> > ceph osd setcrushmap -i /tmp/crush.new > > > > Note: If you are overwriting your current rule, your data will need to > > rebalance as soon as your set the crushmap, close to 100% of your > > objects will move. If you create a new rule, you can set your pool to > > use the new pool id anytime you are ready. And if you have set the "norebalance" flag just before this and have a ceph cluster which uses the upmap balancer, you can use one of the pgremap*) tools to temporarily tell the cluster that all the misplaced PGs are actually at the "correct" place for now, then unset norebalance, and then the balancer will take some ~8 PGs at a time and correct their placements until the whole pool is placed according to the new rules. Good if you don't want movement to impact client IO, but if your cluster is new and not full with data, this might not be needed. Still, a good way to slowly move things around without impact, also when adding or removing OSDs. *) https://github.com/HeinleinSupport/cern-ceph-scripts/blob/master/tools/upmap/upmap-remapped.py (python) or https://github.com/digitalocean/pgremapper (golang) -- 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
 |