Again MCS Cache issue. My VDI images are bloating ...
# citrix-vad
m
Again MCS Cache issue. My VDI images are bloating and bloating. From one day to the other. Yesterday every VM has about 4-5GB Cache file for the whole day. Today - same image, nothing changed - we are at 30GB 4 hours after turning on the machines.
Copy code
VDI-002 : Size = 23.52 GB | LastWrite = 03/27/2026 10:40:05
VDI-003 : Size = 32.3 GB | LastWrite = 03/27/2026 10:39:40
VDI-004 : Size = 11.9 GB | LastWrite = 03/27/2026 10:34:52
VDI-005 : Size = 33.84 GB | LastWrite = 03/27/2026 10:05:37
VDI-006 : Size = 31.31 GB | LastWrite = 03/27/2026 10:40:05
VDI-007 : SKIP (SMB/445 not reachable)
VDI-008 : Size = 31.93 GB | LastWrite = 03/27/2026 10:40:06
VDI-009 : SKIP (SMB/445 not reachable)
VDI-010 : Size = 31.9 GB | LastWrite = 03/27/2026 10:39:56
VDI-011 : Size = 31.09 GB | LastWrite = 03/27/2026 10:40:11
VDI-012 : Size = 12.59 GB | LastWrite = 03/27/2026 10:40:11
VDI-013 : Size = 30.99 GB | LastWrite = 03/27/2026 10:40:07
VDI-014 : Size = 31.43 GB | LastWrite = 03/27/2026 10:36:21
I got a complete DiskMonitor log from @Chris Twiest @Guy Leech for every day and the entire csv with top 500 shows only about 11GB cumulated for writes to C: I dont get it, anyone got a clue?
j
Have you tried opening up one of those vhdx files to see what all is being written into it?
m
Is it possible to open the file somehow? I dont think this is possible? Copied the file from a powered off machine and cant open it. I think basically its a delta-disk and needs the parent disk
l
I didn't think those were readable. When I had this happen before I had to use procmon or one of those sysinternal tools to actively see what was doing the writing. Pretty much in every situation it's been either windows update or ms store apps
💯 1
r
I haven't worked with MCS much but I have seen similar behavior with old school PVS write-cache files when the vDisk image is fragmented. Defragging the PVS vDisk fixed it. No idea if MCS behaves the same.
o
I would combine @langsbr suggestion with perfmon.msc > adding all the MCSIO counters for a live view, could also be automated via sched task + powershell and dump to .CSV
j
Yep, to add to @langsbr, also the SCCM Cache if you're still using that. I could only catch that with Process Monitor. Disable the services such as SCCM, Windows Update, etc, and Scheduled Tasks that may do updates. Try and stabilise your environment. Then you can re-enable it on 1 machine and just monitor it.
💯 1