symmpi errors on Windows 2003 using MSA-1000 LUNS

[Kimik]

Hi There,

I'm trying to troubleshoot some issues with a Windows 2003 server that's using MSA-1000 storage. It's an exchange 2003 server that clients reporting freezing for seconds a few times a day. The only suspect event log entries on the Windows side is:

15 The device, \Device\Scsi\symmpi1, is not ready for access yet.

11 The driver detected a controller error on \Device\Harddisk1.


These occur about 30 times in a row, twice a day. I've correlated this with backup times etc, the only clues I have is 9.00 is about peak time and 4.30 is around the time our exchange backups kick off.

I'm wondering how to troubleshoot this problem?

[Kimik]

I've just found on the Switch2/8 console that portShow 1 returns a lot of loss_of_sync errors for the two HBA's that are in the server.

1: 1.8g 770m 0 0 0 0 0 121k 0 0 36 5 0 0
2: 2.2m 1.1m 0 0 0 0 0 125k 0 0 242 6 0 0


36 and 242 respectively. Is this a concern? The other two hosts connected to the same MSA only log 1 and 2 respectively.

[Uwe Zessin]

The SYMMPI is for a parallel SCSI interface and not directly related to the MSA1000. You might want to verify if you have a current device driver, because there have been problem reports in the past.

About the switch errors...
any QLogic adapters (e.g. FCA2214/DC) in use?

[Kimik]

Yes it's a qlogic.

QLogic PCI to Fibre Channel Host Adapter for QLA2340 :
Firmware version: 3.03.08, Driver version 7.04.00



kim

[Uwe Zessin]

Thought so ;-)

Unfortunately, the adapter spits out nonsense when the driver is not loaded. HP has a advisory which describes what jumper on the adapter to move in order to turn off the laser if that is a problem.

I would check if the errors increase while the server is running the operating system.

[Kimik]

Do yo have any information on this issue that I can use to test in our testing environment?

I'm trying to determine - Could these link sync errors be causing the OS to hang?

Note the Windows server is actually a virtual machine, running on VMWARE ESX...

[Uwe Zessin · Accepted solution]

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_OE050608_CW02&printver=true

I would first check if the sync errors occur while running the system.

What OS are you talking about? ESX server or Windows 2003 in the VM?


Have you already taken a look into the VMware logfiles for the kernel and the VM?

[Kimik]

Yes it's ESX - one of three ESX hosts connected to the same MSA-1000. The whole environment is very light, not too many VMs running. I've spent the last couple of days looking into the vmkernel logs, /proc nodes etc - and nothing untoward is logged regarding excessive HBA queue lengths and the like. Hence why I'm now looking into the MSA-1000 itself....


thanks for your help - I'll look into this advisory...

[Kimik]

Changing the SCSI Adapter driver type from 1020/1030 to PCI-X fixed this. This seems to be some issue with VMWARE Tools installing the XP/.NET driver type instead of the PCI-X version of the symmpi driver.

more info:

http://www.vmware.com/community/thread.jspa?messageID=399737᥹