Hello Mehmet,
On Sat, Sep 3, 2022 at 1:50 PM <ceph@xxxxxxxxxx> wrote:
> Is ceph still backfilling? What is the actual output of ceph -s?
>
Yes:
[truiten@xxxxxxxxxx ~]$ sudo ceph --cluster xxx -s
cluster:
id: 91ba1ea6-bfec-4ddb-a8b5-9faf842f22c3
health: HEALTH_WARN
1 backfillfull osd(s)
3 pool(s) backfillfull
Low space hindering backfill (add storage if this doesn't
resolve itself): 3 pgs backfill_toofull
services:
mon: 5 daemons, quorum a,b,c,d,e (age 6d)
mgr: b(active, since 45h), standbys: a, c, d, e
mds: registration_docs:1 {0=b=up:active} 3 up:standby
osd: 32 osds: 32 up (since 19M), 32 in (since 3y); 12 remapped pgs
mds: 1 daemon active (b)
data:
pools: 3 pools, 1280 pgs
objects: 14.32M objects, 23 TiB
usage: 69 TiB used, 47 TiB / 116 TiB avail
pgs: 543262/42962769 objects misplaced (1.264%)
1268 active+clean
9 active+remapped+backfilling
3 active+remapped+backfill_toofull
io:
client: 5.0 MiB/s rd, 296 KiB/s wr, 10 op/s rd, 0 op/s wr
recovery: 73 MiB/s, 36 keys/s, 44 objects/s
>
> If not backfilling, it is strange that you only have 84 pgs on osd.11 but
> 93.59 percent in use...
>
This morning it wasn't backfilling, but after I did another `osd
reweight-by-utilization`, it started again.
>
> Are you able to find a pg on 11 which is too big?
> Perhaps pg query will help to find. Otherwise you should lower the weight
> of the osd...
>
It's a nautilus cluster, it looks like the pg query command doesn't exist
there. How would I find the large pg(s) on osd.11 and how could I force
them off the osd?
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
