I don't know if it is related to 3.4.0alpha or 3.4.0qa8, but when I upgraded, one of the peers did not restart because its file in /var/lib/glusterd/peers was corrupted this way: uuid=639e4725-eadb-485f-9ffb-930103a466b0 state=3 hostname1=ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ Restoring hostname1 to a sane value fixed the problem. Now it works, but logs are filled with errors: glusterd.vol.log on a peer [2013-02-13 13:45:57.779047] E [socket.c:2767:socket_connect] 0-management: connection attempt failed (Connection refused) [2013-02-13 13:45:57.779300] E [socket.c:2767:socket_connect] 0-management: connection attempt failed (Connection refused) glusterd.vol.log on another peer: [2013-02-13 13:46:52.915638] E [socket.c:2767:socket_connect] 0-management: connection attempt failed (Connection refused) [2013-02-13 13:46:52.915677] E [socket.c:2767:socket_connect] 0-management: connection attempt failed (Connection refused) [2013-02-13 13:46:53.298917] E [rpcsvc.c:448:rpcsvc_check_and_reply_error] 0-rpcsvc: rpc actor failed to complete successfully [2013-02-13 13:46:53.299000] E [rpcsvc.c:448:rpcsvc_check_and_reply_error] 0-rpcsvc: rpc actor failed to complete successfully -- Emmanuel Dreyfus manu@xxxxxxxxxx
