Do you know if you have a session timeout set?
Sometimes in CWP if you don't have an exit session method, I've seen sessions pile up and freeze WPC
Also, I hope you've built in error catching and processing for 401. If it's that common an error I'm sure you have, but you made it sound suspect from your logs.
Lastly, I've also seen WPC lock upon large data sets, especially when lots of finds are involved. Is this a table with a few hundred, thousand, hundreds of thousand of records?
Actually I forgot filemaker addressed the session issue, and does not use persistent sessions. There is still a timeout setting though.
Another thing to look at (besides the error handling Mike alluded to) is your Find request construction. How are you building your Find requests in the PHP code? Are they simple / complex? Are they executing against indexed fields? Number vs. text fields?
Narrowing down the circumstances can help you troubleshoot.
Check the web server logs. We found that robots were responsible for triggering wildcard searches that coincided with WPE problems. We added appropriate instructions in a robots.txt file but a chinese search bot continues to ignore it.
When I look at the logs, I see no indication of why it shuts down. Mostly 401 errors, resulting from a find where no records were found.
I'm at a loss to figure out how to troubleshoot this.
Thanks to you all for the input. You have given me places to look.
That's a good tip, I didn't even think about this because all of my ODBC searches are behind closed doors anyways, as I use a different login system for CWP.