Shell Programming and Scripting

Hi,
in my bash script I start 3rd party app that runs quickly normally and returns some data that I use later for processing. However if there are problems the app doesn't return anything and just hangs... then my script also hangs of course, awaiting for the app to comeback. Was wondering how to avoid that so the script could carry on or at least bailout with an error if that happens. Is there a way to do it in bash ?

Code:

get_proc_stats() {
     port=$1
     tstats=`app $port`
}

I was thinking maybe I could start/fork seperate process in background which would then check the main process somehow, maybe sending a signal USR1 and then the main process would be maybe writing something to some temp. file and forked process runnig in background would be checking if update to temp. file happend, if not then it could wait some time and kill the main process ?
What do you think?

Any hints on that appreciated,
Cheers,
Tomasz

Moderator's Comments:

Please use code tags next time for your code and data.

Cleanest way would be to find out what is producing this hang.
You can always have another script or function that checks for the process. Though you might want to have something like a kind of deadline in there, so there can be "measured" when it is time to kill the process like a time threshold for example.
But something like that should be a workaround if the investigation didn't work or takes too long.

yes, we have got a patch and now the app seem to work stable.
But this topic still is interesting how to achieve in a shell script. In popular languages you can do things just like:

Code:

if (trysys(app))
   code if app fails
else
  proceed with the script

Maybe the simplest way would be to start a subshell (by using &)and run that app in there, the main script would then loop around some time checking the PID of a subshell from that script, or just would wait some time for the subshell to finish - is there a way to check subshell is running or finished ? And then based on that I could proceed in main script. I guess that must be doable in bash. If anyone has done something like that and don't mind sharing I would welcome a template with syntax

Cheers,
Tomasz

Moderator's Comments:

edit by bakunin: Please view this code tag video for how to use code tags when posting code and data.

When you send a process in the background you can use the "jobs" command to get its PID. Using this PID you can terminate it by sending signal 15 (or, if this doesn't help, 9).

The display format of "jobs" might differ slightly from OS to OS, so you will have to fine-tune the following sketch to work on yours. I have not tested it yet, so you might have to adapt it a bit too:

Code:

....

controljob ()
{
typeset    chJob="$1"           # some command line
typeset -i iTimeOut=$2          # time to wait in seconds
typeset -i iPID=0
typeset -i iCnt=0

$chJob &
iPID=$(jobs | sed, awk, ...)    # whatever is necessary for your system

while [ $iCnt -lt $iTimeOut -a $(ps -p $iPID | wc -l) -gt 0 ] ; do
     sleep 1
     (( iCnt += 1 ))
done

if [ $(ps -p $iPID | wc -l) -gt 0 ] ; then
     kill -15 $iPID
     sleep 3                    # the process may need some time to terminate
fi

if [ $(ps -p $iPID | wc -l) -gt 0 ] ; then
     kill -9 $iPID
     exit 1
fi

exit 0
}

Moderator's Comments:

finally i'd like to ask you not to ignore requests by the moderators again. zaxxon has asked you to use CODE-tags and in your next post you didn't use them again. Thank you for your consideration.

I hope this helps.

bakunin