[ Re-adding list ] On Mon, May 5, 2014 at 8:33 PM, [email protected] <[email protected]> wrote: > When I invoke "mark_unfound_lost ", it had a error: > > [root@ceph2 ~]# ceph pg 50.5 mark_unfound_lost revert > Error EINVAL: pg has 43 unfound objects but we haven't probed all sources, > not marking lost > > What can I do?
Well, apparently it thinks there are some extant copies of the PG which are down but still existent. If you've lost an OSD (including by re-imaging and re-adding it), you should mark the OSD as lost. If not, you should turn on the OSDs that are down. -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com > > ________________________________ > [email protected] > > From: Gregory Farnum > Date: 2014-05-06 05:06 > To: [email protected] > CC: ceph-users > Subject: Re: [ceph-users] some unfound object > "Need" means "I know this version of the object has existed at some > time in the cluster". "Have" means "this is the newest version of the > object I currently have available". If you're missing OSDs (or have > been in the past) you may need to invoke some of the "lost" commands > to tell the OSDs to just go with what they have. > -Greg > Software Engineer #42 @ http://inktank.com | http://ceph.com > > > On Mon, May 5, 2014 at 2:01 AM, [email protected] <[email protected]> > wrote: >> Hi everyone, >> my ceph has some object unfound. When I run "ceph pg 50.5 list_missing", >> it >> show me: >> >> ... ... >> { "oid": { "oid": "rbd_data.53ec83d1b58ba.0000000000000740", >> "key": "", >> "snapid": -2, >> "hash": 4097468757, >> "max": 0, >> "pool": 50, >> "namespace": ""}, >> "need": "27229'2203677", >> "have": "27064'2202729", >> "locations": []}, >> ... ... >> >> I want to know, what's the "need" and "have"? Can I change it? Or how to >> fix >> it? >> ________________________________ >> [email protected] >> >> _______________________________________________ >> ceph-users mailing list >> [email protected] >> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >> _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
