[GERONIMO-761] Transient failures during full rebuild (OOM?)

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.0-M3
• Fix Version/s: 1.0-M3
• Component/s: buildsystem
• Assignee: Unassigned
• Reporter: Aaron Mulder
• Created: 15/Jul/05 10:25
• Updated: 07/Aug/06 15:05
• Resolved: 25/Jul/05 14:04

Description

I can't often run a full "maven -o m:rebuild-all" and make it all the way through. Usually, it dies around the time of the Geronimo assembly module. The errors vary, but include:

• bootstrap:bootstrap null
• error in plugin script
  A subsequent "maven -o" completes successfully (having it skip tests/itests to begin with also works), leading me to believe that somehow the amount of work starves the VM of something, probably memory, but slightly smaller quantities of work don't. Though after the plugin error the next build complains about a bad plugin cache and has a lot of stack traces complaining about JEXL errors, but after that things are back to normal.

Someone reported a similar error to the user or dev list, but I think their message was still different from mine.

This happens to me on Linux (SuSE Pro 9.3) with JDK 1.4.2_08 and on Mac (OS X 10.4.1 and 10.4.2) with JDK 1.4.2_07. I have not tried extending the Maven memory settings like we used to recommend for JDK 1.5.

I wonder if making more of the tests fork would help.