Expand thick device leads to corruption?

[haslund]

Hello!,

I created a target with 2 devices.
Both devices were created as 13 GB thick with 1024 MB L1 cache.
I then expanded both devices from 13 GB to 20 GB.

In ESXi (5.5) I added the LUNs as individual datastores (VMFS-3) and started creating new virtual machines.
Up to the first 13 GB I did not experience any issues, however going above the 13 GB usage and I started to see wierd issues with the VMs.
Installation of Windows failing with corrupt ngen.exe file, bluescreens, install finishing but then during boot it said drive needed to be checked.

Is this something you have tested/are aware of?

[haslund]

First boot after unattended installation:

Capture.PNG (144.57 KiB) Viewed 32173 times

[haslund]

Capture.PNG (228.63 KiB) Viewed 32169 times

[haslund]

I have digged a bit more into this, very strange.
Same thing happens with new clean LUN that was not expanded.

If I create a VM on local storage on the ESXi host and then storage vmotion it to the Starwind iSCSI lun it continues to work fine without any corruption.
However if I create a new VM on the Starwind iSCSI lun and install Windows Server 2003 using an unattended ISO it does shows the screenshots above with corruption issues.

If you need access to my lab to perform tests or something like that just let me know.

[Bohdan (staff)]

This is really strange.
Yes, remote session may help us to find what's wrong.
I'll send you e-mail and we schedule a meeting.
Thank you.

[haslund]

Bohdan: I did not see any e-mail from you, did you send it already?

[Bohdan (staff)]

There were 3 e-mails. I sent them to the e-mail you specified for StarWind forum registration. No response from you. Maybe you should check your spam box...

BTW, I have updated the build.
http://starwind.com/tmplink/StarWindBeta.exe

[haslund]

I sent you an e-mail with info on remote access to my lab environment where you can reproduce the issue.
Let me know if you did not get it.

[haslund]

I just installed the updated build.
Before upgrade: 8.0.6441
After upgrade: 8.0.6694

This issue seems to be resolved now.

Thanks!

[Bohdan (staff)]

Thanks for keeping us updated.