Normally if we need to refresh a config in a Container we will spawn a new one, or we will access with sudo docker exec -it /bin/sh mycontainer for instance and force a reload, or we will have to restart the Container.

What if we want to be able to reload the config at any moment without restarting the process, or to trigger a process in our Container (like a dump or a flush) in another way than implementing an API?.

An unexplored way, for many, to communicate with your Container’s main process is to send Signals.

So basically I will show you how you can trap Signals within a Python process which is the main process for your Docker Container, and send them from your Hypervisor with the command:

sudo docker kill --signal=SIGUSR1

I choose to use SIGUSR1 as it is reserved for user defined Signals.

You can clone the project or get the source code from:

https://gitlab.com/carles.mateo/docker-signal

The Dockerfile

FROM ubuntu:20.04

MAINTAINER Carles Mateo

RUN apt update && apt install -y python3 python3-pip vim less && apt-get clean

# This will make sure printing in the Screen when running in dettached mode
ENV PYTHONUNBUFFERED=1

ENV DOCKERSIGNAL /var/dockersignal

RUN mkdir -p $DOCKERSIGNAL

COPY *.py $DOCKERSIGNAL

WORKDIR $DOCKERSIGNAL

# Again to enforce printing to the Screen when running dettached
CMD ["python3", "-u", "/var/dockersignal/dockersignal.py"]

The dockersignal.py file

# By Carles Mateo https://blog.carlesmateo.com

import signal
import time


def handler(signum, frame):
    print('Signal handler called with signal', signum)

    if signum == 10:
        # 10 is the equivalent to SIGUSR1 for most x86/ARM (not for Alpha/Sparc, MIPS, PARISC)
        print("Simulated action: Reload config")


if __name__ == "__main__":

    print("Waiting for a Signal")
    # Listed for this signal, so can listen for more
    signal.signal(signal.SIGUSR1, handler)

    while True:
        # Do Whatever
        time.sleep(1)

A shell file to build and run the Container like a pro

#!/bin/bash

DOCKER_CONTAINER_NAME="docker-signal"
DOCKER_IMAGE_NAME="docker-signal"

printf "Removing old Container %s\n" "${DOCKER_CONTAINER_NAME}"
sudo docker rm "${DOCKER_IMAGE_NAME}"

printf "Removing old Image %s\n" "${DOCKER_IMAGE_NAME}"
sudo docker image rm "${DOCKER_IMAGE_NAME}"

echo "Creating Docker Image"
sudo docker build -t ${DOCKER_IMAGE_NAME} . --no-cache

retVal=$?
if [ $retVal -ne 0 ]; then
    printf "Error. Exit code %s\n" ${retVal}
    exit
fi

echo "Running Docker Container ${DOCKER_CONTAINER_NAME} based in image ${DOCKER_IMAGE_NAME}"
sudo docker run --cpus="1.0" --name ${DOCKER_CONTAINER_NAME} ${DOCKER_IMAGE_NAME}

See it in action

References

https://docs.docker.com/engine/reference/commandline/kill/

https://man7.org/linux/man-pages/man7/signal.7.html

https://docs.python.org/3/library/signal.html

If you want you can buy my Docker Combat Guide book.