Docker - Detached and Interactive?
I think that in general starting up a detached interactive container like this is an unusual use case. It's certainly nothing I ever end up using in my daily use of docker.
Some applications may behave differently when their stdout is associated with a tty vs when it's not (e.g., they may only log to stdout by default when their output is a terminal, or if you're running a shell it may simply exit when not associated with a terminal).
You may want to docker attach
to the running container. This is especially important if you've started some sort of detached shell.
When you run an image with only -d option, the container will exit immediately after the command executed. If you run with -itd option, the container will be detached but running in background, and you can attach back when you need. See the screenshot attached for more clarity.
-i
(interactive) is about whether to keep stdin open (some programs, like bash, use stdin and other programs don't). -d
(detached) is about whether the docker run
command waits for the process being run to exit. Thus, they are orthogonal and not inherently contradictory. A program like bash exits when stdin is closed, so without -i
, it exits immediately.
-t
(tty) allocates a pseudo-tty. You can see the difference from running bash with -it
vs with just -i
. For example, without -t, you don't get any prompt and ls
shows results in one column. This difference is like the difference between running ls
from a normal bash session and running ls | cat
from a normal bash session, where cat
does not have a pseudo-tty.
When you docker run
bash in a container, -it
and -itd
behave differently as follows:
- With
-it
,docker run
gives you the bash prompt immediately. - With
-itd
,docker run
exits immediately, but you candocker attach
after that and get the bash prompt just as if you had just donedocker run -it
.