Hi,
I was planing to activate the pg_autoscaler on a EC (6+3) pool which I
created two years ago.
Back then I calculated the total # of pgs for this pool with a target
per ods pg # of 150 (this was the recommended /osd pg number as far as I
recall).
I used the RedHat ceph pg per pool calculator [1] with the following
settings:
PoolyType: EC
K: 6
M: 3
OSD #: 220
%Data: 100.00
Target PGs per OSD: 150
This resulted into a suggested PG count of 4096, which I used to create
that pool.
(BTW: I get the same using the ceph.io pgcalc [2], setting size to 9
(=EC 6+3))
# ceph osd df shows that I have now 162-169 pgs / osd
Now I enabled the pg_autoscaler and set autoscale_mode to "warn", using
the following commands:
# ceph mgr module enable pg_autoscaler
# ceph osd pool set hdd-ec-data-pool pg_autoscale_mode warn
when i set the target_ratio_size to 1, using:
# ceph osd pool set hdd-ec-data-pool target_size_ratio 1
I get the warning:
1 pools have too few placement groups
The autoscale status shows that pg number in the EC pool would get
scaled up to 16384 pgs:
# ceph osd pool autoscale-status
POOL SIZE TARGET SIZE RATE RAW CAPACITY RATIO TARGET
RATIO BIAS PG_NUM NEW PG_NUM AUTOSCALE
ssd-rep-metadata-pool 11146M 3.0 29805G 0.0011
1.0 64 4 off
ssd-rep-data-pool 837.8G 3.0 29805G 0.0843
1.0 1024 64 off
hdd-ec-data-pool 391.3T 1.5 1614T 0.3636
1.0000 1.0 4096 16384 warn
How does this relate the the calculated number of 4096 pgs for that EC
pool? I mean 16384 is quite a jump from the original value.
Is the autoscaler miscalculating something or is the pg calculator wrong?
Can someone explain it? (My cluster is on ceph version 14.2.7)
Thanks so much
Dietmar
[1]: https://access.redhat.com/labs/cephpgc/
[2]: https://ceph.io/pgcalc/
--
_________________________________________
D i e t m a r R i e d e r, Mag.Dr.
Innsbruck Medical University
Biocenter - Institute of Bioinformatics
Innrain 80, 6020 Innsbruck
Email: dietmar.rieder@xxxxxxxxxxx
Web: http://www.icbi.at
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
