Inline... > -----Original Message----- > From: Moderated discussion of advanced .NET topics. > [mailto:[EMAIL PROTECTED] Behalf Of J. Merrill > Sent: Wednesday, January 28, 2004 4:19 PM > To: [EMAIL PROTECTED] > Subject: Re: [ADVANCED-DOTNET] Garbage collection problem > > > At 03:12 AM 1/27/2004, Bert Roos wrote (in part) > >Hi, > > > >We're having a strange problem that seems to be related to > garbage collection. We're performing load testing through a > load generator tool (LoadRunner). At a certain load level, > the response times show strange and heavy fluctuations. > Request handling times go up by a factor 10 over an extended > period of time (several minutes, up to 10 minutes). Analysis > of performance counters gave some interesting results: > > > >* Context switches go up heavily > >* Contention rate/sec (.NET CLR LocksAndThreads) goes up heavily > >* Time spent in GC (.NET CLR Memory) goes up heavily > >* Gen 0 heap size (.NET CLR Memory) falls back from fluctuating around 5MB to > >a flat 1MB > >* Gen 1 heap size (.NET CLR Memory) falls back from fluctuating around 850KB > >to nearly nearly flat 350KB > >* Gen 2 heap size (.NET CLR Memory) keeps fluctuating around 29MB. At one > >point, it goes up a few MBs in but this might be unrelated. > > > >The test is running on a 4 processor server with 1.5GB RAM. > >[snip] > > I'm not sure I can help, but this info might be useful to > someone -- but it's possible that pointing your finger at GC > is premature. > > Are you using the CLR ThreadPool, or are you doing your > threading "naked"? Is there a correlation between the number > of CLR threads and the number of context switches? (If you > all of a sudden have 350 threads, when at a slightly lower > load level you had 100, you'd expect context switches to go > up significantly.)
The server is accessed through remoting, so the threads come from the thread pool. The number of logical threads goes up with the request duration, which make sense since new requests come in while they're slowly handled. It reaches a maximum of 39, which is not at all dramatical. > > Is the server being tested handling TCP/IP requests (is it a > web server?), or are the requests coming in some other way? > If the former, did you collect any TCP/IP performance counters? It's used through remoting. > > What OS is the server running? I would guess that adding > another 2.5GB RAM would help. It's running Windows 2000 Server. As the private bytes of the process reach a mere 123MB, I don't think additional memory would do any good. > > Good luck! > > J. Merrill / Analytical Software Corp > > =================================== > This list is hosted by DevelopMentor� http://www.develop.com > Some .NET courses you may be interested in: > > NEW! Guerrilla ASP.NET, 26 Jan 2004, in Los Angeles > http://www.develop.com/courses/gaspdotnetls > > View archives and manage your subscription(s) at http://discuss.develop.com =================================== This list is hosted by DevelopMentor� http://www.develop.com Some .NET courses you may be interested in: NEW! Guerrilla ASP.NET, 26 Jan 2004, in Los Angeles http://www.develop.com/courses/gaspdotnetls View archives and manage your subscription(s) at http://discuss.develop.com
