VMs on NAS

[Andrew1234]

Hi,

I am having issues with my VMs corrupting every couple of days on an ESXi bare metal setup.

I have about 7 or 8 VMs, one is a domain controller (windows server 2012), and the rest are dedicated workstations for groups of applications (windows 10), and one Linux Ubuntu VM.

I have had the most corruptions on the domain controller, but the others including the Linux machine have almost all corrupted at some point, which usually has resulted in me having to do a fresh install or full VM restore from a weekly backup.  By corruptions, I mean the OS tries to start goes into some sort of repair mode, or error mode, etc., and cannot load the OS, even in safe mode.

The VMs are stored on a NAS.  The NAS is split up into a iSCSI LUN with Thick Provisioning for my VM datastore, and the rest is just for simple network file storage on a separate volume.  The NAS is set up for RAID 5 and also has a scheduled backup every week onto a USB locally attached HDD.  The corruptions do not seem to coincide with the backing up of the LUN, they just seem random.  The NAS itself I would say is working fine, since there are no faults/errors, and all of my other file storage has had not issues.

Any ideas what could be causing this?

[vmrale · Accepted solution]

Andrew1234​,

how many ESXi hosts do You have?

How many vmkernel ports responsible for iSCSI communication are configured per host?

Are iSCSI vmkernel ports using IP addresses from the same subnet?

[Andrew1234]

Hi vmrale,

Thanks for the response.  By actually thinking about the answers to your questions and trying to think why you wanted this specific information, it jogged my memory - I only have one ESXi host...BUT I had once tested the LUN with my desktop machine prior to connecting it to the ESXi Host (but then disconnected it).  So I checked my desktop machine and lo and behold, my desktop was also connecting to the LUN without me realising it all of this time (months).  I had disconnected it after testing it, but apparently windows 10 doesn't work that way, it just reconnects to the LUN when the pc is rebooted.  I had to delete it from my favourite targets list before it wouldn't automatically start up on boot.

I hope this fixes it, I still need to try and see for a while, but multiple concurrent connections to a LUN can't be good, which is what I think you were alluding to.

Thanks for the help, I'll be back if I still have further issues, but I suspect I wont.

[vmrale]

You did it! 🙂