Hi Yunan and Eugene, I moved up to the latest stable build of TPTP 4.5.2, and I made a step further when I run the profiler, I can now see some profiling data (for my packges/classes) showing up in the views, for Heap and Thread.
But I can't get it to work, for Heap, the JVM still hangs up for a quite number of times I try to rerun the test. I attached two log files, each file name shows if it is a Heap or Thread profile log, Take a look when you get a chance. http://www.nabble.com/file/p21987603/MLog04380_ThreadProf.zip MLog04380_ThreadProf.zip http://www.nabble.com/file/p21987603/MLog04356_HeapProf.zip MLog04356_HeapProf.zip Thanks Meena He, Yunan wrote: > > BTW, we can set two environment variables to output the log file. > > export MARTINI_LOGGER_DIRECTORY = $YOUR_LOG_DIR > export MARTINI_LOGGER_LOG_LEVEL = 5 > > It's worth to try to get the log file first. Thanks. > > Regards, > ---------- > Yunan He > China Runtime Technologies Lab, > SSG/SSD/MRTC, Intel > > -----Original Message----- > From: tptp-tracing-profiling-tools-dev-boun...@eclipse.org > [mailto:tptp-tracing-profiling-tools-dev-boun...@eclipse.org] On Behalf Of > He, Yunan > Sent: 2009年2月12日 10:13 > To: TPTP Tracing and Profiling Tools Project developer discussions > Subject: RE: [tptp-tracing-profiling-tools-dev] TPTP 4.5.1 using too much > disk space...., CPU peaks up to almost 100% > > Hi Meena, > > Could you please try with TPTP 4.5.2 for thread and heap scenario to see > if it can be reproduced? I will send you debug version for experiment > soon. > > Regards, > ---------- > Yunan He > China Runtime Technologies Lab, > SSG/SSD/MRTC, Intel > > -----Original Message----- > From: tptp-tracing-profiling-tools-dev-boun...@eclipse.org > [mailto:tptp-tracing-profiling-tools-dev-boun...@eclipse.org] On Behalf Of > EricMeena > Sent: 2009年2月11日 23:58 > To: tptp-tracing-profiling-tools-dev@eclipse.org > Subject: Re: [tptp-tracing-profiling-tools-dev] TPTP 4.5.1 using too much > disk space...., CPU peaks up to almost 100% > > > Hi Eugene, > > Yes, I have created a filter for the profiler to capture profiling > information only for my code, not other packages and also, I was getting > profiling data shown in the views (for Execution time analysis only) > > I realized that the following property is the one causing the problem: > execdetails=true. > > I changed the VM argument from: > -agentlib:JPIBootLoader=JPIAgent:server=controlled;CGProf:execdetails=true > to -agentlib:JPIBootLoader=JPIAgent:server=controlled;CGProf and every > thing > was working fine. I can run the test until it the end. With that property > set to "true", the CPU goes up to 99%. The disk space is consumed on fast > pace (1 GB is gone every 4 minutes or so), I can't figure out which > folder > this data is being writen to... > > The other thing is that I am not able to get any profiling data for Memory > (HeapProf) and ThreadProf. > I am trying to use the this setting: > -agentlib:JPIBootLoader=JPIAgent:server=controlled;HeapProf:allocsites=true > or -agentlib:JPIBootLoader=JPIAgent:server=controlled;ThreadProf and what > happens is that the JVM hangs up, I can't interact with my environment > (OSGi > command line) once I run the profiler (Attach to Agent), nothing is shown > in > the profile data views, except for the default packages (I guess > primitives) > in case of memory analysis, and my test just hangs up (of course with the > JVM being in that state...). > > Is there any way to start the profiler in a debug mode so that I can tell > what is happening? > > Thanks for your time. > > Meena > > > Eugene Chan wrote: >> >> Hi Meena, >> >> Did you configuration the profiling launch configuration (Attach to >> Agent) >> with filter that collects only the profiling data of your target >> application/package/class? The default filter configuration may include >> profiling data outside of your scope that slow down the process. Did you >> get any profiling data collected and shown in the views? >> >> >> Regards, >> ____________________________________________ >> >> Eugene Chan >> IBM Toronto Laboratory, Canada >> >> >> >> >> EricMeena <mauvv...@yahoo.fr> >> Sent by: tptp-tracing-profiling-tools-dev-boun...@eclipse.org >> 02/09/2009 05:08 PM >> Please respond to >> TPTP Tracing and Profiling Tools Project developer discussions >> <tptp-tracing-profiling-tools-dev@eclipse.org> >> >> >> To >> tptp-tracing-profiling-tools-dev@eclipse.org >> cc >> >> Subject >> [tptp-tracing-profiling-tools-dev] TPTP 4.5.1 using too much disk >> space...., CPU peaks up to almost 100% >> >> >> >> >> >> >> >> Hello, >> >> I am using TPTP Platform 4.5.1 with Eclipse Ganymede, and I am running >> this >> on Windows XP SP2, on Intel Core Duo 2.00GHz . I am trying to profile >> OSGI >> bundles that constitute a messaging infrastructure. >> >> I have configured the profiling agent to be launched in controlled mode >> ("-agentlib:JPIBootLoader=JPIAgent:server=controlled;CGProf:execdetails=true"). >> >> Every thing is good until I start profiling (create an instance of >> 'Attach >> to Agent', in the 'Agents' tab..). >> >> My Test case just generate a message traffic in the system, with routing >> components moving messages around, from/to other components... >> >> Once in profile mode, the whole System (in eclipse workbench) becomes >> extremely slow, CPU time rising too much reaching into high 90%, and the >> serious problem is how the disk free space is being eaten up to the max, >> until the system runs out of disk space. >> >> I found some where that TPTP generates some problem files, where do these >> files go? Is there a way to avoid this behavior? What could be killing >> the >> CPU? >> >> Any help appreciated. >> >> Meena >> >> >> -- >> View this message in context: >> http://www.nabble.com/TPTP-4.5.1-using-too-much-disk-space....%2C-CPU-peaks-up-to-almost-100--tp21920373p21920373.html >> >> Sent from the Eclipse - Tracing and Profiling Tools - Dev mailing list >> archive at Nabble.com. >> >> _______________________________________________ >> tptp-tracing-profiling-tools-dev mailing list >> tptp-tracing-profiling-tools-dev@eclipse.org >> https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev >> >> >> _______________________________________________ >> tptp-tracing-profiling-tools-dev mailing list >> tptp-tracing-profiling-tools-dev@eclipse.org >> https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev >> >> > > -- > View this message in context: > http://www.nabble.com/TPTP-4.5.1-using-too-much-disk-space....%2C-CPU-peaks-up-to-almost-100--tp21920373p21957935.html > Sent from the Eclipse - Tracing and Profiling Tools - Dev mailing list > archive at Nabble.com. > > _______________________________________________ > tptp-tracing-profiling-tools-dev mailing list > tptp-tracing-profiling-tools-dev@eclipse.org > https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev > > _______________________________________________ > tptp-tracing-profiling-tools-dev mailing list > tptp-tracing-profiling-tools-dev@eclipse.org > https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev > > -- View this message in context: http://www.nabble.com/TPTP-4.5.1-using-too-much-disk-space....%2C-CPU-peaks-up-to-almost-100--tp21920373p21987603.html Sent from the Eclipse - Tracing and Profiling Tools - Dev mailing list archive at Nabble.com. _______________________________________________ tptp-tracing-profiling-tools-dev mailing list tptp-tracing-profiling-tools-dev@eclipse.org https://dev.eclipse.org/mailman/listinfo/tptp-tracing-profiling-tools-dev
