Hacker News new | past | comments | ask | show | jobs | submit login

I'd argue that using a terminal multiplexer like tmux or screen on a production server (as this article implied) often means you are doing things wrong.

It is better to set up a service (systemd or other) and configure it to start correctly rather than relying on a tmux session to manage long running processes. By going this route you get logging, auto restarting, starting on boot, monitoring etc for free.




There's difference to daemons and Kong running process. Long process might be tar up all these files, or copy all these. Things that takes 10 s of minutes to hours. Things you don't want to bomb out just cause network died. Things that are one off and variable enough it doesn't make sense to write a server for them.

And of course any you don't want to be interiors part way. Like upgrade or deployment


I just now went through this, which makes this post perfectly apt for my morning. 4-day running process. SSH disconnects on day 3.5.

Reconnected; Started tmux; Started the process; 4 days to go...

I generally know better. Some lessons need to be relearned.


Maybe I'm missing something, but why not nohup?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: