El Miércoles 22/10/2014, Christian Balzer escribió: > Hello, > > On Wed, 22 Oct 2014 17:41:45 -0300 Ricardo J. Barberis wrote: > > El Martes 21/10/2014, Christian Balzer escribió: > > > Hello, > > > > > > I'm trying to change the value of mon_osd_down_out_subtree_limit from > > > rack to something, anything else with ceph 0.80.(6|7). > > > > > > Using injectargs it tells me that this isn't a runtime supported change > > > and changing it in the config file (global section) to either host or > > > room has no effect. > > > > > > Christian > > > > I had a similar problem with 0.80.7 and "mon osd downout subtree limit = > > host" till I realized it's actually "mon osd down out subtree limit = > > host" (notice the space between "down" and "out"). > > That's exactly what it was, thanks. > > And while I feel moderately stupid for not spotting this difference > between the config file and what the active configuration displays I > really, REALLY would love for Ceph to log anything it finds wrong with > its config during startup. Given how massively chatty Ceph is otherwise > that would be a most welcome and USEFUL addition to the log deluge. Yep, that'd be nice. > > After putting that option in ceph.conf and restarting mons and osds I > > can see the change, haven't tested it yet though. > > Yeah, I shall go and test this now. ^^ I tested it earlier today: a disk died so I had 1 osd down and recovering, I shutdown the others osds on the same host and the recovery for the host didn't start. One oddity I see is that after I started the host again without the faulty disk and the recovery finished, ceph says HEALTH_OK but I still see one osd out/down: "53 osds: 52 up, 52 in". > > (Christian: If this is not your case, sorry for hijacking your thread!) > > > > > > Anyway, the docs should to be corrected on this point: > > > > http://ceph.com/docs/master/rados/configuration/mon-osd-interaction/ > > Indeed, that's where I cut and pasted it from as well. I asked on IRC how to go about this, I'll submit a patch tomorrow. > Thanks again, > > Christian > > > Cheers, _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |