docker run -i -t

docker run -i -t --name nodejs ubuntu:latest /bin/bash

So here, -i stands for interactive mode and -t will allocate a pseudo terminal for us.

Some more trivia about these flags.

-i keeps STDIN open even if not attached and it allows piping (or “composition”). Since -i keeps STDIN open even if not attached, it allows for composition (piping).

So for example

docker run ubuntu printf "line1\nline2" \
    | docker run -i ubuntu grep line2 \
    | docker run -i ubuntu sed 's/line2/line3/g'

Or, for a simpler example,

echo hello | docker run -i busybox cat
  hello

The -t option goes back to how Unix/Linux handles terminal access:

In the past, a terminal was a hardline connection. Terminals had physical device drivers which were real pieces of equipment.

Once generalized networks came into use, a pseudo-terminal driver was developed.

The pseudo-terminal driver is a layer of abstraction that helps the end user interact with the terminal.

So, with that background, when you do a docker run container (without any -i or -t flag), you basically get a STDOUT stream. So this works:

docker run ubuntu echo "hello" | cat

With -i you add a STDIN to docker too, so…

echo "hello" | docker run -i busybox cat

…works as well.

And with -t you add the pseudo-tty driver to the container, so you can type stuff from your keyboard into the container process; this is what you want (*typically combined with -i as in -it**) when you want to interact with the container.

The -i -t makes the container start to look like a terminal connection session.

If you are interested in how Docker actually does what it does, there is no better way than reading the source to learn more about it.

nhannguyen95/docker-i-t-trivia.md