On 21/06/18 10:14, Wido den Hollander wrote:
Hi Wido,
>> Note the date stamps, the scrub command appears to be ignored
>>
>> Any ideas on why this is happening, and what we can do to fix the error?
>
> Are any of the OSDs involved with that PG currently doing recovery? If
> so, they will ignore a scrub until the recovery has finished.
>
> Or set osd_scrub_during_recovery=true
>
> Wido
your correct - osd_scrub_during_recovery=false
# ceph config get osd.333 osd_scrub_during_recovery
false
# ceph config set osd.333 osd_scrub_during_recovery true
# ceph config get osd.333 osd_scrub_during_recovery
true
Looking at the osd log, the osd immediately started scrubbing a pg after
I set osd_scrub_during_recovery=true. However it started scrubbing a
different pg (not the one I asked it to scrub).
Perhaps I need to wait? So I waited, but the osd completed scrubbing the
other pg, and then went idle.
2018-06-21 11:37:55.075 7f39a3421700 0 log_channel(cluster) log [DBG] :
4.ce8 deep-scrub starts
2018-06-21 12:14:32.096 7f39a3421700 0 log_channel(cluster) log [DBG] :
4.ce8 deep-scrub ok
Digging deeper, there are some other anomalies that may give a clue:
checking a "known good" pg on this OSD
# rados list-inconsistent-obj 4.ce8
{"epoch":170900,"inconsistents":[]}
then trying to check the active+clean+inconsistent pg on this osd
# rados list-inconsistent-obj 4.1de
No scrub information available for pg 4.1de
error 2: (2) No such file or directory
Another potential anomali, I'm not sure if this is due to the mimic
upgrade, but this cluster doesn't respond to ceph --show-config commands
as I would expect:
Both of these nodes are in the same cluster...
[root@ceph6 ~]# ceph -n osd.333 --show-config
ceph>
[root@ceph1 ~]# # ceph -n osd.219 --show-config
[root@ceph1 ~]#
any thoughts and suggestions appreciated,
thanks again and best regards,
Jake
>
>>
>> Some background:
>> Cluster upgraded from Luminous (12.2.5) to Mimic (13.2.0)
>> Pool uses EC 8+2, 10 nodes, 450 x 8TB Bluestore OSD
>>
>> Any ideas gratefully received..
>>
>> Jake
>>
--
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
