I have a problem where the same 4 SATA based datastores are missing following a reboot of any ESXi 3.5 host. A recan of the HBA's fixes the problem first time, but I'd like to be able to reboot hosts without doing this.There are active VM's on the LUNS, which work perfectly after the rescan.There are 43 LUNS total a mixture of SAS and SATA, with all others working fine.
There are 4 stacks of MSA 2000 based storage attached the 4 hosts, all connected to the same dual SAN fabric. This was all built and configured by myself so I'm fairly sure the storage and zoning is correctly configured. I'm only seeing this problem on LUNS from one of the MSA2000's G2's
I'd try removing and recreating the LUNS but there's a lot of data on them I'd rather not move. (Almost all LUNS are 2TB). Most of the LUNS have a single large VMDK on them so there should be little contention for LUNS between hosts.
Logs are showing "Failing I/O due to too many reservation conflicts" when trying to scan the storage at boot. This only happens on reboot, which is the only time when the LUNS are missing.
Anyone have any suggestions on how I can resolve this?
