I agree. This was just our brute-force method on our test cluster. We won't do this on production cluster.
--Jiten On Nov 11, 2014, at 2:11 PM, cwseys <[email protected]> wrote: > 0.5 might be too much. All the PGs squeezed off of one OSD will need to be > stored on another. The fewer you move the less likely a different OSD will > become toofull. > > Better to adjust in small increments as Craig suggested. > > Chad. _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
