Quote:
Originally Posted by
filosophizer
[...]Broadcast message from root@pr2serv (tty) at 00:01:04 ...
Cluster Verification detected cluster configuration errors on node pr1serv. Detailed clverify output is available in file /var/hacmp/log/autoclstrcfgmonitor.out on node pr1serv.
more autoclstrcfgmonitor.out
autoclstrcfgmonitor.out: A file or directory in the path name does not exist.
# pwd
/var/hacmp/log
[...]
You use /var as filesystem(*) for cluster logs which filled this filesystem to 100% subsequentally leading to no space being available for autoclstrcfgmonitor.out. You are lucky that that node did not crash completely as AIX is not amused when either of /, /var and /tmp is filled a 100%. It will prevent your node from booting or might crash it during operation. Hence create a separate filesystem, size it adequately (start with 512M) mount it on /var/hacmp/log and make ALL cluster logs use that (with cllog command or from CSPOC).
(*) Using /var for logs is really stupid however that is not your fault but IBM's. They should know better but are using /tmp and /var as a default filesystem for clusterlogs for ages now.