Hi Mac
We've also tweaked
osd-recovery-max-single-start => 2
osd-recovery-sleep-hdd => 0.05
to speed things up.
On 2020-10-20 16:04, Mac Wynkoop wrote:
OK, so for interventions, I've pushed these configs out:
ceph config set mon.* target_max_misplaced_ratio 0.05 > 0.20
ceph config get osd.* osd_max_backfills 1 > 4
ceph config set osd.* osd_recovery_max_active 1 > 4
And also ran injectargs to push the changes to the OSDs hot. I'll monitor
it for a bit to see how it reacts to the more aggressive settings.
Thanks,
Mac Wynkoop
On Tue, Oct 20, 2020 at 8:52 AM Eugen Block <eblock@xxxxxx> wrote:
The default for max misplaced objects is this (5%):
ceph-node1:~ # ceph config get mon target_max_misplaced_ratio
0.050000
You can increase this for the splitting process but I would recommend
to rollback as soon as the splitting has finished.
Zitat von Lindsay Mathieson <lindsay.mathieson@xxxxxxxxx>:
On 20/10/2020 11:38 pm, Mac Wynkoop wrote:
Autoscaler isn't on, what part of Ceph is handling the increase of
pgp_num?
Because I'd like to turn up the rate at which it splits the PG's, but if
autoscaler isn't doing it, I'd have no clue what to adjust. Any ideas?
Normal recovery ops I imagine - Bump up the recovery settings, Max
Backfills and Recovery Max Active
--
Lindsay
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
--
Toby Darling, Scientific Computing (2N249)
MRC Laboratory of Molecular Biology
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
