mptscsih: ioc0: WARNING - Device (0:3:0) reported QUEUE_FULL!
 

Hi all.


One of my old virtual RHEL 3 servers (running on Vmware ESX) started giving these messages:

In addition, the CPU usage increased to close to 100%. To me it seems like this message indicates that the RHEL-server lost contact with it's disks for some reason. I'm not very into the details of SCSI operations, but is it so that there is a queue in which SCSI read and write operations are placed in, and that this queue in this case was full?

Check this (very old) post in the vmware communities:
http://communities.vmware.com/messag...B7DAA2EFB5EFD1

It may be worth the time to investigate your physical controller within your ESX server.

Thanks, I'll have our Vmware-admin check the physical controller.

I found this quote on the page you pointed me to:
Do you know what queue that's referred to here?

Sorry, can't seriously help you with this. It can be numerous things.
Judging from the information I found on the subject so far I'd start with the storage behind ESX.

Just some thoughts / questions (fishing in the dark):
- What kind of storage is used by the ESX server (SAN/NAS iSCSI/NFS)?
- if external ESX storage: Is the storage device busy with other tasks (snapshots/dedup/avscans)
- If NAS: how's the network load to that NAS
- Is the disk (0:3:0) in the RHEL3 VM mounted of used by another server (Cluster Filesystem)

Thansk for your pointer. We're using SAN storage, but now clustered file systems. It's a good possibility that the SAN is busy with other tasks, so I'm gonna have the SAN and Vmware admins take a look at it. Thanks for you help so far.