vmotion problems

[iLikeMoney]

Simple setup, two host cluster ESXi 6.5.  I have 15-20 VMs on host A that I want to vmotion to host B in prep for ESXi patching.

I would say 10 of the VMs vmotion just fine.  The other 4 or 5 appeared to vmotion (no errors were generated or anything, shows as 100% completed) but afterward they remained on host A. !

In the web client I right-clicked the header row and added the column 'host' on the VM inventory, sure enough a handful of VMs still remain on host A.

Double-checked for some potential causes, there are no snapshots on these VMs and no ISO mappings.

Went back in and attempted to vmotion one of these stragglers for a second time, now host B is not even showing up in the dialog as an option to migrate to, it only shows 'host A' which is not helpful as it's the host it's already on.

I vmotion it to host A now just for grins. Completes successfully in a millisecond. Stuck in a loop with this... any ideas?

[jhague]

Are the networks identical on both hosts? VSS or VDS? Any hardware affinity? Any DRS anti affinity rules which could interfere/cause to be moved back? Anything in the VMware.log or vpxd.log? Maybe worth restarting the management services if strange things are still happening.

[iLikeMoney]

After further inspection it seems all of the VMs that failed to migrate were on LUN01, and Host B is having intermittent issues 'seeing' LUN01.  In the past I have done some action at the command line to get the signature squared away but apparently that fix did not persist.

Other 20 LUNs are fine, not sure why this one causes an issue, I bought a Compellent SAN from Dell, direct-connected to the hosts on SAS hopefully Compellent doesn't end up being a mistake.