topic Re: Kernel Panic on DL380 G4 Packaged Cluster with MSA 500 in Operating System - Linux

Kernel Panic on DL380 G4 Packaged Cluster with MSA 500

pratik_patel — Wed, 03 Jun 2009 14:11:43 GMT

We have RHEL 3.0 U5 running on a dual DL380 G4 Cluster with dual paths to a HP MSA 500 disk array.

It has Oracle 10g, OCFS and Java 1.5 with our application running on it.

We are getting 'Kernel Panic' Error while booting Array and node with onboard SCSI cable.

Attached is the screenshot of the problem <>

Can you please advise how we start debugging this problem?

Re: Kernel Panic on DL380 G4 Packaged Cluster with MSA 500

Ivan Ferreira — Wed, 03 Jun 2009 18:10:30 GMT

I would start with the installation media in "rescue" mode to try to identify the reson why the "raidstart" command is failing.

Do you have software raid on these nodes? Is the operating system installed on local disks?

Re: Kernel Panic on DL380 G4 Packaged Cluster with MSA 500

Steven E. Protter — Wed, 03 Jun 2009 18:36:36 GMT

Shalom,

Bring down the server, bring down the array. Check the connections.

Bring the array up first, maybe with new cables, then bring up the system.

SEP

Re: Kernel Panic on DL380 G4 Packaged Cluster with MSA 500

pratik_patel — Fri, 05 Jun 2009 06:48:10 GMT

We have two paths to the disk array, one via an onboard SCSI controller and one via a SCSI controller on a PCI card.

=> controller all show
MSA500 G2 at M4NTMT424D (sn: P85AC0DX3RJ0I8)

Smart Array 642 in Slot 1 (sn: P92260P9SRR03N)

Smart Array 6i in Slot 0 ()

=> controller all show status

MSA500 G2 at M4NTMT424D

Controller Status: OK
Cache Status: OK
Battery Status: OK

Smart Array 642 in Slot 1

Controller Status: OK
Cache Status: NotConfigured

Smart Array 6i in Slot 0

Controller Status: OK
Cache Status: OK

We have tried booting the server with only one path connected. Booting with one path works OK, booting with the other path causes the panic.

We can try rescue mode to get more information.

Re: Kernel Panic on DL380 G4 Packaged Cluster with MSA 500

Steven E. Protter — Fri, 05 Jun 2009 09:52:25 GMT

Shalom,

Two general areas you want to check.

SCSI/disk infrastructure. That is my first post.

It could however be application related. You might want to see if Oracle 10g is contributing to the problem. metalink. bugs.

SEP