1

I have a long running tool I’ve written. It, unfortunately, isn’t super reliable and sometimes crashes. I’ll call it myscript.py.

I run it with a launcher script called launch, which sets some env variables for my script.

Is there a way to run ./launch, but where if it crashes, it will automatically restart after two minutes, and if it crashes again, it will wait four minutes, and then if it crashes again, eight minutes, and so on, but it will never wait more than 64 minutes between retries?

Is there a command I can install that can do this restart functionality automatically? Alternatively, I’m open to a custom shell script or possibly doing it with a systemd service if that’s possible.

Raffa
  • 32,237
cocomac
  • 3,394

1 Answers1

0

A Bash shell script:

#!/bin/bash

Set initial wait in miutes


... It will be multiplied by 2 while the result is =< than "$limit"


min="2"


Set maximum wait limit in minutes


limit="64"


Set scrpt filename


name="myscript.py"


while sleep 5
  do
    if ! /bin/pgrep -f "$name" &> /dev/null
      then
        # Script is not running
        sleep "${min}m"
        [[ "$((min * 2))" -le "$limit" ]] && min="$((min * 2))"
        # Add your command to start the script/launcher on the next line
        # ... Send it to the background with "&" so the loop continues
        # example ---> /bin/bash /home/user/dir/launch &
        fi
    done

See more explanation and other ways here and here.

Raffa
  • 32,237