[jira] [Commented] (SUREFIRE-1264) Some tests can be lost when running in parallel with parameterized tests

["Tibor Digana (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SUREFIRE-1264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16192759#comment-16192759 ] 

Tibor Digana commented on SUREFIRE-1264:
----------------------------------------

[~cbeikov]
[~christian.beikov]
Maybe we have talked about it but I would ask again. 
Do you have a Thread Dump when Surefire 2.18 hanged in project {{calcite-core}}? 
Can you attach it?
Please use version 2.18 and not 2.18.1.
The old version definitely did not have new threading mechanism when Maven sent tests to run to Surefire process via native process pipes.
It looks to me like the test class name was swallowed somewhere.
Are you open for building some Surefire branch by yourselves and perform some experiments on your side?
Perhaps it would be nice dumping this cross-process communication when debugging Maven build like this {{mvn -X test}} or {{mvn --debug test}}.
Thx.

> Some tests can be lost when running in parallel with parameterized tests
> ------------------------------------------------------------------------
>
>                 Key: SUREFIRE-1264
>                 URL: https://issues.apache.org/jira/browse/SUREFIRE-1264
>             Project: Maven Surefire
>          Issue Type: Bug
>          Components: Junit 4.7+ (parallel) support
>    Affects Versions: 2.19.1
>         Environment: Maven 3.3.9
> Java 1.8.0_45 (Oracle)
> System: OS X
> Reproduced on Linux too, with both OpenJDK 7 and Java 7 from Oracle.
>            Reporter: Jean-Luc Derrien
>            Assignee: Tibor Digana
>             Fix For: 2.20.1
>
>         Attachments: failure-2.21-snapshot-SUREFIRE-1264.zip, pom.xml
>
>          Time Spent: 144h
>  Remaining Estimate: 24h
>
> Hello,
> It appears some tests can be lost when using the parallel mode with parameterized tests. Here is a [small project|https://github.com/jderrien/surefire-junit-tests/tree/simple-1] I've used to try to reproduce the issue we have seen.
> This is not something that happens on every run, but it's quite frequent.
> With this loop, the problem should appear in less than 2 minutes, maybe on the first run when (un)lucky:
> {code}
> time while true; do mvn clean test > last.log ; tail -25 last.log ; if [ "$(grep -c 'Tests run: 12' last.log)" == "0" ]; then break; fi ; done
> {code}
> Normal run:
> {noformat}
> -------------------------------------------------------
>  T E S T S
> -------------------------------------------------------
> Running [p2]
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p2] - p2 - sleeptime = 1 => start
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p2] - p2 - sleeptime = 1 => stop
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p1] - p1 - sleeptime = 2 => start
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p1] - p1 - sleeptime = 2 => stop
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p0] - p0 - sleeptime = 5 => start
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p0] - p0 - sleeptime = 5 => stop
> Tests run: 6, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 5.006 sec - in [p2]
> Running [p2]
> Tests run: 6, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.011 sec - in [p2]
> Results :
> Tests run: 12, Failures: 0, Errors: 0, Skipped: 0
> {noformat}
> When tests have been lost (here one test lost according to the output):
> {noformat}
> -------------------------------------------------------
>  T E S T S
> -------------------------------------------------------
> Running [p2]
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p0] - p0 - sleeptime = 5 => start
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p2] - p2 - sleeptime = 1 => start
> com.appnexus.viewability.core.surefireJunitTests.ATest.methodA1[p1] - p1 - sleeptime = 2 => start
> Tests run: 5, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.007 sec - in [p2]
> Running [p2]
> Tests run: 6, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.012 sec - in [p2]
> Results :
> Tests run: 11, Failures: 0, Errors: 0, Skipped: 0
> {noformat}
> Note: there are 3 test classes and 18 tests on the [master branch|https://github.com/jderrien/surefire-junit-tests/tree/master] and it's even more frequent/easy to reproduce.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)