Shell Script checking processes

[Carlos A. Garcia]

Hi everybody, haven´t been able to come up with a decent shell script to do this for me:
For a few servers running OS=HP-UX 11.0

1. Check on processes running from a file, third column of a file
2. If running do nothing (and if was previously "not running" send e-mail saying is UP again and exit, once it has notified, don't bother to inform again!).
3. If not running, send e-mail to inform (keep the status to check on next run if process came up).

Hope I was "crystal clear" on my question!
Does anybody have such a script?, since I believe must be a common issue actually, any help would be greatly appreciated.
Couldn't get it to work properly myself.
Thanks a lot

[Graham Cameron_1]

www.bb4.com

-- Graham

[Todd McDaniel_1]

Here is a rudementary script that will handle one process. You can expand it to cover more than one.

Where $VAL -ge 1, the 1 is the number of procs you are validating are supposed to be running. In my case the number is 4, but for you maybe only 1

-------------------------------------------
VAL=0
echo $VAL

VAL=`ps -aef|grep |grep -v grep|grep -v <scriptname> |wc -l`

echo $VAL
if [ $VAL -ge 1 ]
then echo "Services are running";ps -aef|grep -e -e opl |grep -v
grep|grep -v <scriptname>;date
else mailx -s "Services" root < /root/process.notice
fi
--------------------------------------------

[Carlos A. Garcia]

Thanks a lot Graham, actually I'm a HP-OVO's admin, the weird thing about it, is our Operations staff won't consider OV's notifications good enough, and would like to have mail notifications (for processes down) and a single notification whenever a process comes up again.
I'm already using opcmsg to accomplish part of it, but OP would like e-mail notifications.

Thanks again.

[Michael Schulte zur Sur · Accepted solution]

Hi,

here is another script.

greetings,

Michael

[Carlos A. Garcia]

Thanks a lot Todd and Michael!
Todd, your script works great, had my own script working close to that, but still doesn't provide with the information/status, regarding whenever a process previously was DOWN and now is back UP again, I did test it but didn't seem to work that way.

Now, Michael, yours works great, but seems I really have a problem getting correctly the processes from my file, such as:

COMMAND=`cat /var/opt/OV/conf/OpC/procmon.conf | awk '{ print $3 }'`

Is there a better way to do this?A proper way?
Since this file /var/opt/OV/conf/OpC/procmon.conf looks like this:
1 OS inetd Exact 1 Critical 2
1 OS sshd Min 1 Critical 2

Any help is much appreciated guys, thanks for your quick and proper response.
Best regards.

[Michael Schulte zur Sur]

Hi,

if you have more than one process in your config file, you have a problem.
try this

while read F1 F2 PROCESS F4 F5 F6
do
checkproc ${PROCESS} &
done < /var/opt/OV/conf/OpC/procmon.conf

greetings,

Michael

[Michael Schulte zur Sur]

Hi,

I have redone my script.

greetings,

Michael

[Carlos A. Garcia]

Michael, just perfect!
I've only modified this:
"Min") COMPAREON[${NUMPROC}]="-eq";;
Replaced "-gt" with "-eq", so it notifies perfectly whenever a process comes up again.

I really appreciate guys your help, and even more, in such a short period of time!
Thanks again!
If there is a way to assign more point to you guys,let me know!

Best regards,
C.

[Michael Schulte zur Sur]

Hi Carlos,

just curious, can you explain the meaning of the fields in the config file? I thought Min means minimum, so if 2 processes are running then the script will complain again. The script was designed to mail once, when the minimum number of processes are running. Am I missing something?

greetings,

Michael

[Carlos A. Garcia]

Hi Michael, yes, sure (this is a conf file I'm using for Openview's script, so, that is why I keep the same format):

1=check for process
OS=OS process
inetd=name of process
Exact=condition Min-Max or Exact
1=number of times running (Exact 1 time, Min 1 time) etc...
Critical=status IF down
2=frequency

And in order to e-mail once, had to add the following test:
else
test ${NOTIFIED[${COUNT}]}="YES" && echo "Service down...etc...

This way it only sends one e-mail to notify for a process DOWN, and only one e-amil to notify a process is UP again, works great!
Thanks again!

Best regards,
C.