Hi Anthony,

On Thu, 16 Apr 2026, Anthony D'Atri via ceph-users wrote:

...

I might suggest either enabling the autoscaler or setting the mode to off.

I confirm, setting the autoscaler mode from warn to off removes the health warning. Until now I had always used mode=warn as a last resort for the absentminded admin (i.e. cluster expansion / OSD addition).

That's one of the reasons the PG autoscaler was implemented, it is very, very easy to neglect the need to recalculate, and computers should serve US after all.

What I miss from previews ceph versions is, that autoscale-status shows the suggested PG count in the column "NEW PG_NUM", like shown in the docs[1].

You might just periodically set them to warn, capture the info, and turn it back off. But I do have to ask, is there a specific reason you aren't using the autoscaler? It's much better than it used to be, and there are really great improvements in the pipe.

And regarding the ping-pong with TOO_MANY / TOO_FEW. I understand that BIAS=4.0 for the index pool should prevent oscillating suggestions?

The BIAS value means "give this pool more PGs than the amount of data would otherwise indicate". You'll see it on RGW index and some other metadata pools, which benefit from more sharding than does bulk data. What you're thinking of is the per-pool threshold. Historically this has defaulted to 3.0 : the autoscaler won't make a change until the calculated value varies by more than a factor of 3 from the current setting. I think of this as hysteresis, but there's probably a better term. There's been some fairly recent work to do this more surgically, which I _think_ removes the per-pool threshold, but in any event, I believe that recent autoscaler releases are fairly immune to this flapping.

Thanks, Sascha.

[1] https://docs.ceph.com/en/latest/rados/operations/placement-groups/#viewing-p... _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io