Tmux kill server does not kill everything

[poetaman]

>> lsof -U | grep '^tmux'
tmux      99036 reportaman    9u  unix 0x570ea09cc94aadd1      0t0      /private/tmp/tmux-501/default
tmux      99036 reportaman   10u  unix 0x570ea09cc94ae611      0t0      /private/tmp/tmux-501/default
tmux      99036 reportaman   14u  unix 0x570ea09cc94acf69      0t0      ->0x570ea09cc94ad8c9
tmux      99851 reportaman    9u  unix 0x570ea09cc94aad09      0t0      ->0x570ea09cc94aadd1

I recently added this to my alacritty.yml config file.

shell:
  program: /opt/homebrew/bin/zsh
  args:
    - -l
    - -c
    - "tmux attach || tmux new-session -s alacritty"

Then when I open alacritty & close it right away, and repeat the process 2-4 times, I see increase in number of entries returned by lsof -U than I had seen the first time I opened alacritty.

The behavior of >>tmux kill-server also changes. If I had opened alacritty only once, running >>tmux kill-server kills server and closes the window. But after two /private/tmp/tmux-501/default entries appear, running >>tmux kill-server seems to kill one of the /private/tmp/tmux-501/default, and tmux session (with same name) continues to exist. All the while, there seems to be only one session visible in session list loaded by pressing <tmux_prefix>s.

I got this alacritty-tmux recipe from alacritty/alacritty#2956.

[nicm]

Am I meant to guess your tmux version? What does pgrep -lf tmux show after kill-server? Does this still happen if you try without a .tmux.conf?

[poetaman]

@nicm

>> tmux -V
tmux 3.2

Here's what pgrep -lf tmux shows at various points of time:

1. After opening Alacritty first time:

❯ pgrep -lf tmux
4881 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
5674 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
6117 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
6703 tmux new-session -s alacritty
7099 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
7148 tmux new-session -s alacritty
7540 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done

2. After closing and reopening Alacritty 2-3 more times:

❯ pgrep -lf tmux
5674 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
6117 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
7099 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
7148 tmux new-session -s alacritty
7540 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done

3. After performing tmux-killserver from within Alacritty's running tmux for first time (window doesn't close, am returned to another session of same name apparently?)

❯ pgrep -lf tmux
7099 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
7540 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
8066 tmux new-session -s alacritty
8119 tmux new-session -s alacritty
8541 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done

4. After performing tmux-killserver second time (this time window quits):

❯ pgrep -lf tmux
7099 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
7540 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done
8541 sh -c trap 'exit 0' TERM; while :; do nice cut -c3- ~/.tmux.conf | sh -s _uptime; sleep 60; done

Haven't tried without .tmux.conf, can try that later...

[nicm]

If you wait 60 seconds for all the sleeps in your scripts to return, do these processes disappear?

[poetaman]

@nicm Not sure what that means, I am new to shell scripting. I don't know where this: sh -s _uptime; sleep 60; is coming from, and what you want me to try.

I am using https://github.com/gpakosz/.tmux

[nicm]

They are something @gpakosz's script is running. I mean after you do kill-server, wait for 60 seconds and see if the scripts exit after that.

[gpakosz]

They will exit

[gpakosz]

Maybe I should trap SIGHUP as well

[poetaman]

@gpakosz: No, I waited for a 2-3 minutes, one of them lasted till I did tmux kill-server. + @nicm

[gpakosz]

They do exit.

That being said, I think I should trap EXIT and not TERM, this way the background jobs will exit immediately upon kill-server/SIGTERM.

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.