there is memory leak bug in standby replay code, your issue is likely caused by it. Yan, Zheng On Wed, Jul 9, 2014 at 4:49 PM, Florent B <florent at coppint.com> wrote: > Hi all, > > I run a Firefly cluster with a MDS server for a while without any problem. > > I would like to setup a second one to get a failover server. > > To minimize downtime in case of failure, I want the standby server to > replay journal "on the fly". According to documentation, it seems possible. > > Here is my config : > > [mds] > mds data = /var/lib/ceph/mds/mds.$id > keyring = /var/lib/ceph/mds/mds.$id/mds.$id.keyring > mds cache size = 1000000 > debug mds = 0/20 > # debug ms = 1 > # debug journaler = 20 > > [mds.0] > host = host6 > mds standby replay = true > mds standby for name = 1 > > [mds.1] > host = host5 > > > Other configs related to MDS are defaults. > > mds.1 is my active MDS. > > When I start mds.0, after a few minutes, mds.1 is crashing (it was > running for days without any problem). > > Dump data is in attached file. If you need objdump of executable, just > tell me but I use the official Debian package one. :) > > What could have caused this problem ? > > Thank you a lot. > > _______________________________________________ > ceph-users mailing list > ceph-users at lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >
 |