Windows 11 + REFS + Defrag = CPU 100% (ConnectWise Automate)

[sykerzner]

Hi
We have a bunch of small desktops in the field running VBR for our clients. Each has 2 physical disks, for OS, and for storage. The storage disks hold a single VHDX, which is mounted on startup. It contains a volume formatted with REFS. Let's call them C (OS), E (Physical) and F (VHDX with REFS)

Most of our machines are Win10 or server 16/19 which work perfectly.

However, on Win11 machines (22h2 and 23H3), are having the following issues: Every so often the machine freezes up with 100% cpu usage. We were able to track this back to our RMM which runs a defrag when doing a disk inventory. Volumes C and E complete within seconds (in event viewer I can see "The storage optimizer successfully completed analysis on New Volume (E:)" ). SOMETIMES the machine then freezes. Sometimes it eventually completes on it's own with a message "The storage optimizer successfully completed analysis on (F:)", but sometime we have to down it after waiting hours or days.

We get the same result when manually running defrag, SOMETIMES.

I know this is a Windows problem, not a Veeam problem, but I am curious if anyone else encountered this.

[Gostev]

Hi, do you want to include the RMM name as well? I can update the topic name with it.

[sykerzner]

Sure, ConnectWise Automate (Labtech)

[sykerzner]

Here is an update.

We got back a test machine which had been freezing. If no backups where setup, then defrag would not freeze the machine.
Once we added backups, and ran them, the machine would freeze at the next defrag - Consistently. Even after the backups finished.
It seems that rebooting before the defrag prevented the freeze. Stopping Veeam did not prevent the freeze.

I have a ticket with Veeam 07264297

Experimenting more, it seems like this does not happen if the REFS is version 3.10. I am not sure about 3.9, but all the machines I am having trouble with have REFS 3.7. Could be an REFS bug with Veeam backups tripped on and setting up defrag to trigger the freeze? and maybe the bug was fixed in 3.10? I am running some more tests to see.

[ReKe]

We also see Machines that freeze with W11 22H2 and 23H2 in combination with ReFS. They will be absolutely unresponsive.

[ReKe]

@sykerzner Do you have any new findings on this one?

[M-M-C]

Now I got the same issue with Win11 24h2 and Server 2025 (same version for both, build 26100). Win11 22h2 and 23h2 are 22621/22631, so everything newer than Server 2022 (build 20348) seems to be affected. Server 2025 with ReFS (v3.12) is completely unusable right now. I know those Windows versions are not supported yet, I'm just sharing my experiences from my home lab.

[sykerzner]

We had not had this issue on any machine for a while, so it seemed 3.10 solved it. Some of our machines had only 8 GB of RAM, so going to 16 may have helped.
However, it started happening again on one machine. I think it was right after we added some large busy backups...

One workaround: if you reboot before running a defrag, that doesn't happen.

[karsten123]

with defrag you mean Windows Disk Defragmenter?

[ReKe]

How do you upgraded REFS to 3.10?

[Daun2shay]

I have also experienced this we now just use NTFS in our org

[tyler.jurgens]

How do you upgraded REFS to 3.10?

Here is the ReFS version table (according to wikipedia):
https://en.wikipedia.org/wiki/ReFS#Versions

You won't get ReFS above 3.7 on anything other than Windows 11, and looks to only be available on insider preview builds.

[sykerzner]

How do you upgraded REFS to 3.10?

So main workstation is WIn11 enterprise, When I format a volumes with REFS, I get 3.10. To test on machines, I'd frequently create a VHDX and format it with REFS and move it over.

with defrag you mean Windows Disk Defragmenter?

Yes, or our Labtech's build in Disk Intentory also runs a defrag of sorts, and also MS has a weekly scheduled task. Any of those will trigger the issue
(I don't understand fully the different things MS does when it 'defrags' vhdx's, I know it's not a classic defrag. I am being unspecific)

[helpit]

Anything new on this topic?
I have the same issue on 2 virtual repositorys running server 2022.

regards,
Thomas

[sykerzner]

Nothing I tried conclusively fixed the issue
It looks like this post may have pick up the issue:


veeam-backup-replication-f2/b-r-on-wind ... 95062.html