Unable to mount VMFS partition

[bweeks]

We are currently trialing the StarWind product in our lab environment. We have it running in standalone mode on two Dell servers. Just last week one of the datastores went offline and now we are unable to remount it. ESXi discovers it but when trying to mount it only offers the option to format it.
VMware had a look at the issue and have said it is because the underlying LUN has changed. They also confirmed only one specific version of the StarWind software is supported and only on ESX4 classic (not i).
Unfortuantly we had a number of VM's running on this storage which are now lost unless anyone has any ideas on how to repair this?
Here is the VMware analysis

As per our conversation, the partition table is set up correctly on the affected LUN, however, the change in size of the LUN, prevented us from remouting the datastore.
I could see a new signature in the hexdump output, however, even after creating new partition we were unable to mount the volume.
It also appears that this type of storage was not certified or tested with ESXi 5, as per VMware HCL
http://www.vmware.com/resources/compati ... tOrder=Asc
In order to recover the data, please contact your storage vendor to check whether he will be able to recover the data.
You can also contact 3rd party recovery company to assist you with the data recovery
http://kb.vmware.com/kb/1015413

I can confirm no engineers made any changes to the storage.

[anton (staff)]

Pretty classic case of a vendor fingerpointing Please clarify a) what StarWind version you're running EXAXTLY (including license type) and and 2) what target did you create (RAW, dedupe or IBV) and 3) submit failed node log to support@starwindsoftware.com (zipped please) so we could find out what happened. Thank you!

[bweeks]

Hi there and thanks for the response.
We are running an evaluation so are just using the free version at the moment.
The version number is 5.8.1889 and the plugins all report version 5.8
We created a dedupe disk 1.4TB big
This was formatted in ESXi as VMFS 5
I have emailed the log file as requested.

[Anatoly (staff)]

Thank you for your email.

First of all I see that you are using not the latest build of our software, so I`d recommend you to update it and see if something will happen.

Also, you`ve sent the logs starting from February `till now. could you clarify approximate time of the issue?

[bweeks]

Hi Anatoly

Do you have any instructions for performing the upgrade to the latest version?
The event happened around the 25th April

Regards

Barry

[Anatoly (staff)]

Since you are upgrading from 5.8 to 5.8 the procedure is next:
1.Install latest build on StarWind1
2.Wait until the end of synchronization (FastSync should appear)
3.Install latest build on StarWind2

Can we also see the your topology diagram with all the IPs and everything else that is related to SAN for better understanding the situation please?

[derekdolan]

I'm having this same issue in my lab environment.

The Starwind host is a Dell PowerEdge R710 running Windows Server 2012, Starwind version 6.0.4837. The image files are located on an external Fibre Channel array managed by Windows Storage Spaces. The two ESXi hosts are attached via a separate Layer 2 network for iSCSI traffic.

I suspended all VMs and shut down the Starwind host in order to move it from one rack to another. When the server came back up, I rescanned the HBAs in ESXi and the paths came back up. The VMFS volumes did not come back, however. I rebooted the ESXi hosts to see if something was wonky there, but I’m facing the same issue after that. If I try to “Add Storage” in vCenter, it wants to format the volume and start over.

Please let me know what additional information you might need to help point me in the right direction.

[Max (staff)]

In this case you need to connect to the vSphere host direcly, bypassing the vCenter.
Remove the IP's of the SAN from both dynamic and static discovery and conduct a rescan.
After that add the IP address back to the dynamic discovery, run another rescan and re-add the datastore.
It will either ask you to resignature or allow you to leave the old Datastore signature.

[derekdolan]

The vCenter server isn't online (it's on the VMFS volume I can't access), so this is all done via the vSphere client directly on the ESXi hosts.

I went through the procedure as directed. When I went to mount the volume again, I had the option of using the existing signature. I did that and it the datastore did not show up. I went to add it again, and this time picked to re-signature the volume. That process errored out, and now it's back to looking like an unformatted disk to ESX. I went through the process again (as recommended) and also restarted the Starwind service. No luck with any of that.

[Anatoly (staff)]

I have couple of clarification questions if you dont mind:

I did that and it the datastore did not show up

How exactly it was looked like please (i.e. the datastore was without any data, or something else)?

and this time picked to re-signature the volume. That process errored out, and now it's back to looking like an unformatted disk to ESX

Well, good attempt, but desparate clicking is not always the best way. What was the error please?