Our ceph cluster uses 260 osds. The most highest osd usage is 87% But, The most lowest is under 40%. We consider lowering the highest osd's reweight and rising the lowest osd's reweight.

How many Placement Groups do you have per OSD? We had uneven balanced OSDs in the past as well and were able to reduce the imbalance by increasing the placement groups. Best Regards Jan

I’ve thought about this strategy in the past. I think you might enter a cron job to reset any OSDs at 1.0 to 0.5, but really the balancer module or JJ balancer is a better idea than old-style reweight.

On Jun 19, 2024, at 2:22 AM, 서민우 <smw940219@gmail.com> wrote:

Hello~

Our ceph cluster uses 260 osds. The most highest osd usage is 87% But, The most lowest is under 40%. We consider lowering the highest osd's reweight and rising the lowest osd's reweight.

We solved this problem following this workload. This is our workload. 1. Set All osd's reweight to 0.5 2. Rising the lowest osd's reweight to 0.6 3. Lowering the highest osd's reweight to 0.4

** What I'm really curious about is this. ** I want to set the default osd reweight to 0.5 for the new osd that will be added to the cluster. _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io