11

I'm looking for a way to monitor a process, and relaunch the executable if the process dies for whatever reason. Does Linux have a built-in tool that can do this?

Dennis Williamson
  • 60,515
  • 14
  • 113
  • 148
pufferfish
  • 2,660
  • 9
  • 37
  • 40
  • there's some more answers and pointers in this question http://serverfault.com/questions/56526/whats-the-best-way-of-setting-up-a-cron-job-to-check-that-a-long-running-process/ – jrg Aug 27 '09 at 10:53

6 Answers6

11

I have answered a similar question before.

In your case:

#!/bin/bash
while ! <<command>>
do
  sleep 1
  echo "Restarting program..."
done

Replace <<command>> with the command you want to execute. The process has to finish with exit code zero in order to break the loop. Otherwise, it is restarted by the script.

Juliano
  • 5,402
  • 27
  • 28
7

If it's launched from init, you can have it respawned by init. Set the action for the process to 'respawn' for the run levels you want the process at.

Cian
  • 5,777
  • 1
  • 27
  • 40
4

Do you only want to restart it if it dies? The problem with this is that it doesn't handle situations where it freezes. So to just check for the process won't always help. So if this is something like a web server, you want to have a script checking it from the user's perspective.

If you set up Nagios Monitoring, you can then use event handlers. These are just scripts that you write that will run when the service down, so you could have one that restarts something like Apache if the web site is down.

Kyle Brandt
  • 82,107
  • 71
  • 302
  • 444
3

Why not a simple bash script?

#!/usr/bin/bash
while `true`
do
  xeyes
done

Replace xeyes with program of your choice.

djhowell
  • 1,162
  • 7
  • 9
3

Depending on the distro, but we've successfully used "monit" for this task. It's pretty easy. You can write your own checks, monitor the proces PID, etc.

Example monitrc file:

check process sshd with pidfile /var/run/sshd.pid
   start program  "/etc/init.d/ssh start"
   stop program  "/etc/init.d/ssh stop"
   if failed port 22 protocol ssh then restart
   if 5 restarts within 5 cycles then timeout
Ramon Poca
  • 181
  • 3
0

Short Answer: No you have to do it yourself.

Long Answer: You'll need a parent process to start it, and then keep checking that it's alive and restart it if needed.

You could use something like daemontools to manage it.

Amandasaurus
  • 30,211
  • 62
  • 184
  • 246
  • Short answer: you're dead wrong. Long answer: if a Linux distro doesn't have inittab, it will have /etc/event.d/ - either mechanism will restart a process when it dies. – Paul Tomblin Aug 27 '09 at 12:21