We have an Oracle VM Cluster that we are provisioning on a Cisco UCS B series blade chassis. It has 2 fabric interconnects (FC0 and FC1) provisioned to the (3) Oracle VM service profiles and it is set to boot from san with no local storage configured at all. The issue is that each time we reboot the Oracle VM Server, there is about an 80% chance that it will result in the below errors. I have been able to boot all three service profiles up at the same time with no issue at least three times throughout this process.
Setting up Logical Volume Management: /dev/sdq: open failed: No medium found /dev/sdp: open failed: No medium found
Checking filesystems fsck.ext3: Unable to resolve 'UUID=39b6935d-5c93-45aa-b54d-344fa171c40c'
This drops me to a 'Repair filesystem' prompt where a fsck gives me the same error. I am able to mount this uuid and view the data (from withing the Repair filesystem). Running 'blkid' shows that the boot partition and erroring partition's UUIDs match and they are mapped exactly between grub.conf and /etc/fstab.
As a test, I just zoned the fabric switches back to Single initiator, single target for this service profile, but this has provided no positive movement. Are there zoning or specific UCS service profile requirements that need to be met in order to boot from san with Oracle VM? Any help would be appreciated.