Daniel P. Berrange wrote:
> On Tue, Apr 21, 2015 at 03:47:34PM -0600, Jim Fehlig wrote:
>
>> Daniel P. Berrange wrote:
>>
>>> On Fri, Apr 17, 2015 at 03:36:22PM -0600, Jim Fehlig wrote:
>>>
>>>
>>>> + VIR_FREE(priv->lockState);
>>>> + if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
>>>> + VIR_WARN("Unable to release lease on %s", vm->def->name);
>>>> + VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));
>>>>
>>>>
>>> I see a few calls to ProcessPause, but I'm not seeing any to ProcessResume,
>>> so I think you might have missed a few places needing hooks.
>>>
>>>
>> Currently I'm not using ProcessResume and instead calling ProcessStart
>> with paused param set to false. ProcessStart is called in
>> libxlDomainStart, which is called by all paths that start a domain, e.g.
>> start, create, restore, migrate, etc. I was under the impression that
>> ProcessResume is not needed if ProcessStart is called with paused=false.
>>
>
> For virtlockd it doesn't matter, but with the sanlock driver there is a
> tiny difference due to VIR_LOCK_MANAGER_NEW_STARTED being passed to the
> virLockManagerNew() method in the Start case.
>
Ah, I see. Thanks for the explanation. I'll send a V3 to call
ProcessResume right after ProcessStart. I also noticed a small goto
label bug that will be fixed in V3.
Regards,
Jim
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list