This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 170741 - Profiler timeout
Summary: Profiler timeout
Status: RESOLVED INVALID
Alias: None
Product: profiler
Classification: Unclassified
Component: Base (show other bugs)
Version: 6.x
Hardware: PC All
: P3 blocker (vote)
Assignee: issues@profiler
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-08-21 09:36 UTC by artisan
Modified: 2009-10-27 18:21 UTC (History)
0 users

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
while running at 100% cpu (17.93 KB, text/plain)
2009-08-21 09:37 UTC, artisan
Details
log while netbeans is running the profiler. maybe it contains a clue. (96.09 KB, text/plain)
2009-08-21 09:39 UTC, artisan
Details

Note You need to log in before you can comment on or make changes to this bug.
Description artisan 2009-08-21 09:36:37 UTC
Profiling a JBoss 4.2 instance gives a timeout message. The popup dialog asks to disconnect the profiler or to continue 
the session. If the session is not aborted eventually it all starts to work though. But in the meantime a cpu runs at 100% 
until finally the profiler becomes responsive.

As I can't ask netbeans to profile its profiler I'm including a thread dump of jvisualvm. It's taken while one of the 
cpu's ran at 100%
Comment 1 artisan 2009-08-21 09:37:54 UTC
Created attachment 86519 [details]
while running at 100% cpu
Comment 2 artisan 2009-08-21 09:39:14 UTC
Created attachment 86520 [details]
log while netbeans is running the profiler. maybe it contains a clue.
Comment 3 Jiri Sedlacek 2009-10-27 18:21:39 UTC
Actually it works as designed. When profiling memory the profiler needs to change the bytecode for each class loaded in
the JVM. This takes a significant amount of time for large applications like JBoss and consumes a lot of CPU cycles in
native code.

During instrumentation the profiled application doesn't communicate with the profiler tool, that's why the dialog pops
up to give you a chance to stop the process. It doesn't seem like we could speed up the instrumentation so the only way
is to submit the dialog several times until all classes are instrumented.

Based on what's your intention you may just take a heap dump when in Monitor mode (== fast startup) using Profile | Take
Heap Dump... to start analyzing classes/instances and/or searching for a memory leak.