Tuesday, April 16, 2013

VMware vSphere virtual machines constantly runs a CHKDSK after every reboot

Ran into an interesting problem today while upgrading an existing vSphere 4.1 infrastructure to 5.1 with the latest build (1021289). What we noticed was that when all of the virtual machines that were migrated over to the new 5.1 environment would run a chkdsk every the Windows 2008 R2 operating system was rebooted:


… and whenever we ran a chkntfs c:, the return would be that the disk was dirty. After going through trial and error by moving virtual machines from patched to non-patched ESXi 5.1 hosts without any luck, we opened up a support ticket with VMware and ended up discovering that the issue did not have anything to do with VMware but because the servers had the security update 2823324 patch installed just a week ago. More information on this patch can be found here:

MS13-036: Description of the security update for the Windows file system kernel-mode driver (ntfs.sys): April 9, 2013

What we ended up doing was the following:

  • Uninstall the security patch
  • Reboot
  • Let the chkdsk run (the difference we noticed after having the patch uninstalled was that this chkdsk ran much longer as if it did fix something during the process)
  • Reboot again

This resolved the issue.


VMware KB said...

Noticed your blog post here.

We are going to publish a KB article of our own for this, as we noticed there's not much to be found on the topic in Google.



Unknown said...

Great post Terence, it saved me tons of potential headaches.
You've done well young Jedi!

Running Machine said...

Thanks dear for share this, I am looking for this information. you discuss valuable topic.

Rotary12 said...

The VMWare KB is not there and we are still seeing the issue on ESXi 6.x. We are seeing all seeing all windows server guest going into chkdsk at system start up no matter the how we set the autocheck. I have reset autochk, disabled it checked for the dirty bit, (which was never set) and run chkdsk before a reboot and the servers still run chkdsk and get caught in a loop.
I am very frustrated and our customers are not happy either.