Le 1/31/13 7:46 PM, Mondain a écrit : > I'll see if I can get some unit tests together to possibly locate anything > that may be leaking in Mina 2.0.8.
What would be interesting is to runa profiler to see if we have some memory leaks. Regarding the reported 'bug', this is clearly not a bug, but a normal behaviour. Even under a heavy load - and I conducted some performance tests two weeks ago - the GC never ever get higher than 1 or 2 %. In fact, most of the CPU is consummed in the select(à method. It sounds to me that the guy who reported the issue has made some blind guess about what could occur when an iterator is used frequently. IMO, this is totally irrelevant. > > > On Thu, Jan 31, 2013 at 8:57 AM, Emmanuel Lécharny <elecha...@gmail.com>wrote: > >> Le 1/31/13 5:52 PM, Mondain a écrit : >>> I was talking to Jon Valliere this morning about GC with IOPS and I >>> wondered what the opinion of the Mina dev team is on this issue: >>> >> http://bugs.sun.com/bugdatabase/view_bug.do;jsessionid=ca40fa265e28a3ccb91527b258ff?bug_id=6924217 >>> How well is this handled by 2.x or 3.x? I personally use 2.0.8 with the >>> -XX:+UseParNewGC JVM option and it seems to work ok; but I'm fairly sure >>> there are still some leaks. >> The bugs just says that there are a lot of GCs, not that there is a leak >> anywhere. >> >> Is that a real problem ? >> >> -- >> Regards, >> Cordialement, >> Emmanuel Lécharny >> www.iktek.com >> >> > -- Regards, Cordialement, Emmanuel Lécharny www.iktek.com
