Hi, After some testing, it looks like when the deep scrubbing is running, the performance is greatly affected. Even when only 1 active deep scrubbing is running. After some googling, suggestion to enable the kernel CFQ scheduler. https://ceph.com/geen-categorie/lowering-ceph-scrub-io-priority/
ceph tell osd.* injectargs '--osd_disk_thread_ioprio_priority 7' ceph tell osd.* injectargs '--osd_disk_thread_ioprio_class idle' However on the ceph config documentation http://docs.ceph.com/docs/mimic/rados/configuration/osd-config-ref/ "Since Jewel scrubbing is no longer carried out by the disk iothread, see osd priority options instead " Does anyone have info on this osd priority setting for Luminous release? Thanks On Wed, Jun 27, 2018 at 9:47 AM, Phang WM <ph...@silverliningsys.com> wrote: > >> On Wed, Jun 27, 2018 at 4:02 AM, Anthony D'Atri <a...@dreamsnake.net> > wrote: > >> Have you dumped ops-in-flight to see if the slow requests happen to >> correspond to scrubs or snap trims? >> >> > Hi Anthony, > Yes, we have tried the ops-in-flight, what we get is osd_op with > flag_point=delayed and event initiated, queued_for_pg, reached_pg, waiting > for rw locks... > There is no scrubs or snap trims. > > Thanks > > This electronic mail transmission and any accompanying attachments contain > confidential information intended only for the use of the individual or > entity named above. Any dissemination, distribution, copying or action > taken in reliance on the contents of this communication by anyone other > than the intended recipient is strictly prohibited. If you have received > this communication in error please immediately delete the E-mail and notify > the sender at the above E-mail address. Thank you. >
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
