check me for a process and restarting if the process is not found

[Deoncia Grayson_1]

hello everyone

I relatively new to the scripting world and would like a script that would go out and check a process and if it doesn't find it, the restart it. The process is called rclproc, any help would be appreciated.

[Scott Palmer_1]

Deoncia,

there are many ways to do this

this example is one that will start a program that will run for ever and check ever 5 minutes to see if the program is running. If not it restarts the program.

#! /bin/sh

while true; do # infinite loop
var=$(ps -ef | grep -v grep | grep rclproc) # check to see if program is running

if [[ -z "$var" ]]; then
/usr/local/bin/rclproc > /dev/null 2>&1 & # run program if nothing is found inthe ps
else
var="" # reset the variable to null
fi
sleep 300 # sleep 5 minutes
done


you could always eliminate the infinate loop, and run it from cron ever 5 mintues or so

Hope this is helpfull
Sincerely

--Scott Palmer

[Scott Palmer_1]

to start the program

create a script foo2 with the code snippet in it.

chmod 750 foo2
./foo2 > /dev/null 2>&1 & # run pipe output to /dev/null and background process.

or you could do
echo "./foo2" | at now

etc.

Sincerely

--Scott Palmer

[Sundar_7]

Deoncia,

You can try something like this

#!/usr/bin/sh
PROCNAME="rclproc"
while true
do
ps -ef | grep -q "$PROCNAME" >/dev/null 2>&1
if [[ $? -ne 0 ]]
then
# Process is not there
/path/to/the/procname
[ $? -eq 0 ] && echo "$PROCNAME successfully started"
fi
sleep 300
done

- Sundar

[Sridhar Bhaskarla · Accepted solution]

Hi,

This script does little bit of checking. Modify the lines with appropriate details whereever I indicated them with # mark before running. It basically does the following

1. check if the process is running. If so, do nothing.
2. If the process is not running, then log a message in /tmp/rclproc.log file and start it. If it starts, then log a successful message into the log file.
3. If the startup was not successful, then send you a mail and creates file /tmp/rclproc.nocheck and will stop monitoring the process. Fix the errors, remove the stamp file and it will start monitoring it again.

Put it in 'cron' to run every 5 mins. Test it prior to implementing it.

#!/usr/bin/ksh

PROC="rclproc"
STARTCMD="your_command_to_start_rclproc" #<==modify
LOGFILE=/tmp/rclproc.log
EMAIL=your_mailid@yourdomain.com #<==Modify
STAMP=/tmp/rclproc.nocheck
touch $LOGFILE

check_pid()
{
MESG=$1
STATUS=$2

UNIX95= ps -C "$PROC" > /dev/null 2>&1
if [ $? != 0 ]
then
echo "$(date): $PROC $MESG" >> $LOGFILE
$STARTCMD
else
echo "$(date): $PROC has been successfully started" >> $LOGFILE
STATUS=0
fi

}

if [ ! -f $STAMP ]
then
check_pid "not running" 0
check_pid "is still not running" 1

if [ $STATUS = 1 ]
then
echo "Unable to start $PROC. Fix the errors
and remove $STAMP file to resume monitoring"| mailx -s "$PROC failure" $EMAIL
touch /tmp/envd.nocheck
fi
fi



-Sri

[Rodney Hills]

The /etc/inittab file (do man inittab) has entries to start scripts at boot time.

One of the options is "respawn" which will start a script at boot, and if it terminates, will relaunch it.

HTH

-- Rod Hills

[Deoncia Grayson_1]

Hey Sri,


Thanks for the script but your script doesn't really just stops at checking for the process it goes out and restart it as well.

[Sridhar Bhaskarla]

Yes. It does and I believe that's your primary requirement. If it couldn't start your program successfully, then it will send you a mail and then will stop monitoring until you fix the problem and remove the stamp file. This is one of the simple logics I use for in my serviceguard scripts.

-Sri

[Cheryl Griffin]

This is a common question.
Be sure to check out the answer here as well:
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=199874

There might be some other alternatives as well.

[RAC_1]

kill -0 $(ps -ef|grep [r]clproc|awk '{print $2}' || "start_process"

"start_process" refers to starting the process. If it is being started from /sbin/init.d/rclproc, then the above command look like follows.

kill -0 $(ps -ef|grep [r]clproc|awk '{print $2}' || /sbin/init.d/rclproc start

Anil

[Fred Ruffet]

You should consider running this program as respawn in /etc/inittab. Such programs are automaticaly re-run when they end (kill or normal end).

Regards,

Fred

[Deoncia Grayson_1]

Hey Sri,

I think you misunderstood me, I don't want it to restart the process, if the process is already running, if the process is still running then it should just sleep and if its not found then restart it.

[Sundar_7]

Deoncica,

"I don't want it to restart the process, if the process is already running, if the process is still running then it should just sleep and if its not found then restart it. "

The above description is confusing for sure, atleast to me. :-)

- Sundar.

[Sundar_7]

OK, I think I understand what you meant.

Sri script is meant to be executed from cron. Schedule that script to run once in every N minutes via cron.

If you dont want to do that, you can use the script excerpt from my post, which uses an infinite loop with sleep command.

[Sridhar Bhaskarla]

Hi,

I guess I know what you meant. It doesn't really restart the process but it only displays the message. Replace the function check_pid like this

check_pid()
{
MESG=$1
STATUS=$2

UNIX95= ps -C "$PROC" > /dev/null 2>&1
if [ $? != 0 ]
then
echo "$(date): $PROC $MESG" >> $LOGFILE
$STARTCMD
else
STATUS=0
fi

}

[Deoncia Grayson_1]

Hey Everyone,

Thanks for your help. I got it now. :)

De

[Deoncia Grayson_1]

Thanks for all your help, the scripts helped a lot!