Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 248166

Summary:

many JUnit's crash VM on Linux i386

Product:

[WebTools] Web Tools

Reporter:

David Williams <david_williams>

Component:

General

Assignee:

jst-inbox <jst-inbox>

Status:

RESOLVED DUPLICATE

QA Contact:

David Williams <david_williams>

Severity:

normal

Priority:

CC:

cbridgha, deboer, neil.hauge, thatnitind

Version:

3.0.1

Target Milestone:

---

Hardware:

OS:

Linux-GTK

Whiteboard:

Bug Depends on:

248210

Bug Blocks:

Attachments:

Description	Flags
dump file	none
another dump file	none
another dump file	none
another dump file	none
another dump file	none
another dump file	none

Description David Williams

2008-09-22 12:27:23 EDT

Created attachment 113151 [details]
dump file

After moving to Eclipse I20080909-1121, I noticed a common set of JUnit's "crashing" on my local build machine, which is Linux, but i386 based (not PPC, as our production build server is). 

They all crashed with a similar error. The VM dump saying something about UnixProxyProvider. 

So, makes me wonder if related to bug 245850. 

The special file I/O module is not provided for PPC, so would make sense it would not appear on production machine, even if but on Linux, i386. 

The following JUnit suites are the ones that "crashed" (repeatedly). 


org.eclipse.jpt.core.tests.internal.JptCoreTests
org.eclipse.jst.j2ee.tests.bvt.AutomatedBVT
org.eclipse.jst.jsf.core.tests.AllTests
org.eclipse.jst.jsp.core.tests.JSPCoreTestSuite
org.eclipse.jst.server.tomcat.core.tests.AllTests
org.eclipse.wst.common.frameworks.tests.bvt.AutomatedBVT
org.eclipse.wst.css.tests.encoding.CSSEncodingTestSuite
org.eclipse.wst.html.tests.encoding.HTMLEncodingTestSuite
org.eclipse.wst.server.core.tests.AllTests

Comment 1 David Williams

2008-09-22 12:29:07 EDT

Created attachment 113152 [details]
another dump file

Comment 2 David Williams

2008-09-22 12:29:45 EDT

Created attachment 113153 [details]
another dump file

Comment 3 David Williams

2008-09-22 12:30:12 EDT

Created attachment 113154 [details]
another dump file

Comment 4 David Williams

2008-09-22 12:31:06 EDT

Created attachment 113155 [details]
another dump file

Comment 5 David Williams

2008-09-22 12:32:35 EDT

Created attachment 113156 [details]
another dump file

There was a few more ... but I think the pattern is very similar so will stop with this one.

Comment 6 David Williams

2008-09-22 12:35:56 EDT

I've added several of you Project Leads hoping one of you has or tests on a Linux i386 box and can decide if this is a problem we observe elsewhere. 

If so ... we should open bug on platform as I think something went wrong with their bug 245850.

Comment 7 Tim deBoer

2008-09-22 13:15:01 EDT

I can confirm the server JUnits have never failed here before. It's crashing when trying to create a file in the workspace, which has nothing to do with the test. I agree this looks like a regression in the platform.

Comment 8 David Williams

2008-09-22 14:19:09 EDT

I tried the HTML Encoding test suite, from a dev. environment, and there was no
crash. 

But that was with the final M2 from platform ... so either not easy to
reproduce, or maybe it was just that one driver. 

We'll move to M2 and give that a try.

Comment 9 David Williams

2008-09-22 18:24:15 EDT

I am still seeing the JUnit failures after moving up to M2 platform. 

I've opened bug 248210 against Platform resources.

Comment 10 David Williams

2008-09-23 00:56:03 EDT

I am not, however, able to re-create this in my dev. env., which is a different machine than what the build/JUnits are running on ... my Linux dev. env., is on a slow, single processor machine. 

Tim, or others, if you are able to reproduce in dev. env. with M2, you might give "steps to reproduce and describe your machine some. Maybe it's only on dual processor's?  Does it happen with plain 'ol SDK? Or only with WTP installed? (where more classes would be loading). 

Thanks

Comment 11 David Williams

2008-10-17 08:40:32 EDT

FYI, on my test build machine, I've used a work around as noted in bug 248210#c14

Comment 12 David Williams

2008-10-17 19:13:07 EDT


*** This bug has been marked as a duplicate of bug 248210 ***