On 10/9/07, Maciej Wisniowski <maciej.wisniowski@coig.katowice.pl> wrote:
I set the number of threads: zserver-threads 25 I set: pool-size 27 on <zodb_db main> I think you should rather use ZEO than raise number of threads to such a big value. Also there are some rules about pool-size. I don't remember exactly but seems to me that it is too small, but in general I suppose it has nothing to do with your problem...
Situation you're seeing may appear if there is something that blocks all your threads, like all threads are waiting for RDBMS response...
How can i run some kind of debug on zope (debug-mode on dont log anything more) to check this next time happen? Use DeadlockDebugger and if this doesn't help try: http://www.zopelabs.com/cookbook/1073504990
DeadlockDebugger certainly helped me in a similar situation. I went through similar blind flailing machinations - modifying threads and pool sizes (in the end I found not a Zope or Plone coding problem but some of my Javascript code was inadvertently occasionally causing a tight looped remote access of a generated page --effectively a self inflicted DOS attack!)