On Fri, Nov 24, 2017 at 4:59 PM, Zhang Qiang <dotslash.lu@xxxxxxxxx> wrote: > Hi all, > > To observe what will happen to ceph-fuse mount if the network is down, we > blocked > network connections to all three monitors by iptables. If we restore the > network > immediately(within minutes), the blocked I/O request will be restored, every > thing will > be back to normal. > > But if we continue to block it long enough, say twenty minutes, ceph-fuse > will not be > able to restore. The ceph-fuse process is still there, but will not be able > to handle I/O > operations, df or ls will hang indefinitely. > > What is the retry policy of ceph-fuse? Is it normal for ceph-fuse to hang > after the > network blocking? If so, how can I make it restore to normal after the > network is > recovered? If it is not normal, what might be the cause? How can I help to > debug this? you can use 'kick_stale_sessions' ASOK command to make ceph-fuse reconnect, or set 'client_reconnect_stale' config option to true. Besides, you need to set mds config option 'mds_session_blacklist_on_timeout' to false. > > Thanks. > > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |