Datastores not seen after 3.5u3 upgrade

[Jam111]

Hi,

After ages looking (and a call open with HP/VMware) I thought I'd throw this out there...

I have 2 Farms, both with an upgraded VC (2.5u3). 1 farm has been upgraded from 3.0.1 to ESX3.5u4 (via u3) and is working fine. The 2nd farm has had 2 test ESX servers upgraded from 3.0.1 to 3.5u3 and 3.5u4. These two servers have lost sight of the datastores. They were previously fine before the upgrade and the VMs are still running on other 3.0.1 hosts, although the storage did disappear after a reboot and the 3.0.1 host would have to be rescanned.

The hosts use the same physical kit in both farms (HP DL585 G2) with same HBAs (HP branded FC1242 Emulex1150). Both SANs are similar DMX2, the only difference being seen is the working farm is connecting using ANSI SCSI Revision 04, broken one ANSI SCSI Revision 02. The ESX Host knows it is connecting to a SYMM6 device, so it can talk to the SAN, it just cannot see the datastore. I have had the masking removed and replaced.

Only other major difference is in fabric. Working farm is using Brocade switches, broken farm using CISCO m9500. I don't see any VMware guides to fabric, just HBA and SAN array. I've tried rescanning, LUN resignaturing, HBA firmware update etc. I can only think logically and compare the differences in setup. I read from the VMware Fibre Channel SAN Configuration Guide that SCSI 3 (SC3) should be enabled. Is this related to the above ANSI revisions? Any thoughts? If it is a storage issue, I need to prove it to them...

Cheers, James

[AndreTheGiant]

Have you tried to create a new LUN and see whats happen?

Andre

**if you found this or any other answer useful please consider allocating points for helpful or correct answers

[Chamon]

Can you post the results from

fdisk -l

When you look into the /vmfs/volumes/ directory do you see anything?

[Jam111]

Have you tried to create a new LUN and see whats happen?

Andre

**if you found this or any other answer useful please consider allocating points for helpful or correct answers

Good thought. I have had a new LUN created with no formatting etc. It is not seen by the Host, rescanned and rebooted. I don't think the host even sees the "storage gatekeepers".

[Jam111]

Can you post the results from

fdisk -l

When you look into the /vmfs/volumes/ directory do you see anything?

Hi Chamon,

I have attached an fdisk screenshot. Only sees the local harddrive and whatever /dev/sda might be. Under /vmfs/volumes is just the local drive and associated shortcut.

Cheers

James

[AndreTheGiant]

Here there is only you local disk an a small disk of 47M ?! (is suppose some internal flash?)

Are you you that your hardware is in HCL?

Andre

**if you found this or any other answer useful please consider allocating points for helpful or correct answers

[Jam111]

Hi Andre,

The HBAs and the DMX/DMX2 are in the HCL. The small disk is on every build and I have seen it on Linux builds as well. I'm just confused as to why everything is fine in one farm and not in another. The only differences appear to be the fabric and some config within the DMX. But trying to convince the storage guys to cross check everything is tedious! The fact it worked as 3.0.1 but not 3.5 is more strange. I am thinking of rebuilding a server directly to 3.5 to see if that proves anything.

Cheers

Jam

[AndreTheGiant]

The small disk is on every build and I have seen it on Linux builds as well.

But is a SAN disk? (is like a LUN0)?

In this case zoning (switch fabric) is working, and your problem is with LUN masking (on storage side).

Andre

**if you found this or any other answer useful please consider allocating points for helpful or correct answers