vmhba1:0:1:0 Retry (unit attn) ??

[rminick]

Anyone get this before? I can't find any problems with our fabric or fiber switches. It does not appear to loose connectivity to it's LUNS either.

I thought I would check here before opening a support request. I'm seeing a few of these error on various LUNS in the vmkernel logs every few days. The LUNS are presented to our ESX servers from an HP EVA8000.

Dec 18 20:54:26 hostname vmkernel: 19:03:10:36.443 cpu3:1044)SCSI: 8062: vmhba1:0:1:0 Retry (unit attn)

[bobross]

FWIW, you're not the only one seeing UA (unit attention) messages when using EVA 8000 arrays.

http://communities.vmware.com/message/817339

While there are many possible causes for the UA, without getting into sordid details with SCSI, the most likely culprit is a change in the EVA LUN configuration and/or pathing. The definitive way to troubleshoot these is to use a Fibre Channel analyzer.

Are you seeing these on all hosts that the EVA is connected to, or only this particular one?

[rminick]

I'm seeing it on all hosts. Our DR site has an EVA5000 and I see it there as well. Maybe the path is changing. There is 9 ESX hosts in 2 clusters.

[bobross]

That makes sense (9 hosts on 2 clusters). You rarely see this in a standalone single-path setting unless the LUNs are actually changing. UAs don't just happen for no reason. If you saw it on one and only one host, you might suspect its HBA, but it seems all HBAs are seeing the UA. What does your EVA log itself show?

You might also want to correlate your FC switch logs (and/or turn on lower-level error messages) to see if the switches are changing pathing and/or seeing the UA come through as well, and make sure your zoning is all copasetic.