Bash: check pid for active process (FreeBSD)

Harpune · 05-16-2006, 01:44 AM

I have a bash script, this one:

Code:

#!/usr/local/bin/bash

pidfile=/var/dchub/pid
hubscript=/var/dchub/DCHub.py
python=/usr/local/bin/python

if [ -e $pidfile ]
then
  #check for running process
  hubpid=`cat $pidfile`
  if [ `kill -CHLD $hubpid >/dev/null 2>&1` ]
  then
    exit 0
  else
    echo "The hub is dead; long live the hub!"
    $python $hubscript
  fi
else
  echo "The hub is dead; long live the hub!"
  $python $hubscript  
fi

Its purpose is to check py-dchub's pid file and if it is non-existant or if their is no process associated with the pid in the file, to restart py-dchub. I have been able to gather that sending the SIGCHLD signal to a process will be ignored by the process, but you can test the exit status of kill to see if the process exists or not, but for some reason, kill is always returning '0'. Whenever I run the script it attempts to start py-dchub, whether or not it is already running. What am I doing wrong?

I am using bash 3.1.10_1 straight from the FreeBSD RELEASE-6.1 packages.

Thanks

ataraxia · 05-16-2006, 01:28 PM

The backquotes around your "kill" make your "if" check the output of the kill, instead of its return code. Remove them.

Harpune · 05-16-2006, 01:33 PM

EDIT: I took out the backticks and it still behaves similarly. I also checked other scripts online that claim to do the samet hing, and they all have backticks around the 'kill' command.

Also, I realized that I think I had my logic backwares in the internal 'if' statement. I think it should read something like this:

Code:

  if [ `kill -CHLD $hubpid >/dev/null 2>&1` ]
  then
    echo "The hub is dead; long live the hub!"
    $python $hubscript
  else
    exit 0
  fi

With the hub attempting to restart if the kill command exits '1' and not the other way around. I have made those changes, though, and my script is still not doing what it should be.

Any more ideas?