Which BOINC version you run (See Help->About BOINC in BOINC Manager). We'd just like the version number, please: for instance v7.8.2
8.2.6
What your problem is exactly, write it out in as much detail as you can.
When running tasks, it uses all available memory on my Unraid server. I have to physically go to the server to stop it. I have web prefs set to 50% memory but it still uses more than this.
I can easily reproduce the error by simply turning the Docker container back on.
What you have done so far to try to remedy the problem.
Updated the container - no go
If applicable, the project(s) that you have the problems with and if they have multiple applications, which ones give you problems.
Einstein at Home
Post the first 40 - 100 lines from your BOINC Event Log (CTRL+SHIFT+E), including it showing what your BOINC is doing, which project tasks it is running.
In case the log doesn't show this, please give us:
• the CPU brand and model
12th Gen Intel® Core™ i5-12600K @ 3663 MHz
• the GPU brand and model
iGPU, Intel i915 Alderlake-s GT1
• the driver version(s) you have installed
I think it's in the kernel 6.12.54-Unraid
• the operating system that you use
Unraid /Slackware

You don't tell us how much RAM (memory) your computer has and how many tasks you are trying to run at the same time, nor are you showing what specific type of tasks at Einstein you are trying to run.

Einstein O4 tasks can sometimes require up to 5Gb (GIGA BYTE) RAN to run PER TASKS.

I'm going to reference you to the Einstein forums so you can better understand what probably is going on and you should read the whole topic, not just the post I referenced below and ask any further questions there because it's RELATED TO THE EINSTEIN PROJECT.

https://einsteinathome.org/de/content/multi-directional-gravitational-wave-search-o4-data-o4md1?page=2#comment-240655
Bernd Machenschalk wrote:
The main reason why we run these WUs on CPUs is that these require more memory than most GPUs have, up to <5GB in higher frequencies. However, here the correct value should be written into the WU limitations and the client should be aware of this and schedule these accordingly. Then again, nothing is perfect ...

I have 32GB of RAM.
Note that I've not had this problem with Einstein for years. It just started happening on my server in the last week or two.

Looking in the Einstein forum I see that another user is having the same problem. Looks like since 11/26.2025 there was a bug that the project admin is aware of and continued anomalies.

But I'll go to that forum going forward since this appears to be with that particular project and not the BOINC client itself.