Jenkins seems to pass
-Dmaven.test.failure.ignore<http://maven.apache.org/plugins/maven-surefire-plugin/test-mojo.html#testFailureIgnore>=true
as default. If I set
-Dmaven.test.failure.ignore<http://maven.apache.org/plugins/maven-surefire-plugin/test-mojo.html#testFailureIgnore>=false,
the build breaks on the failed unit test. I think, this is done to let
jenkins finish the build with whatever else needs to be done (collecting
data, clean up, ...).

But still I found some strage behaviour in maven projects:

- If I use the "Build other projects"-option in "Post-build
Actions"-section with "Trigger only if build succeeds" selected, the
projects to build are NOT triggered if there were unit test failures. Good.

- If I use the same option "Trigger only if build succeeds" for the "Post
Steps"-section the configured post steps are still executed, even if there
were unit test failures. Bad.

Shouldn't the two switches have the same effect?

If you agree, I open a bug for this.

Regard,

Frederik




2012/4/11 Wannes Sels <[email protected]>

> Maven reports BUILD SUCCESS even though there are test failures, this is
> the reason why your build succeeds.
> Maybe you already pass 
> -Dmaven.test.failure.ignore<http://maven.apache.org/plugins/maven-surefire-plugin/test-mojo.html#testFailureIgnore>=true
> as a parameter to the maven calls? You could try adding 
> help:effective-pom<http://maven.apache.org/plugins/maven-help-plugin/effective-pom-mojo.html>as
>  a maven call, to see if the property is set.
>
> --
> Wannes Sels
>
>
> On Wed, Apr 11, 2012 at 11:45, Frederik Fromm <
> [email protected]> wrote:
>
>> Thanks for the quick answer. But perhaps I need to add more detail. In
>> the same log I find:
>> ...
>> [ERROR] There are test failures.
>> ...
>> [INFO]
>> ------------------------------------------------------------------------
>> *[INFO] BUILD SUCCESS*
>> [INFO]
>> ------------------------------------------------------------------------
>> [INFO] Total time: 3:26.538s
>> [INFO] Finished at: Wed Apr 11 10:09:40 CEST 2012
>> [INFO] Final Memory: 37M/90M
>> [INFO]
>> ------------------------------------------------------------------------
>> Waiting for Jenkins to finish collecting data
>> channel stopped
>> *Triggering projects: childProject
>> *Finished: SUCCESS
>> The "*Triggering projects: childProject"* should not be the result, if
>> "Run only if build succeeds" is set.
>>
>> I want that failed unit tests break the build and no child jobs should be
>> triggered.
>>
>> Any more thoughts?
>>
>> Thanks again,
>>
>> Frederik Fromm
>>
>>
>> 2012/4/11 Adrien Lecharpentier <[email protected]>
>>
>>> I don't think this is a bug because if you run the same goals on a dev
>>> computer, you will see that with tests failure, maven set the build result
>>> to "build failure".
>>>
>>> You can add an option to maven to prevent it to set the build result to
>>> "build failure" : -Dmaven.test.failure.ignore=true.
>>>
>>> I think this is not a really good idea because a test failure should be
>>> a critical point and you process should end when there are tests
>>> failure/error as well as when there are compilations errors.
>>>
>>> --
>>> Adrien Lecharpentier
>>>
>>>
>>> Le 11 avril 2012 11:23, ffromm <[email protected]> a écrit :
>>>
>>> Hi,
>>>>
>>>> I have a question about the execution of post build steps in maven
>>>> builds with test failures:
>>>>
>>>> In a maven project there are 3 options to run post steps:
>>>>
>>>> - Run only if build succeeds
>>>> - Run only if build succeeds or is unstable
>>>> - Run regardless of build result
>>>>
>>>> The last one is the default.
>>>>
>>>> I understand these option that choosing the first one "Run only if
>>>> build succeeds" no post steps are executed if the build is unstable.
>>>> No shell scripts, no trigger of other builds, etc. Is that right?
>>>>
>>>> Because, right now, having "Run only if build succeeds" set, I find
>>>> "[ERROR] There are test failures." in the build's log an and got a
>>>> yellow ball, but still the configured child job gets triggered and
>>>> executed using the parameterized trigger plugin.
>>>>
>>>> Is this intended to be like this or is this a bug?
>>>>
>>>> Any explication is appreciated. Thanks.
>>>>
>>>> Frederik Fromm
>>>
>>>
>>>
>>
>

Reply via email to