We are getting the same exception in case of publishing JUnit results. But in our case "Max # of builds to keep" is not set. Builds which throw this exception are run on Windows slaves.

As soon as we switched for these jobs from xUnit plug-in to standard "Publish JUnit test result report" the problem disappeared.

Jenkins ver. 1.501
xUnit plug-in ver. 1.52

I have the same NPE running on Windows 2008R2 64bit, Jenkins 1.504 and NUnit plugin 0.14. This issue only arose when I upgraded from 1.501 to 1.504 (the next day), but it didn't start immediately and I have other NUnit publications that are working without issue. I do not have the job currently configured to "Discard Old Builds" and I have a around 170 builds in my build history.

Recording NUnit tests results
ERROR: Publisher hudson.plugins.nunit.NUnitPublisher aborted due to exception
java.lang.NullPointerException
at hudson.model.Run.getRootDir(Run.java:935)
at hudson.tasks.junit.TestResultAction.getDataFile(TestResultAction.java:91)
at hudson.tasks.junit.TestResultAction.load(TestResultAction.java:147)
at hudson.tasks.junit.TestResultAction.getResult(TestResultAction.java:97)
at hudson.tasks.junit.TestResultAction.getResult(TestResultAction.java:55)
at hudson.tasks.test.AbstractTestResultAction.findCorrespondingResult(AbstractTestResultAction.java:183)
at hudson.tasks.test.TestResult.getPreviousResult(TestResult.java:145)
at hudson.tasks.junit.SuiteResult.getPreviousResult(SuiteResult.java:296)
at hudson.tasks.junit.CaseResult.getPreviousResult(CaseResult.java:375)
at hudson.tasks.junit.CaseResult.freeze(CaseResult.java:486)
at hudson.tasks.junit.SuiteResult.freeze(SuiteResult.java:338)
at hudson.tasks.junit.TestResult.freeze(TestResult.java:604)
at hudson.tasks.junit.TestResultAction.setResult(TestResultAction.java:74)
at hudson.tasks.junit.TestResultAction.<init>(TestResultAction.java:67)
at hudson.plugins.nunit.NUnitPublisher.recordTestResult(NUnitPublisher.java:150)
at hudson.plugins.nunit.NUnitPublisher.perform(NUnitPublisher.java:109)
at hudson.tasks.BuildStepMonitor$3.perform(BuildStepMonitor.java:36)
at hudson.model.AbstractBuild$AbstractBuildExecution.perform(AbstractBuild.java:814)
at hudson.model.AbstractBuild$AbstractBuildExecution.performAllBuildSteps(AbstractBuild.java:786)
at hudson.model.Build$BuildExecution.post2(Build.java:183)
at hudson.model.AbstractBuild$AbstractBuildExecution.post(AbstractBuild.java:733)
at hudson.model.Run.execute(Run.java:1592)
at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:46)
at hudson.model.ResourceController.execute(ResourceController.java:88)
at hudson.model.Executor.run(Executor.java:237)

It appears that I have managed to work around the issue by deleting the build where the issue first arose. While doing this I noticed something strange:

• the build that first failed had a time stamp of 20:39
• the successful build before that was time stamped 19:30 (failed-1)
• the successful build before that was time stamped 15:50 (failed-2)
• the 15:50 folder contained a junitResult.xml time stamped at 19:30 ???
• the build at 19:30 did not contain a junitResult.xml
• the build at 19:30 stated that "Notifying upstream projects of job completion" was a "SUCCESS"

It appears that the build at 19:30 physically wrote its results to the previous builds folder, looking at the Test history it shows that build at 15:50 had twice as may tests run as normal.

Deleting the 19:30 folder resolved the issue for me.

Tried reproducing the issue by the max # of builds to keep but that didn't work.

From reading the other comments here, looks like it is not trivial to reproduce this. Maybe by having used xunit and/or junit plug-in in older builds?

Could someone with this issue provide some example test data, or a scenario that can be reproduced? If so I can debug it and confirm what's happening.

Thanks