In addition I've found that if I have 3 child builds and 2 pass and 1 builds, but is unstable. I find that something kills the child build before the email is sent.
{{
10:12:49 Recording test results
10:12:51 Archiving artifacts
10:13:10 Email was triggered for: Unstable
10:13:10 Sending email for trigger: Unstable
10:13:11 ERROR: Could not send email as a part of the post-build publishers.
10:13:11 java.lang.InterruptedException
10:13:11 at java.lang.Object.wait(Native Method)
10:13:11 at hudson.remoting.Request.call(Request.java:122)
10:13:11 at hudson.remoting.Channel.call(Channel.java:630)
10:13:11 at hudson.EnvVars.getRemote(EnvVars.java:202)
10:13:11 at hudson.model.Computer.getEnvironment(Computer.java:779)
10:13:11 at hudson.model.Run.getEnvironment(Run.java:1766)
10:13:11 at hudson.model.AbstractBuild.getEnvironment(AbstractBuild.java:732)
10:13:11 at hudson.plugins.emailext.ExtendedEmailPublisher.createMail(ExtendedEmailPublisher.java:276)
10:13:11 at hudson.plugins.emailext.ExtendedEmailPublisher.sendMail(ExtendedEmailPublisher.java:220)
10:13:11 at hudson.plugins.emailext.ExtendedEmailPublisher._perform(ExtendedEmailPublisher.java:212)
10:13:11 at hudson.plugins.emailext.ExtendedEmailPublisher.perform(ExtendedEmailPublisher.java:174)
10:13:11 at hudson.tasks.BuildStepMonitor$3.perform(BuildStepMonitor.java:36)
10:13:11 at hudson.model.AbstractBuild$AbstractRunner.perform(AbstractBuild.java:649)
10:13:11 at hudson.model.AbstractBuild$AbstractRunner.performAllBuildSteps(AbstractBuild.java:625)
10:13:11 at hudson.model.AbstractBuild$AbstractRunner.performAllBuildSteps(AbstractBuild.java:603)
10:13:11 at hudson.model.Build$RunnerImpl.cleanUp(Build.java:170)
10:13:11 at hudson.model.Run.run(Run.java:1405)
10:13:11 at hudson.matrix.MatrixRun.run(MatrixRun.java:137)
10:13:11 at hudson.model.ResourceController.execute(ResourceController.java:88)
10:13:11 at hudson.model.Executor.run(Executor.java:145)
10:13:11 Finished: UNSTABLE
}}
I encounter the same "problem" on my matrix builds.
Not sure if this is the expected behaviour ? because it doesn't alwasy occurs.
Sometimes the failure of one configuration in the matrix will stop the other configurations and sometimes the other configurations will continue until their end.
Anyway it would be interesting either not to interrupt the other configurations or to have a parameter to control that.