Hello everyone.
I recently ran into some issues with my Server 2012 R2 Hyper-V cluster where CSV's were running low on space causing the Cluster service to crash on a host. I cleaned up space then decided to start monitoring the volumes more closely to see what changes were happening from day to day. I noticed that over the last 5 days, one particular volume (the worst offender) had grown over 300 GB's in size over 4 days and was back under 10% free space again. I started running Powershell scripts to report on VHD/vhdx sizes so I could see if there was any change as maybe one of our servers were expanding VHD size. (I do know dynamic disks aren't recommended but I obtained this environment well after it was implemented so converting those to fixed is a problem for another day.)
I found that the disk changes were very minimal over the course of that time and did not jive with the decline I was seeing in size. After checking VSS shadows (none) and searching the web, I ran across this post from a few years back https://social.technet.microsoft.com/forums/windowsserver/en-US/7afd6a36-8b10-43c2-b99b-d7a4b695b747/disk-space-on-csvs-disappearing
This is the exact issue I am seeing. I tried changing the owner of the offending CSV to another node and low and behold I went from 340 GB free to over 3.6 TB!. My concern is that there is a cache that is not cleaning itself up properly and just continues to grow. Im not sure what to do to even check this out (commands or scripts etc) and checking disk space each and every day and manually moving the CSV seems counter productive to the idea of the cluster itself.
Has anyone else experienced something similar and if so, what did you do to check, monitor and even fix it?
Thanks!