On Wed, 11 Aug 2010, Henry C Chang wrote: > After adding osd3, only osd2 is alive. > But the whole cluster can be recovered after restarting all dead osds. > > I read the wiki and reversed the order of my procedure: starting osd > before setting crush map. > It seems to work ok now. Hmm, the order shouldn't matter--there's definitely something going wrong. Are you able to reproduce the crash with the old order? (With logs? :) sage > > Thanks, > Henry > > > On Wed, Aug 11, 2010 at 4:56 AM, Sage Weil <sage@xxxxxxxxxxxx> wrote: > > Hi Henry, > > > > Your osd add procedure looks correct. Did the osd start correct after > > this point? Are you able to reproduce the problem with osd logging > > turned up (debug osd = 20 in [osd])? > > > > I just tried this with the latest unstable and wasn't able to reproduce > > the problem. > > > > sage > > > > On Wed, 11 Aug 2010, Henry C Chang wrote: > > > >> Hi Wido, > >> > >> I am running unstable branch (commit: b72c1bb6e9b77e1ab6c2) dated July 30. > >> I just uploaded the core dumps to > >> > >> http://home.anet.net.tw/cycbbb/coredumps/core.2580.gz > >> http://home.anet.net.tw/cycbbb/coredumps/core.2602.gz > >> > >> > >> > >> On Tue, Aug 10, 2010 at 11:42 PM, Wido den Hollander <wido@xxxxxxxxx> wrote: > >> > Hi Henry, > >> > > >> > Is there a core-dump of these crashes in /? If so, these could help > >> > finding the cause of this. > >> > > >> > See: http://ceph.newdream.net/wiki/Troubleshooting > >> > > >> > And btw, which version of Ceph are you running? > >> > > >> > Wido > >> > > >> > On Tue, 2010-08-10 at 21:09 +0800, Henry C Chang wrote: > >> >> Hi, > >> >> > >> >> I have a ceph cluster: 3 (mon+osd) and 2 (mds). > >> >> When I tried to add the 4th osd to the cluster, osd0 and osd1 crashed. > >> >> The error logs are attached. > >> >> > >> >> My procedure to add the 4th osd is: > >> >> > >> >> add [osd3] in the conf file: /etc/ceph/ceph.conf > >> >> ceph -c /etc/ceph/ceph.conf mon getmap -o /tmp/monmap > >> >> cosd -c /etc/ceph/ceph.conf -i 3 --mkfs --monmap /tmp/monmap > >> >> ceph -c /etc/ceph/ceph.conf osd setmaxosd 4 > >> >> osdmaptool --createsimple 4 --clobber /tmp/osdmap.junk --export-crush > >> >> /tmp/crush.new > >> >> ceph -c /etc/ceph.conf osd setcrushmap -i /tmp/crush.new > >> >> /etc/init.d/ceph -c > >> >> /etc/ceph/d2c5d946-b888-40b3-aac2-adda05477a81.conf start osd > >> >> > >> >> Is my procedure to add an osd incorrect? > >> >> > >> >> Thanks, > >> >> Henry > >> > > >> > > >> -- > >> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > >> the body of a message to majordomo@xxxxxxxxxxxxxxx > >> More majordomo info at http://vger.kernel.org/majordomo-info.html > >> > >> > > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > >
