VMware checkunitready returns errors (HP EVA8400 Storage)

[SAKET_5]

Hi,

We are seeing a large number of the following warning messages in the vmkernel log on ESX 3.5 servers:
Apr 26 12:03:05 udcvm4 vmkernel: 144:15:50:02.586 cpu0:1047)WARNING: SCSI: 2959: CheckUnitReady on vmhba1:8:9 returned Storage init
iator error 0x7/0x0 sk 0x0 asc 0x0 ascq 0x0
Apr 26 12:03:05 udcvm4 vmkernel: 144:15:50:02.586 cpu0:1047)WARNING: SCSI: 2959: CheckUnitReady on vmhba1:9:9 returned Storage init
iator error 0x7/0x0 sk 0x0 asc 0x0 ascq 0x0
Apr 26 12:03:05 udcvm4 vmkernel: 144:15:50:02.586 cpu1:1047)WARNING: SCSI: 2959: CheckUnitReady on vmhba1:10:9 returned Storage initiator error 0x7/0x0 sk 0x0 asc 0x0 ascq 0x0

A number of VMs went offline and had to be rebooted in order to bring them back. The guest OS logs contain these errors:

"The device, \Device\Scsi\symmpi1, is not ready for access yet."
"The driver detected a controller error on \Device\Harddisk0."

EVA8400 logs look clean - no hard drive failures, SAN Fabric logs look clean too - no events indicating Fabric rebuilds, etc.

The EVA Vdisks used by the VMs as RDMs are in a CA replication relationship and also get snapcloned at various schedules.

Any suggestions? We are in the process of logging calls to both VMware & HP.

Regards.

[Uwe Zessin]

http://kb.vmware.com/kb/1003433

mentions this error, but does just advise to involve support - which you're already doing. Let us know if anything useful is found.

[CLEB]

I had this issue with ESX 3.5 update 4 and EVA4400. VMware support told me it is a known issue which is fixed in Vsphere 4.