There's been a number of times where a misconfigured Docker container has taken over my Unraid server. For example, some containers have an automatic restart on failure. When this occurs, the misconfigured container takes over my server and the server is no longer responding. I can't do anything to stop the container because the server is no longer responsive. I also cannot login directly to the physical machine because the login timeout of 60 seconds. The only solution has been to reset the machine.

Is there a configuration setting to prevent any docker container from doing whatever it is doing that prevents the server from responding?

I usually reduce available available cpu to my dockers so Unraid has dedicated cores, this generally stops the server getting locked out due to compute heavy processes.

With Advanced View toggled (or on cpu pinning page in settings), you can either do this in the CPU pinning section by selecting all cores except the first 2 cpu, or you can do the same by adding something like this to Extra Parameters instead: 

--cpuset-cpus=4-12

Note adjust cpu numbers as per your available cores.

This invokes the same command, but I personally find it more convenient for many dockers on high cpu count machines.

If you are hitting memory limits, there are options for that also, but restricting cpu usually does the job.

Edited June 7, 20251 yr by tjb_altf4

Is there a configuration to make sure the docker service is not a higher priority than critical Unraid/OS processes?

4 hours ago, Jaybau said:

Is there a configuration to make sure the docker service is not a higher priority than critical Unraid/OS processes?

Not that is part of Unraid, no.
I believe you could do something with cgroups to deprioritise processes, but the simpler option is the one I mentioned already.

Looks like one of the containers(s) has a memory leak. So I'm limiting the memory per container.

Would be great for a better/safer way to limit memory/cpu.