A container is an ecosystem for your application and exists as long as your application is running. If you application crashes or never gets off the ground, the Docker container will terminate. It would be helpful to monitor your app and do some scratching around while/after you application crashes in order to track down the cause of the failure. However, because a Docker container is kind of like a little server/machine, if it is off/terminated by the time you want to investigate and won’t be able to log into it.
We need a way to keep the container running.
Method 1 - using docker
You can use Docker’s -t or --tty flag to keep the container running even if your application crashes. This flag allows the container to emulate a terminal session - like the one you’re running this command from. Like so:
docker run -d -t alpine
Method 2 - tail in the Dockerfile
The most common method however, is to use the Dockerfile’s ENTRYPOINT to execute some long-running command like tailing a file with the “follow” option:
# Dockerfile
FROM alpine:latest
ENTRYPOINT ["tail", "-f", "/dev/null"]
The tail command prints out the last few lines of a file. By default it prints of 10. But when using the -f flag, it will watch a file and print out any new data added to the end of it until it is manually terminated with a SIGKILL which can be done with a CTRL-C. In this case tail is watching the null device which won’t ever print anything but won’t ever close/end either allowing tail to keep running.
Method 2 - inlined tail
You could inline the above tail command in the Docker run command like so:
docker run -d alpine tail -f /dev/null
Method 4 - using sleep
In a similar vein as method 2 and 3 you could use the sleep command. Give it a long duration and voila:
In a Dockerfile
# Dockerfile
FROM alpine:latest
ENTRYPOINT ["sleep", "infinity"]
Inlined in docker run
docker run -d alpine sleep inifinty
And there you go :)