Hello,

I have a bash script the checks a directory ever 1 (for testing) and if a file(s) is found if gets the old file and then executes another script by"

if condition

/path/postscript arguments ....

fi

I seem to hit a bottleneck of some short ... my logs show that every 1-2 seconds I am calling the other script.

But I can never get more then 40-45 instances of the postscript to run ... I have a max value of 99 for testing so I want to try and hit that mark.

Any suggestions on how I could increase the execution time ??

Michael.

the only think i can think of is to redirect stderr and stdout from the second script to /dev/null

There must be something inside that script making it slow. Calling a minimal script on my 450 MHz Pentium-II with FreeBSD-4.x one tousand times takes less than two seconds.

In general, you can speed things up by not calling, but sourcing in new script. But watch for side-effects, variable settings are effective in the parent script and exit() will exit the outer shell.

That means you write
. foobar/newscript.sh
instead of
sh foobar/newscrip.sh

That is called "sourcing" the new script instead of "executing". Doing so is making it a factor of 16 faster for me.

But obviously the target script needs to be written with this in mind because of the side effects.

Now, here's the real MyCoy. You can shield yourself from the side effects by executing a subshell first:
(. foobar/newscript.sh)

That is much slower than just sourcing, but it is still twice as fast as just calling the script. The reason is that you have to fork() which is expensive but at least you don't exec(). Since the minimal script is almost empty this cuts the systemcalls in half.

Again, here's the timings:
executing: 500 calls/second
sourcing: 8000 calls/second
sourcing in subshell: 1000 calls/second

Hello,

After placing the script in debig mode I found a performance problem and corrected the issue. It is much better now. Thanks for all the pointers.

Michael.