karaf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grzegorz Grzybek <gr.grzy...@gmail.com>
Subject Re: [UPDATE] MavenTest itest fails randomly but often
Date Wed, 29 Nov 2017 05:16:20 GMT
@Mikael - that's what I observed too - 2.20.1 had this (intermittent) 30
seconds timeout after test finishes.

I don't have time (before Code Europe conf.) to check what's the problem
now, but I won't just leave it unexplained...

best regards
Grzegorz Grzybek

2017-11-28 23:10 GMT+01:00 Mikael Åsberg <m.asberg.watch@gmail.com>:

> I've had problems with Pax Exam and surefire after version 2.18.1 of
> surefire. Version 2.18.1 was working fine, but versions after that would
> often fail with errors along these lines:
> [ERROR] org.apache.maven.surefire.booter.SurefireBooterForkException:
> ExecutionException The forked VM terminated without properly saying
> goodbye. VM crash or System.exit called?
> Version 2.20.1 finally fixed the above problem but now there is a 30
> second wait after each test, but no crash...so I am still stuck at 2.18.1
> for Pax Exam. :(
> I've been meaning to report it, but didn't get around to it.
>
>
>
>
> On 2017-11-27 08:28, Grzegorz Grzybek wrote:
>
>> Hello
>>
>> I found that maven-surefire-plugin 2.20+ has some problems with pax-exam.
>> I'd love to investigate the exact cause, but didn't have time yet...
>> The surefire/failsafe issue is
>> https://issues.apache.org/jira/browse/SUREFIRE-1374
>>
>> And in my (jboss fuse) integration tests I stayed with 2.19.x.
>>
>> regards
>> Grzegorz Grzybek
>>
>> 2017-11-27 7:32 GMT+01:00 Jean-Baptiste Onofré <jb@nanthrax.net>:
>>
>> Hmm, not sure it's related as we use the same KarafTestSupport in bunch of
>>> other itest.
>>>
>>> I will exclude the MavenTest for now, and investigate.
>>>
>>> Thanks anyway !
>>>
>>> Regards
>>> JB
>>>
>>>
>>> On 11/26/2017 09:50 AM, Giuseppe Gerla wrote:
>>>
>>> Hi JB
>>>> although the tests worked on my PC, very often they went wrong on the
>>>> jenkins machine.
>>>> I did several experiments to find the problem, but it was very difficult
>>>> to
>>>> understand it.
>>>> What I understood is that for some reason the container is too slow to
>>>> come
>>>> up and the timeout of pax expired before. Why? I don't know.
>>>> At the end, I clone the activemq project on the jenkins machine, I
>>>> start a
>>>> build and it works. So I did a diff between activemq itest configuration
>>>> and my itest configuration. The result was that I removed some
>>>> initialization from my itest:
>>>>
>>>>           customOptions.add(junitBundles());
>>>>           customOptions.add(KarafDistributionOption
>>>>               .editConfigurationFilePut("etc/
>>>> org.apache.karaf.features.cf
>>>> g",
>>>> "featuresBoot",
>>>>                                         "(aries-blueprint, bundle,
>>>> config,
>>>> deployer, diagnostic, feature, instance, jaas, kar, log, management,
>>>> package, service, shell, shell-compat, ssh, system, wrap)"));
>>>>
>>>> then also:
>>>>
>>>>          MavenUrlReference karafStandardRepo =
>>>> maven().groupId("org.apache.karaf.features")
>>>>
>>>> .artifactId("standard").classifier("features").type("xml").
>>>> versionAsInProject();
>>>>
>>>>    customOptions.add(KarafDistributionOption.features(
>>>> karafStandardRepo,
>>>> "scr"));
>>>>
>>>> and I replaced "activemq-broker-noweb" feature with "activemq" one.
>>>> I rememeber that also surefire plugin version has a rule in this
>>>> problem.
>>>> So I changed my poms as following:
>>>>
>>>> PARENT-POM
>>>>
>>>>
>>>>    <surefire.plugin.version>2.16</surefire.plugin.version>
>>>>                                  <pluginManagement>
>>>> <plugins>
>>>> <plugin>
>>>> <artifactId>maven-surefire-plugin</artifactId>
>>>> <version>${surefire.plugin.version}</version>
>>>> <configuration>
>>>> <excludes>
>>>> <exclude>**/*AcceptanceTest.java</exclude>
>>>> </excludes>
>>>> </configuration>
>>>> </plugin>
>>>> </plugins>
>>>> </pluginManagement>
>>>>
>>>>
>>>>
>>>> ITEST-PROJECT-POM
>>>>
>>>>                           <plugin>
>>>> <artifactId>maven-surefire-plugin</artifactId>
>>>> <configuration>
>>>> <forkCount>1</forkCount>
>>>> <reuseForks>false</reuseForks>
>>>> <redirectTestOutputToFile>true</redirectTestOutputToFile>
>>>> <argLine>-Xmx512m</argLine>
>>>> <excludes>
>>>> </excludes>
>>>> <systemPropertyVariables>
>>>> <activemqVersion>${activemq.version}</activemqVersion>
>>>> <karafVersion>${karaf.version}</karafVersion>
>>>> </systemPropertyVariables>
>>>> </configuration>
>>>> </plugin>
>>>>
>>>>
>>>> I hope this can help you.
>>>>
>>>>
>>>> Regards
>>>> Giuseppe
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> 2017-11-26 7:17 GMT+01:00 Jean-Baptiste Onofré <jb@nanthrax.net>:
>>>>
>>>> Hi Giuseppe,
>>>>
>>>>>
>>>>> not sure I follow you. You mean you change the itest on your local
>>>>> copy ?
>>>>>
>>>>> Regards
>>>>> JB
>>>>>
>>>>>
>>>>> On 11/25/2017 11:05 PM, Giuseppe Gerla wrote:
>>>>>
>>>>> Hi JB
>>>>>
>>>>>> If you remember i had the same problem.
>>>>>> After several week and several investigativo, i discovered some
>>>>>> difference
>>>>>> beteeen my itest and activemq itest. So i changed my itest sturtup.
>>>>>> Now
>>>>>> It
>>>>>> seems solved.
>>>>>>
>>>>>>
>>>>>> Regards
>>>>>> Giuseppe
>>>>>>
>>>>>>
>>>>>> Il 25 nov 2017 17:44, "Jean-Baptiste Onofré" <jb@nanthrax.net>
ha
>>>>>> scritto:
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>>
>>>>>>> randomly (only for this itest), the container never came up (pax
>>>>>>> exam).
>>>>>>>
>>>>>>> Regards
>>>>>>> JB
>>>>>>>
>>>>>>> On 11/25/2017 09:56 AM, Giuseppe Gerla wrote:
>>>>>>>
>>>>>>> Hi JB
>>>>>>>
>>>>>>> what type of failure?
>>>>>>>>
>>>>>>>> Regards
>>>>>>>> Giuseppe
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> 2017-11-25 7:20 GMT+01:00 Jean-Baptiste Onofré <jb@nanthrax.net>:
>>>>>>>>
>>>>>>>> Hi guys,
>>>>>>>>
>>>>>>>>
>>>>>>>> Our Jenkins jobs are not very stable (both master and PR)
due to the
>>>>>>>>> MavenTest which randomly fails (but often).
>>>>>>>>>
>>>>>>>>> I will remove this test from master and move it to a
dedicated PR
>>>>>>>>> to
>>>>>>>>> investigate (but at least it won't impact our nightly
builds and
>>>>>>>>> PRs).
>>>>>>>>>
>>>>>>>>> Regards
>>>>>>>>> JB
>>>>>>>>> --
>>>>>>>>> Jean-Baptiste Onofré
>>>>>>>>> jbonofre@apache.org
>>>>>>>>> http://blog.nanthrax.net
>>>>>>>>> Talend - http://www.talend.com
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>>
>>>>>>>>
>>>>>>>> Jean-Baptiste Onofré
>>>>>>> jbonofre@apache.org
>>>>>>> http://blog.nanthrax.net
>>>>>>> Talend - http://www.talend.com
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>
>>>>> Jean-Baptiste Onofré
>>>>> jbonofre@apache.org
>>>>> http://blog.nanthrax.net
>>>>> Talend - http://www.talend.com
>>>>>
>>>>>
>>>>>
>>>> --
>>> Jean-Baptiste Onofré
>>> jbonofre@apache.org
>>> http://blog.nanthrax.net
>>> Talend - http://www.talend.com
>>>
>>>
>>
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message