Podman escalates to SIGKILL because containerized process does not exit on SIGTERM

[x80486]

When stopping a docker.io/datawookie/medusa-proxy:0.2.2 container, Podman logs that SIGTERM failed and resorts to SIGKILL.

$ podman compose down --volumes 
WARN[0015] StopSignal SIGTERM failed to stop container tor-proxy in 15 seconds, resorting to SIGKILL 
tor-proxy
tor-proxy
d9d7e172468d21c2279178694a9734d4d1912d3dea054922ac033931c6acfb67
tor-network

Increasing stop_grace_period (timeout) does not help.

I can only guess that somehow, the the process inside the container is not quite sending the correct signal back.

[x80486]

If you are fine with this, the suggested fix could be something around these lines:

• Include tini as PID 1 in the image
• Set it as the ENTRYPOINT so it becomes PID 1 and forwards signals correctly

Another approach could be to tell the Python script to handle SIGTERM explicitly — but I'm not entirely sure about this one.