I've read over your message several times, but I can't figure out how my version of p4pr.perl could be at fault. I'm sure I could be missing something, but I don't even have any idea where to start looking.

The way Unix in general and Linux in particular works, if a process spans child processes and then exits, then child processes are automatically inherited by init. If they're defunct, "init" will "wait" for them and they'll go away.

If the defunct processes you are seeing persist after p4pr.perl has exited, there is pretty much no way they could have been started by p4pr.perl in the first place. That is, unless your "init" program is extremely broken, which is unlikely.

Your observation that these processes go away when you restart httpd suggests that in fact the processes are children of httpd, not children of p4pr.perl. You can confirm this by running "ps -f -p <pid>" on one of the defunct processes to get its parent's PID, and then running the same command on that PID to find out what the parent process is.

I don't have any guesses about why you might be seeing this problem with my version of the script and not with Bob Sidebotham's version. I just reviewed the full diff between his version and mine, and I don't see anything that might explain that particular difference in behavior.

I wish I could help, but I don't see anything more I can do at this point.

jik

P.S. I'm not on perforce-user, so please CC me if you respond in this thread and want me to see it.

P.P.S. Perforce at Cisco? I thought Cisco was a ClearCase shop? At least, they were pushing ClearCase heavily when I was there.