Peter N Lewis wrote:
I already have:
MaxRequestsPerChild 1000
So the worker processes should die and clean that out.
I'm suspicious that the worker threads are not dieing because I've seen cases where there are worker threads listed in ps but not listed in server-status (that will probably show up tomorrow or the next day when the server has been running a bit longer, and I'll post details of that if I haven't got a response by then).
You are talking about threads. What MPM are you using, on which platform ? Read http://httpd.apache.org/docs/2.2/mpm.html and try ./httpd -l You may need MaxRequestsPerThread instead.Also, there are subtleties as to when Perl interpreter threads/instances are created/re-used/killed, depending on the mod_perl configuration.
You may want to investigate that a bit more in-depth.All this is still a plaster over a real issue though. The real issue is that some module is leaking memory, and you should really find out which one and go from there. There may be an alternate module available or not, depending on the case. It may also just be an application, not a Perl library module, that leaks memory. I would consider that in fact much more likely.
What /are/ your "mod_perl pages" and what do they do ? Is there any possibility to narrow down the search to some likely suspects ?In the order of probabilities for memory leaks, I would consider, from most to least likely culprits :
1) perl application modules/pages 2) perl standard library modules handling XML 3) other perl standard library modules 4) Apache standard modules 5) Apache core code --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See <URL:http://httpd.apache.org/userslist.html> for more info. To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|