On Wed, 11 Mar 2015, Dan van der Ster wrote: > Hi Sage, > > On Tue, Mar 10, 2015 at 8:34 PM, Sage Weil <sweil@xxxxxxxxxx> wrote: > > Adjusting CRUSH maps > > -------------------- > > > > * This point release fixes several issues with CRUSH that trigger > > excessive data migration when adjusting OSD weights. These are most > > obvious when a very small weight change (e.g., a change from 0 to > > .01) triggers a large amount of movement, but the same set of bugs > > can also lead to excessive (though less noticeable) movement in > > other cases. > > > > However, because the bug may already have affected your cluster, > > fixing it may trigger movement *back* to the more correct location. > > For this reason, you must manually opt-in to the fixed behavior. > > > > In order to set the new tunable to correct the behavior:: > > > > ceph osd crush set-tunable straw_calc_version 1 > > > > Since it's not obvious in this case, does setting straw_calc_version = > 1 still allow older firefly clients to connect? Correct. The bug only affects the generation of internal weight values that are stored in the crush map itself (crush_calc_straw()). Setting the tunable makes the *monitors* behave properly (if adjusting weights via the ceph cli) or *crushtool* calculate weights properly if you are compiling the crush map via 'crushtool -c ...'. There is no dependency or compatibility issue with clients, and no need to set tunables to 'firefly' to set straw_calc_version. sage _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |