Re: Blocked Ops > 32 sec

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


On Thu, 12 Jan 2017, Chris Jones (BLOOMBERG/ 120 PARK) wrote:
> Hey,
> Our monitoring checks the HEALTH_OK/WARN to trigger alerts which are almost
> always 'X requests are blocked > 32 sec'. This is multi-petabyte cluster of
> only rgw using spinners. 

Out of curiousity, which version is this?  There was a bug in early jewel 
that could cause scrub to block the OSD thread for extended periods (and 
trigger these sorts of warnings).

It migght be worth checking the ceph.log to see if it is the same OSD(s) 
that are hitting the slow requests.


> What do you guys find to be the best way to
> trigger/monitor? What criteria etc? BTW, most of the 'X' requests are
> usually only '1'. Thanks in advance.
> -Chris
Ceph-large mailing list

[Index of Archives]     [Information on CEPH]     [Linux Filesystem Development]     [Ceph Development]     [Linux USB Development]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFS]

  Powered by Linux