Bugzilla – Full Text Bug Listing
| Summary: | Eclipse project Hudson builds should invoke build with architecture appropriate vm | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | Paul Webster <pwebster> | ||||||||||||||
| Component: | Releng | Assignee: | Kim Moir <kim.moir> | ||||||||||||||
| Status: | RESOLVED FIXED | QA Contact: | |||||||||||||||
| Severity: | normal | ||||||||||||||||
| Priority: | P3 | CC: | aniefer, d_a_carver, john.arthorne, kim.moir | ||||||||||||||
| Version: | 3.6 | ||||||||||||||||
| Target Milestone: | 3.7 M1 | ||||||||||||||||
| Hardware: | PC | ||||||||||||||||
| OS: | Linux | ||||||||||||||||
| Whiteboard: | |||||||||||||||||
| Attachments: |
|
||||||||||||||||
|
Description
Paul Webster
Yes, it's an error in our scripts. We should check for the architecture and invoke the build with the appropriate vm. (In reply to comment #1) > Yes, it's an error in our scripts. We should check for the architecture and > invoke the build with the appropriate vm. Yeah, I think Athena common builder may be running into some of this same issue as well. Created attachment 168425 [details]
patch
Created attachment 168430 [details]
patch
build2 is actually x86_64 arch
e4 build is running now on build2 with the correct vms. It will be interesting to see how fast it runs :-) https://build.eclipse.org/hudson/job/eclipse-e4-test/72/console It was relatively fast, but it failed: https://build.eclipse.org/hudson/job/eclipse-e4-test/72/consoleText /opt/users/hudsonbuild/workspace/eclipse-e4-test/builds/I201005131401/src/compile.master.xml:37: The following error occurred while executing this line: /opt/users/hudsonbuild/workspace/eclipse-e4-test/builds/I201005131401/src/plugins/org.eclipse.swt/buildFragment.xml:84: The following error occurred while executing this line: /opt/users/hudsonbuild/workspace/eclipse-e4-test/builds/I201005131401/src/plugins/org.eclipse.swt/buildFragment.xml:63: The following error occurred while executing this line: /opt/users/hudsonbuild/workspace/eclipse-e4-test/builds/I201005131401/src/plugins/org.eclipse.swt/buildFragment.xml:184: The following error occurred while executing this line: /opt/users/hudsonbuild/workspace/eclipse-e4-test/builds/I201005131401/src/plugins/org.eclipse.swt/buildFragment.xml:408: java.lang.StringIndexOutOfBoundsException: start > length() PW The build failed in the part where it replaced the ints with longs in the swt fragments. Hard to debug because I can't login to build2 with ssh and see the build artifacts. Investigating... When I made this change, I updated the bootstrap script that invokes the Hudson build to use a newer one. There have been changes to include the s390x platform recently. I think that this is where the build is failing - updating the build files for this platform from int to long in the swt fragment. John, could you sync with the map files from HEAD to include the recent changes for the e4 build in R4_HEAD of org.eclipse.releng. Some of the tags look out of date and this is configured as an integration build. (In reply to comment #8) > files for this platform from int to long in the swt fragment. John, could you > sync with the map files from HEAD to include the recent changes for the e4 > build in R4_HEAD of org.eclipse.releng. Some of the tags look out of date and > this is configured as an integration build. I've merged the releng map files into R4_HEAD, I'll launch a build after the e4 build completes. PW I started https://build.eclipse.org/hudson/job/eclipse-e4-test/74/ around 10:00 PW Removed the qnx fragment and tagged org.eclipse.platform-feature I've started another build - https://build.eclipse.org/hudson/job/eclipse-e4-test/75/ PW When I merged i accidentally took the sdk-tests feature from HEAD. I've fixed that and started build #77 PW I switched back to the slave for build 78, and I'm back to comment #6 PW I'm going to revert this builder back to the one we used to use for this e4 build and see what happens. All the feature changes you merged should still be okay, you don't need to make any changes. So Andrew and I looked at this issue today an we're not sure why it's happening. It looks like the replace of the int with longs already occurs in the swt fragments source. I'm going to try to run with ant in verbose mode. If this doesn't reveal the source of the problem I'll disable the parallel compilation we use in pde build. (Although this shouldn't be the source of the problem). Faster hardware sometimes reveals problems in build scripts, I recall a similar problem the last time I switched to a faster machine :-) Andrew, here's is the longer stack trace https://build.eclipse.org/hudson/job/eclipse-e4-test/88/console java.lang.StringIndexOutOfBoundsException: start > length() at org.apache.tools.ant.ProjectHelper.addLocationToBuildException(ProjectHelper.java:508) at org.apache.tools.ant.taskdefs.Ant.execute(Ant.java:418) at org.apache.tools.ant.UnknownElement.execute(UnknownElement.java:288) at sun.reflect.GeneratedMethodAccessor2.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) at java.lang.reflect.Method.invoke(Method.java:597) at org.apache.tools.ant.dispatch.DispatchUtils.execute(DispatchUtils.java:106) at org.apache.tools.ant.Task.perform(Task.java:348) at org.apache.tools.ant.Target.execute(Target.java:357) at org.apache.tools.ant.Target.performTasks(Target.java:385) at org.apache.tools.ant.Project.executeSortedTargets(Project.java:1337) at org.apache.tools.ant.helper.SingleCheckExecutor.executeTargets(SingleCheckExecutor.java:38) at org.eclipse.ant.internal.core.ant.EclipseSingleCheckExecutor.executeTargets(EclipseSingleCheckExecutor.java:30) at org.apache.tools.ant.Project.executeTargets(Project.java:1189) at org.apache.tools.ant.taskdefs.Ant.execute(Ant.java:416) ... 102 more Created attachment 169993 [details]
patch
Try removing the parallel compilation parameters and see if this makes a difference. This only impacts the builder for the e4 SDK build, not the regular 3.6 stream build.
Disabling parallel compilation didn't work. Back to square one. Sorry, can we turn verbose back on, I missed comment #17 and build 88 is gone now. Running another build now in verbose mode https://build.eclipse.org/hudson/view/Eclipse%20and%20Equinox/job/eclipse-e4-test/98/ (In reply to comment #21) > Running another build now in verbose mode > > https://build.eclipse.org/hudson/view/Eclipse%20and%20Equinox/job/eclipse-e4-test/98/ When I ran #99 I switched it back to master. Feel free to switch the config back to slave. PW Andrew requested a new build with Ant in verbose mode and parallel compilation disabled. It's running here https://build.eclipse.org/hudson/view/Eclipse%20and%20Equinox/job/eclipse-e4-test/100/ The root exception is java.lang.StringIndexOutOfBoundsException: start > length() at java.lang.AbstractStringBuilder.replace(AbstractStringBuilder.java:799) at java.lang.StringBuffer.replace(StringBuffer.java:391) at org.apache.tools.ant.taskdefs.Replace$Replacefilter.replace(Replace.java:288) at org.apache.tools.ant.taskdefs.Replace$Replacefilter.process(Replace.java:258) at org.apache.tools.ant.taskdefs.Replace.processFilterChain(Replace.java:668) at org.apache.tools.ant.taskdefs.Replace.processFile(Replace.java:613) at org.apache.tools.ant.taskdefs.Replace.execute(Replace.java:496) at org.apache.tools.ant.UnknownElement.execute(UnknownElement.java:288) I looked at the ant code and I don't see how this can happen so my only explanation is a vm bug. Probably jit related. I would suggest trying a different vm if possible Got past the vm issue, not I have this issue bug 320045. Once these utilities are installed by the webmaster, I'll run another build. Created attachment 174630 [details]
patch to fix java15-home path
Created attachment 174682 [details]
patch
Created attachment 174683 [details]
patch
patch
eclipse-equinox-test build is now successfully running on build2 which takes 92 minutes compared to 186 minutes on build.eclipse.org. I'll also backport these changes to the R4_branch so the 4.0 build can also run on the new hardware. Closing, changes also released to R4_HEAD stream. |