Try to handle multiple connections better. The session database is a hot spot. When multiple requests (e.g. 20) come in at the same time session database contention can get great. The original code didn't retry session database access when the open failed. This resulted in errors at the client. The second pass delayed 0.01 seconds and retried. It was better but we still had multiple second stalls. I think the first request got in, everybody else backed up and then retried at the same time. Again they stepped on each other. With logging I would see many counters go all the way to low single digits or to -1 indicating falure. This pass uses randomint to generate delays from 0-.125 seconds in 5ms increments. This performs better in testing. I rarely saw a counter less than 13 (2 failed retries). Current logging starts after 6 failures and counts down until success or failure.

See <a href="http://dev.zope.org/Wikis/DevSite/Projects/ZPT">the
ZPT project Wiki</a> for more information about Page Templates, or
<a href="http://www.zope.org/Members/4am/ZPT">the download page</a>
for installation instructions and the most recent version of the software.

This Product requires the TAL and ZTUtils packages to be installed in
your Python path (not Products).  See the links above for more information.