Jump to content
Welcome to our new Citrix community!
  • 0

CITRIX, (Hosted Desktops-Server 2016)-Memory gets filled up.


Darman

Question

Hi Everyone,

 

We have Shared Desktops; Windows sever 2016 -non-Persistent, Provisioned in MCS-7.15LTSR.

Every week , we have to reboot the servers or else they became very slow/unresponsive, and unregistered.

We suspect the Memory gets filled up, each Machine has  allocated the 32GB per server.

Any one know how to tackle this issue, other than restarting servers.

 

Thanks,

Akiyo.

Link to comment

10 answers to this question

Recommended Posts

  • 0

Greetings:

 

I've spent some time tracking down a similar problem, might be the same -- Windows Server 2016, phsysial -- 192GB of RAM.  What I found was machines started to under-perform, then slowly got worse and worse until the servers would lock up and become unresponsive.

 

I was able to observe that the machines were using a rather large amount of non-paged pool memory.  As the Non-paged pool memory grew, the available memory shrank causing the machine to page to disk.  Eventually "something" would consume 100% of the RAM and -- lockup.

 

Using Poolmon.exe, I was able to identify the tag as PsIn.  I've sent memory dumps and xperf logs off to Microsoft for Analysis.  At this point in time, they believe the probelm is being caused by UPMjit.sys.  This is a little bit of unfamiliar territory for me, but Microsoft mentioned that PsIn was a type of buffer where executable names were stored.   Process are supposed to put data into this "pool" and then free it -- it's not supposed to stay herre very long.  What they see is that the data is stored but never freed, causing the machine to leak memory and drive the machine out of memory.

 

If you're curious if you're seeing the same, you could try either poolmon from the Windows Driver Kit or the open source version, PoolMonx

 

With PoolMonX, you can just sort by bytes and then watch it keep going up and up.    When data is green, it means it went up from the last time it was polled, and red means it went down.  You SHOULD see PsIn going up and down, but in my case it only ever went up.

 

I'm curious if this is similar behavior to what you're seeing, but its been a while so it might be too late!

Link to comment
  • 0
On 6/4/2022 at 1:25 AM, Joe Robinson said:

Greetings:

 

I've spent some time tracking down a similar problem, might be the same -- Windows Server 2016, phsysial -- 192GB of RAM.  What I found was machines started to under-perform, then slowly got worse and worse until the servers would lock up and become unresponsive.

 

I was able to observe that the machines were using a rather large amount of non-paged pool memory.  As the Non-paged pool memory grew, the available memory shrank causing the machine to page to disk.  Eventually "something" would consume 100% of the RAM and -- lockup.

 

Using Poolmon.exe, I was able to identify the tag as PsIn.  I've sent memory dumps and xperf logs off to Microsoft for Analysis.  At this point in time, they believe the probelm is being caused by UPMjit.sys.  This is a little bit of unfamiliar territory for me, but Microsoft mentioned that PsIn was a type of buffer where executable names were stored.   Process are supposed to put data into this "pool" and then free it -- it's not supposed to stay herre very long.  What they see is that the data is stored but never freed, causing the machine to leak memory and drive the machine out of memory.

 

If you're curious if you're seeing the same, you could try either poolmon from the Windows Driver Kit or the open source version, PoolMonx

 

With PoolMonX, you can just sort by bytes and then watch it keep going up and up.    When data is green, it means it went up from the last time it was polled, and red means it went down.  You SHOULD see PsIn going up and down, but in my case it only ever went up.

 

I'm curious if this is similar behavior to what you're seeing, but its been a while so it might be too late!

 

Hi Joe,

 

Did you get anywhere with the PsIn non-paged pool memory leak? I'm getting the same symptoms on a Windows 10 VDI image and any clues would be much appreciated.

 

Thanks,

 

Steve.

 

Link to comment
  • 0
On 2/28/2023 at 7:03 AM, Stephen Jeffrey1709163295 said:

 

Hi Joe,

 

Did you get anywhere with the PsIn non-paged pool memory leak? I'm getting the same symptoms on a Windows 10 VDI image and any clues would be much appreciated.

 

Thanks,

 

Steve.

 

 

I did; it was Session Recording.  There was a bug that took forever to get fixed, but session recording was causing the heavy memory usage which would eventually drive the machine out of memory and crash.

 

The latest version is supposed to have the fix, but I completely removed it from my environment so I can't be 100% certain.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...