Skip to content
Blog

Working with Stdin, TTY, and Interactive Shells in Docker

Master interactive Docker containers using -i (stdin) and -t (TTY) flags, pseudo-terminal allocation mechanics, entering running containers, and piping streams into docker run/exec.

Published on • November 17, 2026

AI Assistant

Running interactive terminal shells or streaming input into Docker containers is a fundamental daily task for developers and sysadmins. When running commands like docker run -it alpine sh or docker exec -it web bash, flags like -i and -t play distinct roles.

Understanding standard input (stdin), pseudo-terminal (TTY) allocation, and stream piping mechanics allows you to troubleshoot issues, interact with containerized applications, and automate shell pipelines.


1. Deconstructing -i (Interactive / Stdin) and -t (TTY)

While -i and -t are frequently combined as -it, they control two entirely separate operating system mechanics:

FlagNameFunctionality
-i--interactiveKeeps standard input (stdin) open even if not attached, enabling data piping into the container.
-t--ttyAllocates a pseudo-terminal (pty) inside the container, providing terminal formatting, colored text, and prompt handling.

2. When to Use -i, -t, or -it

Combination A: Interactive Terminal (-it)

When you want an interactive REPL shell (e.g., bash, sh, python, node) where you type commands and expect formatted prompts:

docker run --rm -it ubuntu:latest bash

Here:

  • -i connects your keyboard input to the shell’s stdin.
  • -t provides terminal capabilities (cursor movement, color ANSI codes, Ctrl+C interrupt handling, line editing).

Combination B: Piping Data into Containers (-i without -t)

When piping file streams or script contents into a container process via stdin, do not pass -t:

# Pipe host SQL script directly into a containerized PostgreSQL database
cat schema.sql | docker run -i --rm postgres:alpine psql -U postgres -d mydb

Passing -t during piped operations forces pseudo-terminal allocation, causing line-ending translation errors (CRLF vs LF) or binary stream corruption.

Combination C: Headless Script Execution (Neither -i nor -t)

For background automated scripts where output is captured without user interaction:

docker run --rm alpine echo "Build completed successfully"

3. Entering Running Containers with docker exec -it

To open an interactive diagnostic shell inside an already running container instance, use docker exec -it:

# Execute bash inside a running NGINX container
docker exec -it web-server bash

If the target container does not include bash (e.g., minimal Alpine Linux images), fall back to POSIX sh:

docker exec -it web-server sh

4. Detaching from Interactive Sessions Without Stopping

When running an interactive container session (docker run -it ...), pressing Ctrl+C or typing exit sends a signal to PID 1, which stops the container.

To detach from a running interactive session without stopping the container, use the default escape key sequence:

Ctrl + P, followed by Ctrl + Q

The container detaches and continues running in the background. You can re-attach to it later using docker attach:

docker attach web-server

5. Summary and Best Practices

Takeaway: Use -it together when opening interactive terminal shells like bash or sh. Use -i alone when piping raw text or binary data streams into containerized commands, omit both for automated non-interactive tasks, and use Ctrl+P + Ctrl+Q to detach from interactive sessions cleanly without terminating the container.


Sources