multiprocessing.cpu_count() gives incorrect number of available cores when used within a docker container

[KaneRodriguez]

As discussed here:

If your container has been allocated (say) a single core and 512 MB of RAM on a machine with 64 cores, multiprocessing.cpu_count() thinks it has 64 cores instead of just 1, and you end up spawning 129 workers. This is way more than the container has resources for, and ends up causing out of memory crashes.

I ran into this issue with too many workers being spawned in a docker container on kubernetes, and my application would be in a constant CrashLoopBackoff state.

Debug logs

{"loglevel": "info", "workers": 64, "bind": "0.0.0.0:80", "graceful_timeout": 120, "timeout": 120, "keepalive": 5, "errorlog": "-", "accesslog": "-", "workers_per_core": 1.0, "use_max_workers": null, "host": "0.0.0.0", "port": "80"}

Docker Image

FROM tiangolo/uvicorn-gunicorn-fastapi:python3.7

Relevant Line of Code

uvicorn-gunicorn-docker/docker-images/gunicorn_conf.py

Line 21 in 622470e

cores = multiprocessing.cpu_count()

Solution

I've modified the default gunicorn_config.py to assign cores to an optional AVAILABLE_CORES variable; This is set when I define the number of cpus for the container.

cores = os.getenv("AVAILABLE_CORES", multiprocessing.cpu_count())

[tiangolo]

Great! I think you solved your problem, right?

If that's correct, then you can close this issue ✔️

[github-actions]

Assuming the original issue was solved, it will be automatically closed now. But feel free to add more comments or create new issues.