On 07/18/2012 02:32 PM, Liu Yuan wrote: > On 07/18/2012 02:11 PM, Dietmar Maurer wrote: >> Jul 18 07:57:37 [main] cluster_sanity_check(501) joining node epoch too >> large: 13 vs 12 >> Jul 18 07:57:37 [main] cluster_wait_for_join_check(524) transfer mastership >> (13, 12) >> Jul 18 07:57:37 [main] sd_check_join_cb(921) 192.168.2.2:7002: ret = 0x3, >> cluster_status = 0x4 >> Jul 18 07:57:37 [main] __corosync_dispatch_one(312) failed to join sheepdog >> cluster: please retry when master is up >> Jul 18 07:57:37 [main] crash_handler(402) sheep pid 27848 exited unexpectedly > >> Which one do you mean? > > I just saw above log and didn't look into the issue yet. Node joining > logic has been vastly refactored by you, so I think you would know better. > > Thanks, > Yuan >
Is it the correct logic we expected when master transfer happens? As a node with smaller epoch starts firstly, starting a node with bigger epoch would cause master transfer, this would make the node with smaller epoch exit and user should try to start it again. thanks, levin -- sheepdog mailing list [email protected] http://lists.wpkg.org/mailman/listinfo/sheepdog
