Le 15/05/2016 15:35, Emmanuel Lacour a écrit : > Dear ceph users, > > > I have clusters running debian wheezy with dumpling. > > I upgraded one cluster from dumpling to firefly, then to hammer without > problem. > > Then I upgraded a second cluster from dumpling to firefly without > problem, thought I forget to restart 2 osds/10 so they stayed in dumpling. > > I then did the firefly->hammer upgrade. 8/10 osd restart without problem > (including those that where running dumpling). > > 2 osd does not wan't to start and despite a rep_size of 3, I have lots > of pg down+peering :( > > > Here is the log (debug 20) of one of those osd: > > > http://owncloud.home-dn.net/index.php/s/jROrqr2nuuVL8kZ > > Any help will be really appreciated! > I finally just found the solution, the osdmap was corrupted on those osd (don' t know why), I copied all current/meta/DIR* from another osd et health is OK know. I still check if there is no data loss ... I found this solution after reading this: http://tracker.ceph.com/issues/14406, which has the same symptoms _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |