CentOS - Apr 2018 - BUG: soft lockup - CPU#0 stuck for 36s! [swapper/0:0]

Adam Tauno Williams wrote:
> On Mon, 2017-08-07 at 15:26 +0000, KM wrote:
>> All,This happens on all of our CentOS 7 VMs.? but as stated in the
>> email trail, the file softlockup_thresh does not exist.? Should it be
>> added?? What is the best way to get rid of this behavior.
>> Thanks in advance and sorry if I missed something along the way.KM
>
> Yes, I see this behavior as well.  Never have found a solution - other
> than increasing the threshold and pretending it doesn't happen.
>
We see it a fair bit, and this is on server running on bare metal, not VMs.

      mark

On 24 April 2018 at 17:16,  <m.roth at 5-cent.us>
wrote:
> Adam Tauno Williams wrote:
>> On Mon, 2017-08-07 at 15:26 +0000, KM wrote:
>>> All,This happens on all of our CentOS 7 VMs.  but as stated in the
>>> email trail, the file softlockup_thresh does not exist.  Should it
be
>>> added?  What is the best way to get rid of this behavior.
>>> Thanks in advance and sorry if I missed something along the way.KM
>>
>> Yes, I see this behavior as well.  Never have found a solution - other
>> than increasing the threshold and pretending it doesn't happen.
>>
> We see it a fair bit, and this is on server running on bare metal, not VMs.
>
On bare metal is usually means some hardware has gone into an
uninteruptable IRQ and the CPU is waiting for it to go away. I saw
this with systems with Green disk drives a while ago. Something going
to talk to the drive would just sit for long times while the drive
spun up, the cache was validated etc. Other things would be drives on
USB disks too when some other USB item started needing input.. since
it is a hub environment they can spew for a while and the CPU would
report a soft-lockup.
>       mark
>
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos


-- 
Stephen J Smoogen.

On 04/24/18 17:33, Stephen John Smoogen wrote:
> On 24 April 2018 at 17:16,  <m.roth at 5-cent.us> wrote:
>> Adam Tauno Williams wrote:
>>> On Mon, 2017-08-07 at 15:26 +0000, KM wrote:
>>>> All,This happens on all of our CentOS 7 VMs.  but as stated in
the
>>>> email trail, the file softlockup_thresh does not exist.  Should
it be
>>>> added?  What is the best way to get rid of this behavior.
>>>> Thanks in advance and sorry if I missed something along the
way.KM
>>>
>>> Yes, I see this behavior as well.  Never have found a solution -
other
>>> than increasing the threshold and pretending it doesn't happen.
>>>
>> We see it a fair bit, and this is on server running on bare metal, not
VMs.
> 
> On bare metal is usually means some hardware has gone into an
> uninteruptable IRQ and the CPU is waiting for it to go away. I saw
> this with systems with Green disk drives a while ago. Something going
> to talk to the drive would just sit for long times while the drive
> spun up, the cache was validated etc. Other things would be drives on
> USB disks too when some other USB item started needing input.. since
> it is a hub environment they can spew for a while and the CPU would
> report a soft-lockup.
> 
Not hardly. We discovered green drives were nothing we wanted right after they 
came out. And I'm talking at work, with servers, all drives are either 
enterprise, as we bought them, or NAS-rated (e.g. WD Red).

	mark