Server 2025 ReFS Dedup job consumes all memory and hangs server

micce 41 Reputation points
2025-02-20T12:56:33.94+00:00

Hi!

Server 2025 with Hyper-V role and a couple of guests.

Got a 12TB volume with ReFS. 10TB allocated.

ReFS debup was enabled with this:

Enable-ReFSDedup -Volume L: -Type Dedup

But starting a job that do a dedup consumes all memory and ends up hanging the server.

This is after starting the job. And before it hangs the server I stopped it (a little before it exhaust the memory). But if I let it run everything comes to a halt and process start failing because of lack of memory.

Start-ReFSDedupJob -Volume L: -CpuPercentage 30
Stop-ReFSDedupJob -Volume L:

How can I limit the dedup process to work with "what it have" in memory and not taking down the server?

Memory running out with refs dedup job

Get-Volume -DriveLetter L | fl
ObjectId             : {1}\\HV\root/Microsoft/Windows/Storage/Providers_v2\WSP_Volume.ObjectId="{baef148e-8ffc-11eb-a21e-806e6f6e6963}:VO:\\?\Volume{4385d
                       0fb-a232-444c-a2e0-8a7f06507baa}\"
PassThroughClass     :
PassThroughIds       :
PassThroughNamespace :
PassThroughServer    :
UniqueId             : \\?\Volume{4385d0fb-a232-444c-a2e0-8a7f06507baa}\
AllocationUnitSize   : 4096
DedupMode            : Disabled
DriveLetter          : L
DriveType            : Fixed
FileSystem           : ReFS
FileSystemLabel      : p02v03sim-ded
FileSystemType       : ReFS
HealthStatus         : Healthy
OperationalStatus    : OK
Path                 : \\?\Volume{4385d0fb-a232-444c-a2e0-8a7f06507baa}\
ReFSDedupMode        : Dedup
Size                 : 13194072424448
SizeRemaining        : 1460263964672
PSComputerName       :


Get-ReFSDedupStatus L: | fl
Volume           : L:
Enabled          : True
Type             : Dedup
Status           : Dedup Cancelled
Used             : 10,67 TiB
Deduped          : 45,7 GiB
ScannedOnLastRun : 11,07 GiB
DedupedOnLastRun : 953,93 MiB
FullRun          : False
LastRunTime      : 2025-02-20 13:32:22
LastRunDuration  : 01m:09.681s
NextRunTime      : N/A


Windows for business | Windows Server | User experience | Other

2 answers

Sort by: Most helpful
  1. Evgeny Vasilenko 40 Reputation points
    2026-08-03T15:31:43.0666667+00:00

    Microsoft Support acknowledged the issue over a year ago and said a hotfix was expected within a few months, but nothing has changed since then; the service regularly leaking to exceeding 500 GB ram usage on a 10 TB volume size.

    From: Samuel Atanasy (Convergys International Europe) *Redacted*@microsoft.com Sent: Monday, June 23, 2025 16:40
    Subject: Re: [EXTERNAL] Re: Stability and Memory Issues REFS Volume with De... - TrackingID#2506060050000010 Dear Evgeny, Thanks for your confirmation and partnering with me on your case. Since you I’ll be closing your case. Please keep in mind that when your case is closed, we can reopen it . Below, you can see the summary of your case for your records. Confirmed Issue: **Stability and Memory Issues REFS Volume with Deduplication
    Outcome: Issue reported, and we are waiting for a hot fix for the issue next few months.

    User's image

    Is any fix for this bug expected?..

    Was this answer helpful?

    0 comments No comments

  2. Anonymous
    2025-02-24T07:15:41.94+00:00

    Hi micce,

    Thanks for your post. Generally speaking, if you are using deduplication, there is a formula for how much RAM you need to handle it. If this server has been in place for a while, I’m wondering if you hit a limit of data to RAM ratio that it can no longer handle. You need 1GB of RAM for each 1TB of data for deduplication to run smoothly.

    Reference: Optimize storage with ReFS deduplication and compression in Azure Local - Azure Local | Microsoft Learn

    Best Regards,

    Ian Xue


    If the Answer is helpful, please click "Accept Answer" and upvote it.

    Was this answer helpful?


Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.