How to monitor some UNIX process and send notification in every 10 minutes?

Shell Programming and Scripting
1

Hi Unix Members,

Can anyone guide me to write one shell script to monitor the attach screen processes and when interrupted mail us. ,
like the processes - /bin/ciserver , /bin/clock , /bin/cserver , /bin/main

Please looking forward you guys help.

2

Any ideas/attempts from your side?

3

At a (very) high level, you want:

while true
do
  for processname in $ListOfProcessesToLookFor
  do
    if processnotrunning $processname
    then
      echo "Oh dear!  The process $processname appears to have stopped!  This message was brought to you by the letter C at `date`" | SendMeAnEmail $MyEmailAddress "Warning: Process $processname is not running!"
    fi
  done
  sleep $TimeToWaitBetweenChecks
done
1 Like
4

Hi Unix Gurus,

I want a very simple script. Like

Ps -ef|grep /home/unispool/bin/main

and then check if the process is running or not.

5

If you want a simple script, then it will only be able to achieve simple things.

  • What does your one-line code (now wrapped in CODE tags) give you as output?
  • How would you as a human reader decide if you need to alert or not?
  • How do you plan to alert?
  • Did the suggestion by Smiling Dragon give you a logic to work with?
  • Where are you stuck?
  • What output/errors are you getting?
  • What OS and version are you using? Output from uname -a would be great.

Sorry for all the questions.

Robin

6

Hi Robin,
Thanks for the questions . I am new in the shell scripting.
So the processes I need to monitor. Like when the process get stopped unexpectedly , I need an alert. The Simple script like ,
Will this code work?? like or something more can i do ?

#!/bin/ksh
while true ; do
  isRunning=`ps -ef | grep -v grep | grep /home/unispool/bin/main`
  if [ $isRunning -eq 3 ]
    then
      echo "running"
    else
      echo "not running"
      mailx -s "Main process not running correctly" biswajit.nitd@yahoo.in < /dev/null
  fi

  isRunning1=`ps -ef | grep -v grep | grep /home/unispool/bin/cserver`
  if [ $isRunning1 -eq 4 ]
    then
      echo "running"
    else
      echo "not running"
      mailx -s "Cserver process not running correctly" biswajit.nitd@yahoo.in < /dev/null
  fi

  isRunning2=`ps -ef | grep -v grep | grep /home/unispool/bin/ciserver`
  if [ $isRunning2 -eq 3 ]
    then
      echo "running"
    else
      echo "not running"
      mailx -s "ciserver process not running correctly" biswajit.nitd@yahoo.in < /dev/null
  fi
done
7

I fail to see how:

isRunning=`ps -ef | grep -v grep | grep /home/unispool/bin/main`

(which sets the variable isRunning to the lines reported by ps that contain the string /home/unispool/bin/main (not the number of line that contain that string; the actual lines themselves) would ever be a string that is a numeric string equal in value to the number 3. The same thing can be sad for the other two ps pipelines. So, this script is logically equivalent to:

#!/bin/ksh
while true ; do
      echo "not running"
      mailx -s "Main process not running correctly" biswajit.nitd@yahoo.in < /dev/null
      echo "not running"
      mailx -s "Cserver process not running correctly" biswajit.nitd@yahoo.in < /dev/null
      echo "not running"
      mailx -s "ciserver process not running correctly" biswajit.nitd@yahoo.in < /dev/null
done

which will mostly just fill your mailbox with meaningless messages as fast as your system can run ps , grep twice, and mailx .

If you're trying to count the number of lines produced by ps as filtered by grep in those pipelines, you need to either add a wc -l to the end of the pipeline or add a -c option to the last grep command in the pipeline. (I strongly prefer the latter option since it uses fewer system resources and will run faster.)

But, I fail to see how draining system resources with a continuous loop to slow down your system when it is working well (and to further slow it down and fill up your mailbox rapidly when something isn't running as expected) is a good idea.

Have you considered putting a sleep in your loop (so your system can perform some meaningful work in addition to looking for an imbalance in the three things you seem to think are the only things that matter on your system)?

Have you considered exiting your loop once mail has been sent so you can find one message in your mailbox, fix the problem, and then restart the loop to wait for the next occurrence of a problem?