OraFAQ Forum: RAC & Failsafe » Oracle 9i RAC node failed to take over

Home » Server Options » RAC & Failsafe » Oracle 9i RAC node failed to take over

Show: Today's Messages :: Polls :: Message Navigator
E-mail to friend

Oracle 9i RAC node failed to take over [message #127530]

Tue, 12 July 2005 01:51

sianz99
Messages: 3
Registered: July 2005

Junior Member

Current Situation
DB version : Oracle 9i RAC
2 node cluster sharing raw file system disk with only the archive log cross mounted in each local node.

Question:
However, it seemed that the RAC is not working as when 1 node is down, the other node is not responding as well. I suspect it could be due to the cross mounting of the archive log. How to reconfigure the archive log to be in the same share disk or is there any ways for the cluster node to take over?

Report message to a moderator

Re: Oracle 9i RAC node failed to take over [message #127541 is a reply to message #127530]

Tue, 12 July 2005 02:39

Frank Naude
Messages: 4579
Registered: April 1998

Senior Member

The best solution would be to use a Cluster Filesystem.

If you need to use NFS, ensure that each instance archives to a local filesystem. These filesystems can then be cross-mounted to the other systems.

Best regards.

Frank

Report message to a moderator

Re: Oracle 9i RAC node failed to take over [message #127544 is a reply to message #127541]

Tue, 12 July 2005 02:44

sianz99
Messages: 3
Registered: July 2005

Junior Member

Hi Frank,

This is implemented but however it seemed that the cluster do not work when 1 node is down..

Report message to a moderator

Re: Oracle 9i RAC node failed to take over [message #127548 is a reply to message #127544]

Tue, 12 July 2005 02:59

Frank Naude
Messages: 4579
Registered: April 1998

Senior Member

What is implemented - NFS or CFS?

If a node goes down, its NFS mounts on all other systems will hang. However, if you use LOCAL filesytems for archiving (instead of NFS mounted filesystems), it should not matter.

Please explain how your fileystems are setup in detail.

Best regards.

Frank

Report message to a moderator

Re: Oracle 9i RAC node failed to take over [message #127560 is a reply to message #127548]

Tue, 12 July 2005 04:08

sianz99
Messages: 3
Registered: July 2005

Junior Member

Hi Frank,

The detail setup is that the 2 nodes have their local archive logs but just that they are cross mounted so that when a backup is performed, both sides are being backup through 1 node.

Report message to a moderator

Re: Oracle 9i RAC node failed to take over [message #129888 is a reply to message #127530]

Wed, 27 July 2005 03:25

bluenectar
Messages: 4
Registered: July 2005
Location: Jakarta

Junior Member

hi frank,
I have same problem like sianz99 and I used OCFS for my shared disk. But when one of nodes goes down the other won't take over so users cannot login to Oracle Application. The ultimate solution is always restart all services on each machine APPS, CM and database.
Below is an excerpt of alert.log

Jul 24 04:13:38 RAC2 kernel: (25816) ERROR: status = -12, Common/ocfsgenmisc.c, 1106
Jul 24 04:13:38 RAC2 kernel: (25816) ERROR: status = -12, Common/ocfsgencreate.c, 627
Jul 24 06:12:35 RAC2 kernel: (25816) ERROR: status = -12, Common/ocfsgenmisc.c, 1106
Jul 24 06:12:35 RAC2 kernel: (25816) ERROR: status = -12, Common/ocfsgencreate.c, 627
Jul 25 04:02:03 RAC2 syslogd 1.4.1: restart.
Jul 25 04:11:08 RAC2 kernel: (5470) ERROR: status = -12, Common/ocfsgenmisc.c, 1106
Jul 25 04:11:08 RAC2 kernel: (5470) ERROR: status = -12, Common/ocfsgencreate.c, 627
Jul 27 14:07:59 RAC2 kernel: I/O error: dev 08:80, sector 0
Jul 27 14:08:02 RAC2 kernel: ocfs: Adding RAC1 (node 0) to clustered device (8,32)
Jul 27 14:08:12 RAC2 kernel: ocfs: Mounting device (8,32) on RAC2 (node 1)

is there any relation with kernel or OCFS version ?
thanks for advance.

best regards,
Hery

Report message to a moderator

Previous Topic:	recovery on rac
Next Topic:	RAC without shared storage

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Thu Apr 18 23:24:34 CDT 2024