Question
tmux ↔ Docker: Propagating Environment Variables to Existing Panes?
Ira Indigo
0 reputation · 26 Sept 2021, 08:45 UTC
124.4K views
Loading…
0 reputation · 26 Sept 2021, 08:45 UTC
The goal in many containerized workflows is to create a repeatable development environment that is shared across all tmux panes. In a Docker container, each pane typically starts a shell that inherits the container’s environment at launch time. When the container is running, developers often need to adjust environment variables to point to different services or credentials.
tmux exposes the set-environment command, optionally with the -g flag, to define global variables. However, the documentation and observed behavior indicate that these variables are only applied to windows or panes that are created after the command is issued. Existing panes retain the environment state they were launched with and do not see subsequent changes.
Because there is no built‑in mechanism to push an updated environment variable to all running panes, developers resort to re‑sourcing the configuration or manually recreating panes. This limitation becomes a bottleneck in truly repeatable environments, especially when Docker containers expect consistent variables across active sessions.
Is there a tmux command or extension that can retroactively propagate environment variable updates to all existing panes? Can a hook or plugin be employed to automatically reload the shell environment in each pane when a variable changes? What are the implications of such a feature for Docker exec sessions and the isolation of container processes?
31775 reputation · 26 Sept 2021, 09:18 UTC
No, there is no built-in tmux command or extension that can retroactively inject environment variables into the process space of an already running shell. In Linux, a child process (the shell in your tmux pane) inherits a copy of the parent's environment at the moment of creation; it does not maintain a live link to the parent's environment table.
When you use set-environment -g, you are updating the tmux server's internal environment table. While any new panes created after this command will inherit the updated value, existing panes are independent processes. To change a variable in an active pane, you must execute a command inside that specific shell to modify its own memory.
To pull a variable from the tmux server into an active shell, you can use tmux show-environment. Run the following inside the target pane:
docker
tmuxexport MY_VAR=$(tmux show-environment MY_VAR | cut -d= -f2)For Dockerized development environments where variables change frequently, the most reliable pattern is to use a shared environment file (e.g., .env.shared) rather than relying on tmux's internal state:
echo "API_KEY=123" > /tmp/.env.shared.bashrc or .zshrc), add a hook or a manual alias to source this file: alias reload-env='set -a; source /tmp/.env.shared; set +a'synchronize-panes feature:
Ctrl+b then :set-window-option synchronize-panes onreload-envCtrl+b then :set-window-option synchronize-panes offUsing docker exec to enter a container creates a new process. If you set variables via docker exec -e VAR=VAL, those variables only exist for that specific session. If you are running tmux inside a container, the tmux server becomes the parent process. Any attempt to "push" variables from the Docker host into a running tmux pane would require sending a signal or a command string to the shell, as the kernel prevents one process from arbitrarily modifying the environment of another for security and stability reasons.
Are you using a specific shell (e.g., Bash, Zsh, or Fish) across all panes? The syntax for exporting variables from tmux show-environment varies slightly between shell dialects.
Use comments to ask for clarification. Post a solution as an answer.
31,775 reputation · 26 Sept 2021, 13:09 UTC
One practical way to retroactively apply a tmux‑global environment variable to all already‑running panes is to combine a tmux hook with run-shell and send-keys. For example:
tmux set-hook -g focus-gained "run-shell 'for p in $(tmux list-panes -F "#{pane_id}"); do tmux send-keys -t $p "export MY_VAR=\$(tmux show-environment -gv MY_VAR)" C-m; done'"This hook fires whenever the tmux client gains focus (e.g., after you switch clients or press Prefix+l). It iterates over every pane in the current session, sends the keystrokes to re‑export the variable from the tmux server’s environment, and ends with a newline (C-m).
Because send-keys injects keystrokes into the pane’s input buffer, it is safest to trigger the hook when you know each pane is at a shell prompt; otherwise the keystrokes could interrupt a running program. A simple guard is to wrap the command in a shell function that checks $PS1 or uses tmux display-message -p '#{pane_current_command}' to verify the foreground process is a shell before sending the keys.
Note that this approach only updates the environment of the shell processes themselves. Any child processes that were started before the hook ran retain the old environment and must be restarted or re‑sourced to see the new value.