uccq@uhnix2.UUCP (Alan Pfeiffer-Traum) (11/10/87)
We have a VAXCluster with 4 nodes sharing a common system disk. We do a
moderate amount of logging on each system to the operator log via the SET
AUDIT facility. We also have a VAXCluster Console system functioning as the
console for each of these systems. Since all operator messages from each node
in a VAXCluster go to all of the consoles in the cluster, and all operator log
entries from each node in a VAXCluster go to all of the operator log files on
each node, my questions are:
1 - Has anyone come up with a good way to not log all of the operator log
messages to all of the consoles in a VAXCluster? In our case all of the
consoles are the same terminal; and even worse, all of the messages are being
kept on the console's hard disk.
2 - Has anyone come up with a good way to not log all of the operator log
messages on each node to all of the other system's operator log files? If we
could keep all of the operator log messages for a VAXCluster in one place, that
would be fine.
By "a good way to..." I mean a method that will tolerate members of the
VAXCluster being removed and added; and I cannot tell my student-operators to
watch closely and take action whenever there is a possibility of operator
messages and/or logs being lost.
I understand that I can disable the message to consoles by doing
$ REPLY /DISABLE, and that I can stop logging by using $ REPLY /NOLOG, but I
need to be able to tolerate VAXCluster state transitions.
Alan Pfeiffer-Traum @ University of Houston
UUCP: ...{shell,academ}!uhnix1!uccq
BITNET: UCCQ@UHVAX1
PHONE: (713) 749-2761