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.
Summary: | Can't profile WebApp on GlassFish if OOME detection heapdump directory has space in path | ||
---|---|---|---|
Product: | profiler | Reporter: | Ivan Sidorkin <ivansidorkin> |
Component: | Base | Assignee: | issues@profiler <issues> |
Status: | VERIFIED FIXED | ||
Severity: | blocker | CC: | pjiricka, vkraemer |
Priority: | P1 | ||
Version: | 6.x | ||
Hardware: | All | ||
OS: | Windows XP | ||
Issue Type: | DEFECT | Exception Reporter: | |
Bug Depends on: | 106620 | ||
Bug Blocks: | |||
Attachments: |
server log
domain.xml with -XX:HeapDumpPath="C:\Documents and Settings\jis\My Documents\TomcatJSPExample\nbproject\private\profiler" causing the failure |
Description
Ivan Sidorkin
2007-04-24 13:01:25 UTC
Created attachment 41540 [details]
server log
In fact this IS somehow related to Issue 92919 - it's introduced by the OOME detection mechanism, but is a separate problem. This happens when OOME detection is enabled and heapdump target is set to a directory containing space in path. As a workaround, disable OOME detection in Profiler Options or set the heapdump destination to a folder not containing spaces in path. This may be a GlassFish bug with incorrectly parsed quoted JVM arguments, needs further investigation for 6.0. Created attachment 41546 [details]
domain.xml with -XX:HeapDumpPath="C:\Documents and Settings\jis\My Documents\TomcatJSPExample\nbproject\private\profiler" causing the failure
Vince made me aware of this issue - this sounds really nasty. Changing priority to P1 - please evaluate whether this is a J1 showstopper. Vince, what do you think? P1 - Product feature does not work, no workaround exists P2 - Product feature doesn't work, a workaround may exist but it's difficult to use or impractical Sorry but no word "nasty" in P1 definition. Do you have any concrete reason which overrides the above definitions? I think this is a p2 by the criteria. It is going to be tricky for folks on the show floor... If M9 is going to be 'pushed' at JavaOne those floor headaches will transfer to a fair number of users. As a workaround we could disable the OOME detection for J2EE projects for M9 until this is fixed. That sounds like a good short-term "fix"... We still need to work through to a real solution for the trunk. > I think this is a p2 by the criteria.
Ok, changing to P2. (Previously this was a P3, which was not appropriate either.)
Implemented a workaround fixing this issue - OOME detection is disabled for all J2EE applications. Will file new issue for not working OOME detection. Fixed for M10. Based on discussion with the team this can be seen as a serious problem for M9 as J1 preview release, profiling on GlassFish is broken on Windows when default settings are used. It would be better to have this fixed for J1. Marking as J1_STOPPER and raising priority to P1 in order to be able to fix for M9. Fix merged into release60_m9 branch. Checking in profiler-cluster.zip; /cvs/profiler/external/profiler-cluster.zip,v <-- profiler-cluster.zip new revision: 1.44.2.1; previous revision: 1.44 done verified in 070502 Reopening the issue. This is glassish or glassfish-plugin related bug. Filed an issue to glassfish plugin (the issue #106620) fixed as of NB build 200706201800 Verified with Beta 1 |