Oracle 8.1.7 OPS stability issue
We managed to install Oracle OPS 8.1.7.0.1 (no 8.1.7.4 installed yet) on Hp Proliant DL380 cluster.
However, it is VERY unstable, almost every time we try to shutdown database in one node, the other
node will reboot. watchdogd seems doing this for consistency / integrity reasons, but it renders
the whole thing not ready for prime time.
We've got the watchdogd timeout set to 60 seconds, nowayout=1 with Linux Watchdog modules.
This is on RH7.2, LVM 1.0.5, shared SCSI storage.
name = /tmp/.watchdog/cl_sock_14883_21513,
pid = 14883,
tid = 21513,
margin = 60000,
level = 1,
option = 0,
description = ClientProcListen.
Time: Thu Aug 22 11:08:09 HKT 2002 (910212)
UTC: Thu Aug 22 03:08:09 GMT 2002 (910212)
wddSendRegisterReply: info: sent register ack to client.
Time: Thu Aug 22 11:09:03 HKT 2002 (933625)
UTC: Thu Aug 22 03:09:03 GMT 2002 (933625)
wddProcAbortAllPacket: fatal: client (name=cl_sock_14878_16388) requested abort.
Time: Thu Aug 22 11:09:03 HKT 2002 (933816)
UTC: Thu Aug 22 03:09:03 GMT 2002 (933816)
Shutting down the entire node...
Time: Thu Aug 22 11:09:09 HKT 2002 (060017)
Any help and comment is appreciated.
Regards,
Chris