I have an automated process that uses cron and wget to hit a method in Zope daily. Normally in wget, I first get a 401, which prompts wget to send basic auth, which it sends upon resending the request with the right user/pw in headers, and then gets a 200 (success) after a 15 to 20 minutes (it's a big, slow process). Lately, though, I have been getting an HTTP 500 almost instantaneously on occasion, and the logs seem to verify this. Retrying the process usually solves this problem, but I am trying to figure out why the 500 error is occurring. I am running Zope 2.3.2 with ZEO (1.0b3 + modifications from CVS earlier this summer), and one of my 2 ZEO clients is having this problem and not the other. I have squid in front of this setup for Virtual host support and Squid load-balances the 2 ZEO clients, and the requests, when sent to one of my ZEO clients works fine, but seems to fail every time. The box that seems to be having the problem seems to work fine in other respects. Anybody have any thoughts on what problems might trigger an HTTP 500 in Zope? Sean ========================= Sean Upton Senior Programmer/Analyst SignOnSanDiego.com The San Diego Union-Tribune 619.718.5241 sean.upton@uniontrib.com =========================
sean.upton@uniontrib.com writes:
Anybody have any thoughts on what problems might trigger an HTTP 500 in Zope? Provided that the 500 response is generated by Zope and not the SQUID in front, it means that there has been an exception that ZPublisher is unable to map to a different response code (such as 'Unauthorized' and 'redirect').
Catch exceptions in your process and print them. Then you get a better feeling what goes wrong. If the response is generated by SQUID, then Zope died (or timed out). In this case, use the "-M" option (-> mailing list archives), to analyse the circumstances of the deaths. Dieter
participants (2)
-
Dieter Maurer -
sean.upton＠uniontrib.com