RHEL4 bl45p qla2xxx and oracle

[Marius Pana_1]

Hi All,

I recently installed a few HP blade servers *BL45p) running RHEL4 Update 2 with QLogic FC HBA (2312) with the qlogic driver version 8.01.03-p4-fo.

Recently ORACLE installed oracleasm (not sure what it actually does) and they seem to be getting kernel panics when doing IO intensive operations when using the SAN storage (MSA100) attached via the qlogic adapters. I only have a couple of screenshots to go on and they are attached.

I am wondering on what kind of tests I could do to see if it is a problem with the qlogic drivers or whether it is with the ORACLE software?

Any suggestions will be greatly appreciated.

[Marius Pana_1]

Cant seem to be able to upload pictures. Trying again..

[Marius Pana_1]

And the other screenshot...

[Alan_152]

The 1st things I would do would be to upgrade to RHEL4U3, go to the latest QLA drivers, flash the latest QLA firmware, make sure that the server's firmware is up to date, and make sure that both the MSA1000's firmware is up to date and its ACU is up to date.

Another thing to look at would be to sniff around hazard.boi.hp.com . hazard is a storage testing suite, and a lot of people find issues with various HBAs that they talk about on the boards there.

[Ivan Ferreira]

You should open a call to metalink or Red Hat If you have a subscription.

You may analyze the oops using ksymoops, not a easy task. You may generate I/O using:

hdparm -Tt
dd if=/dev/zero of=bigfile bs=8192 count=131072 &
dd if=/dev/zero of=bigfile bs=8192 count=131072 &
dd if=/dev/zero of=bigfile bs=8192 count=131072 &
dd if=/dev/zero of=bigfile bs=8192 count=131072 &
dd if=/dev/zero of=bigfile bs=8192 count=131072 &