Am Dienstag, den 20.06.2006, 15:02 +0200 schrieb Bernd Fondermann: > Norman Maurer wrote: > > Hi guys, > > > > any has ever test the memory usage of james when used in big enviroment > > and cannot deliver mail. We otice that if we use james and it try to > > deliver emails is seems that it never release the memory after timeout. > > So the load and memory usage get bigger with every message which can not > > get delievert. If we use a stadardgateway for outgoing email and james > > will so never gt a timout the memoryusage and load keeps constant. > > Its not uncommon for Java VMs to _visibly_ free memory only when really > needed. This is totally under the control of the JVM. In real life, when > you see constantly growing heap sizes it may or may not be due to > unreleased memory or leaks. You can only definitivly say you ran short > of memory when you are seeing an OutOfMemory exception. Memory > organisation and garbage collection in Java is a science of its own.
Yes thats right but i set -Xmx256M . So it should not use more then that. But java consumn all of me free 1,8 GB and after that also consumn 2 GB swap. That not normal! > > > Bernd, have you ever tried this in postage ?Anyone else have expirence? > > Sure, I am tracking memory usage under Postage. If you have any > questions concerning that, I'll be happy to answer them. > > Depending on the configuration (repository types) the 'memory > fingerprint' of James varies, but not dramatically. Sooner or later, > with reasonable load, you run out of memory. Optimisation in this area > should be a topic for a future release. > > > We use the current 2.3-branch from svn. > > With high enough load, you run into problems in 2.2 and 2.3. > > Bernd > Yeah thats right. But i not seen a OOM yet. It only consums all Memory and setup high load. Thats all! Here a bit more infos: Kernel: 2.6.15-23-amd64-generic OS: Ubuntu Dapper amd64 java: 1.5 DeliveryTrheads: 1x5 , 1x3 ,1x1 Spools: mysql 4.1 bye Norman
signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil