VMFS volume keeps changing apparent size

[tomaddox]

A little while back, I made a mistake on the VMFS volume that houses most of my VMs. Namely, I extended the VMFS volume onto a new LUN, realized that I had overprovisioned the volume, and deleted the LUN. Now the VMFS volume shows up as being the size it was when the LUN was added (2.2 TB), but if I rescan the storage adapters, the volume shows up as being the correct size, 1.7 TB. After a few minutes, however, the apparent size returns to 2.2 TB. This is problematic for a couple of reasons: first of all, it would be nice to see the correct size of the volume, but more seriously, I have started to see problems accessing newly-written files.

Does anyone know of a way I can permanently resolve this problem?

[Jae_Ellers]

Scary stuff. I'd open an SR while I backed up my VMs.

I bet support can help you recreate the partition table. If you can get your hands on VMworld 2007 TA48 http://www28.cplan.com/cbv_export/PS_TA48_288131_166-1_FIN_v2.pdf that has instructions on backing up the VMFS header & partition data. You might also be able to figure out how to resize the lun.

Do you have any vm-support output from the time when you lun was the right size? vm-support outputs have a copy of the necessary info for putting the lun boundaries back in the right place.

Of course I'm sure you don't have space for the obvious copy it someplace and fix it answer.

[tomaddox]

Alas, I don't have any vm-support output from that period. The apparent size of the datastore has remained steady (and correct) over the weekend, however, and I'm hoping that it will remain so. If it doesn't, I definitely will open a case.

[tomaddox]

The answer that I got from VMWare support (HP support, actually, through whom I have my service contract) is that there is no solution other than to migrate my data off the datastore and recreate it or get rid of it.