Memory leak in openapi.json schema generation.
The issue was reproduced in the Automation Hub project.
openapi.json schema cause uncontrolled growth of memory usage by gunicorn workers and eventually gunicorn worker being killed by OOMKiller.
Steps to reproduce:
Local test environment:
- Minikube instance.
- galaxy-api container configured with memory request\limit of 4 GiB of RAM
- 4 gunicorn workers
After ~1200 requests to
openapi.json gunicorn worker was killed due to OOM and restarted by gunicorn master process.
Please register to edit this issue