> > 2. I tried to create the split-brain on a 2-node cluster and then > recovered from it. I did see the crmd killed and started again. I did see > the core file generated. But this time, I only saw the back trace of: > > Core was generated by `/usr/lib64/heartbeat/crmd'. > Program terminated with signal 6, Aborted. > #0 0x00000039d642e21d in ?? () > (gdb) bt > #0 0x00000039d642e21d in ?? () > #1 0x00000039d642fa1e in ?? () > #2 0x0000000000000020 in ?? () > #3 0x0000000000000000 in ?? () > > Therefore, I am not sure if it is the same error I got like previous. > Anything I can do to get more info from the core file? Also, in the cluster > we got the first core file, it is not likely to have the split-brain > problem. Any other possible scenario could cause this problem? > > Please ignore the above. That was the stack without any resource managed. With resources added, I got the same stack. heartbeat[20053]: 2008/02/08_11:49:07 ERROR: glib: Unable to send [-1] ucast packet: No such device heartbeat[20053]: 2008/02/08_11:49:07 ERROR: write failure on ucast eth1.: No such device heartbeat[20049]: 2008/02/08_11:49:08 CRIT: Cluster node ha2.roundbox.comreturning after partition. heartbeat[20049]: 2008/02/08_11:49:08 info: For information on cluster partitions, See URL: http://linux-ha.org/SplitBrain heartbeat[20049]: 2008/02/08_11:49:08 WARN: Deadtime value may be too small. heartbeat[20049]: 2008/02/08_11:49:08 info: See FAQ for information on tuning deadtime. heartbeat[20049]: 2008/02/08_11:49:08 info: URL: http://linux-ha.org/FAQ#heavy_load heartbeat[20049]: 2008/02/08_11:49:08 info: Link ha2.roundbox.com:eth1 up. heartbeat[20049]: 2008/02/08_11:49:08 WARN: Late heartbeat: Node ha2.roundbox.com: interval 81190 ms heartbeat[20049]: 2008/02/08_11:49:08 info: Status update for node ha2.roundbox.com: status active heartbeat[20049]: 2008/02/08_11:49:09 info: all clients are now paused heartbeat[20049]: 2008/02/08_11:49:09 debug: hist->ackseq =4599 heartbeat[20049]: 2008/02/08_11:49:09 debug: hist->lowseq =4597, hist->hiseq=4700 heartbeat[20049]: 2008/02/08_11:49:09 debug: heartbeat[20049]: 2008/02/08_11:49:09 info: all clients are now resumed heartbeat[20049]: 2008/02/08_11:49:11 WARN: 1 lost packet(s) for [ ha2.roundbox.com] [4782:4784] heartbeat[20049]: 2008/02/08_11:49:11 info: No pkts missing from ha2.roundbox.com! heartbeat[20049]: 2008/02/08_11:49:11 WARN: Exiting /usr/lib64/heartbeat/crmd process 20088 killed by signal 6 [SIGABRT - Abort]. heartbeat[20049]: 2008/02/08_11:49:11 ERROR: Exiting /usr/lib64/heartbeat/crmd process 20088 dumped core heartbeat[20049]: 2008/02/08_11:49:11 ERROR: Respawning client "/usr/lib64/heartbeat/crmd": heartbeat[20049]: 2008/02/08_11:49:11 info: Starting child client "/usr/lib64/heartbeat/crmd" (502,503) heartbeat[23765]: 2008/02/08_11:49:11 info: Starting "/usr/lib64/heartbeat/crmd" as uid 502 gid 503 (pid 23765)
The log file confirms the crmd was killed and respawn. Question remained: it is not like to lose the connection between the two nodes in our environment. Is there any other scenarios will cause this? _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
