Thanks for the quick reply. Ok, so at this time looks like better to avoid split networks across network interfaces. Where can I find list of all issues related to the concrete version?
On Mon, Mar 11, 2013 at 5:16 PM, Gregory Farnum <[email protected]> wrote: > On Monday, March 11, 2013, Igor Laskovy wrote: > >> Hi there! >> >> I have Ceph FS cluster version 0.56.3. This is 3 nodes with XFS on disks >> and with minimum options in ceph.conf in my lab and I do some crush >> testing. >> One of the of several tests is lost connect to replication network only. >> What expect behavior in this situation? Will mounted disk on client >> machine frozen or so? >> >> Look like in my case whole cluster have gone crazy. >> > > Yeah, this is a known issue with the way Ceph determines if nodes are up > or down. Basically the OSDs are communicating over the replication network > and reporting to the monitors that the disconnected node is dead, but when > they mark it down it finds out and insists (over the public network) that > it's up. > > I believe Sage fixed this issue in our development releases, but could be > misremembering. Sage? > -Greg > -- Igor Laskovy facebook.com/igor.laskovy Kiev, Ukraine
_______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
