Aug 6, 2024
3
0
10
Hello, I'm starting to be a system admin for a beefed-up windows server and lately have noticed a large amount of memory usage.
Background info-

OS: Microsoft Windows Server 2022 Standard v. 10.0.20348 Buitld 20348
RAM: 144 GB
CPU: 36 total cores spread apart 36 single-core Intel Xeon Platinum 8268 CPUs
(I do not get to change or manage the hardware)
This server is running a web application, several standalone scripts, and a PostgreSQL 16 database.

I've noticed the memory usage continues to go up throughout the week and can only be fixed from getting to high by rebooting the machine which brings the usage back to normal levels.

I've become slightly more familiar with VMMAP, RMMAP, and Process Explorer (screenshots below). My major concern and where I feel my memory is being soaked up is in the screenshot of RMMAP which shows 65,013,548 KB as "Active" within the "Unused" usage.

Any assistance finding where this memory is going would be great.

Task Manager

Resource Monitor

RMMAP

Process Explorer
 

USAFRet

Titan
Moderator
You need to talk to your developers and see what their code is doing, and how it can maybe be optimized.
Database admin, front end, etc, etc.

This is not something we out here can diagnose.
 
Aug 6, 2024
3
0
10
Hello @kanewolf and @USAFRet I'm not particularly worried about the memory shown in the screenshots regarding Postgres and the web application, from my understanding after speaking to developers, the numbers being monitored are expected. But its a 144 GB RAM machine so I don't understand how a little below 10 GB between postgres and IIS could be using 80% of our RAM. As I mentioned my largest concern is the large amount within RAMMAP for "Unused" "Active". Any advice on how to pinpoint where that RAM is remaining unused but active would be great so I can bring back to development team.
 

kanewolf

Titan
Moderator
Hello @kanewolf and @USAFRet I'm not particularly worried about the memory shown in the screenshots regarding Postgres and the web application, from my understanding after speaking to developers, the numbers being monitored are expected. But its a 144 GB RAM machine so I don't understand how a little below 10 GB between postgres and IIS could be using 80% of our RAM. As I mentioned my largest concern is the large amount within RAMMAP for "Unused" "Active". Any advice on how to pinpoint where that RAM is remaining unused but active would be great so I can bring back to development team.
I have seen some references to VMs which were stopped but not shutdown properly contributing to this problem. Could you have something like that happening? Do you have an orchestration suite that is starting and not properly exiting VMs ?
 
You might want to look at SQL as its a memory hog. We run our SQL data base in its own 32GB VM and it will take all of it. I notice in your RAMMAP list 6 of the top 10 programs are SQL related.

To be truthful we run every piece of software at work in its own VM. Say there was and issue with one software and needs to be rebooted. Rebooting then only takes down one VM instead of rebooting the entire server and taking everything down.

I also have task scheduler setup to reboot every VM on Sunday at 30min intervals and after the last one reboots to wait an hour then reboot the physical server. No one works Sunday's so it takes all day which is fine by me.
 
Aug 6, 2024
3
0
10
Thank you for all your feedback. After testing each project running on the server, I’ve come to the conclusion that it is a C# Selenium chrome bot.

Further analysis with my developer has determined that when exiting the program, resources for the chrome are not being correctly released.