Dieter Maurer wrote:
Richard Ettema wrote at 2004-6-4 13:21 +0100:
... but is there anything I should be looking at that could be the cause of these problems?
I gave you some advice how to analyse the SIGSEGV -- but you have ignored it...
Your advice was not ignored. I asked the guy running our server about using ulimit, he tells me he has done this but not found anything obvious leading to a cause.
To analyse leaks, Shanes "LeakFinder" product might help you.
From what I understand Leakfinder will not work on zope2.6.2/python2.1. I even tried installing it just incase what I had read was old. I modified leakfinder as mentioned in another thread to fix a change to a cache method taking 2 arguments rather than the 3 it use to. Although I could get leakfinder to install I could not get it to run correctly, so came to the conclusion what i read about it not working was true.
Moreover, there are several threads about tracking down memory leaks. Maybe, you search the archives?
I have read alot of threads on memory leaks, tried alot of things including removing Products and areas of the site to narrow down the problem area with no success. removing use of request.set and session.set, remove all ram caching, simplyfying/cleaning up scripts. I posted this new info hoping that finding something as significant as running zope with a single thread hides both problems might mean something obvious to someone else. (well significant to me anyway as its the most progress I have made on this issue for a while) I have read about the posibility of this situation being linked to a c extension but have not come across much in the way of going about solving it. Thank you for your ideas and I apologise for not replying to your previous message. It's a bit mad here trying to deal with this and keep everything running. Cheers, Richard.