Check what is your osd_max_backfills value. If it is more than 1, try lowering it to 1 first and see if your performance issue disappears. J. Ján Senko Proton AG Lead Storage Engineer ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Tuesday, December 7th, 2021 at 9:23, Janne Johansson <icepic.dz@xxxxxxxxx> wrote: > Den tis 7 dec. 2021 kl 09:16 skrev José H. Freidhof > > harald.freidhof@xxxxxxxxxxxxxx: > > > Hello together > > > > question: i repaired some osd and now the rebalance process are running. we > > > > suffer now performance problems. Can i pause the ongoing rebalance job and > > > > continue it at night? > > Yes, "ceph osd set norebalance" should pause them (might leave some to > > finish, but stop new rebalance jobs from starting) > > As long as you don't have degraded/recovery this is perfect to > > start/stop PGs that are in good health but misplaced/remapped > > currently. > > If you have other operations in queue also (or if you get > > degraded/undersized while rebalancing) they can end up behind one of > > those ops in the per OSD queue*) and those would not start until this > > particular rebalance finishes, but otherwise it should be fine. > > *) This is just my personal experience, haven't looked at the code > > logic or anything, just left a mimic cluster with PGs that either > > needed rebalance or backfill+degraded/backfill+undersized and the > > second ones would not run after I paused rebalances. > > Perhaps fixed in more recent versions, or perhaps my hunch is slightly off. > > ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- > > May the most significant bit of your life be positive. > > 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
 |