LinuxQuestions.org - [SOLVED] monitoring the time in a process with shell

I have a shell script that is doing stuff, submitting other processes mainly.
Now I have a timer withing a while loop that monitors if the stuff has been running too long it will exit.

The problem is that I am running the program overnight and I now had a problem due to change of month.

I post few snippets from the script:

Code:

#!/bin/sh



gettime () {



years=`date +"%y"`

months=`date +"%m"`

days=`date +"%d"`

hours=`date +"%H"`

mins=`date +"%M"`

secs=`date +"%S"`

gettime=`echo $years*31104000+$months*2592000+$days*86400+$hours*3600+$mins*60+$secs | bc`



echo $gettime



}



starttime=`gettime`



submit_run_process.sh



actualtime=`gettime`

time_diff=`echo $actualtime - $starttime | bc`



echo "Runtime: " $time_diff

  if [ $time_diff -lt 7200 ] ; then

    echo "Runtime less than 2 hours from submit, suspicious - exit"

    exit

  fi

The problem is that my script assumes that month is 30 days and the multiplicator in seconds is 30*24*3600=2592000, now in May we had 31 days so the script had suddenly negative runtimes between May 31 and June 1 around midnight (which is then less than required 7200 seconds), ok I could take a multiplicator 31*24*3600 which would pass the if sentence, but in months with less than 31 days I would always have more than the required 2 hrs even if it was not so.

I am not keen on putting the whole calender into my subroutine and I do not want to bind another language into this, I would rather stick to shell, but at the moment I am a bit short of ideas. Any ideas? I recall there was some command that counts seconds from start? use uptime or so instead of date?