 4.8) and known to be buggy version of
> Credit2.
>
> We have other tests and benchmarks, done on equally big machines which
> proves the scheduler is 100% functional.
>
Yes, I have done extensive stress tests on XenSever(IIRC Xen - 4.7) with 
Credit2 on 32 or more VCPU guests. Didn't see any hangups.

Malcolm, did find hangup and crash issues with 32VCPU guests, but the 
issue was highlighted and fixed. So, during my testing I specifically 
foucussed on this scenario and found no problem.

> Regards,
> Dario
>

Anshul

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel