You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zookeeper.apache.org by Patrick Hunt <ph...@apache.org> on 2015/04/20 23:18:27 UTC
Re: Jenkins pre-commit build is bogus?
Look near the bottom of the console output.
It failed for two reasons - no tests as part of the patch, and the c client
tests failed. Jenkins doesn't know how to report the c client tests, just
the java.
Patrick
On Mon, Apr 20, 2015 at 2:04 PM, Flavio Junqueira <
fpjunqueira@yahoo.com.invalid> wrote:
> While looking at ZK-2124, I checked the report #2640 and it says no test
> failures (https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/ <
> https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/>). The
> jenkins summary on the jira correctly reports that there has been core test
> failures, though. Any clue of what needs to be fixed?
>
> -Flavio
Re: Jenkins pre-commit build is bogus?
Posted by Chris Nauroth <cn...@hortonworks.com>.
The JUnit tests I mentioned were in fact passing, but JUnit4ZKTestRunner
was logging a failure message anyway. The root cause is that the runner
was treating any exception as a failure, without awareness for tests that
had been annotated @Test(expected=...) to indicate an exception is
expected. I filed a patch on ZOOKEEPER-2174 to make JUnit4ZKTestRunner
check the annotation. Hopefully that will prevent similar confusion in
the future.
I guess that just leaves the C client tests as actual failures from this
pre-commit run.
Chris Nauroth
Hortonworks
http://hortonworks.com/
On 4/20/15, 2:32 PM, "Chris Nauroth" <cn...@hortonworks.com> wrote:
>In addition to the C client test failures, there were also JUnit failures.
> One example is ByteBufferInputStreamTest. Here we can see that Jenkins
>reports it as Passed, but the console output shows that it failed.
>
>https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/testReport/or
>g
>.apache.zookeeper.server/ByteBufferInputStreamTest/testReadArrayOffsetLeng
>t
>h_0Length/
>
>
>As I recall, the Jenkins JUnit reporting mechanism works via string
>parsing of the console output. I wonder if something about an exception
>propagating out of JUnit4ZKTestRunner causes console output that the
>Jenkins reporting doesn't understand.
>
>Chris Nauroth
>Hortonworks
>http://hortonworks.com/
>
>
>
>
>
>
>On 4/20/15, 2:18 PM, "Patrick Hunt" <ph...@apache.org> wrote:
>
>>Look near the bottom of the console output.
>>
>>It failed for two reasons - no tests as part of the patch, and the c
>>client
>>tests failed. Jenkins doesn't know how to report the c client tests, just
>>the java.
>>
>>Patrick
>>
>>
>>On Mon, Apr 20, 2015 at 2:04 PM, Flavio Junqueira <
>>fpjunqueira@yahoo.com.invalid> wrote:
>>
>>> While looking at ZK-2124, I checked the report #2640 and it says no
>>>test
>>> failures (https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/
>>><
>>> https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/>). The
>>> jenkins summary on the jira correctly reports that there has been core
>>>test
>>> failures, though. Any clue of what needs to be fixed?
>>>
>>> -Flavio
>
Re: Jenkins pre-commit build is bogus?
Posted by Chris Nauroth <cn...@hortonworks.com>.
In addition to the C client test failures, there were also JUnit failures.
One example is ByteBufferInputStreamTest. Here we can see that Jenkins
reports it as Passed, but the console output shows that it failed.
https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/testReport/org
.apache.zookeeper.server/ByteBufferInputStreamTest/testReadArrayOffsetLengt
h_0Length/
As I recall, the Jenkins JUnit reporting mechanism works via string
parsing of the console output. I wonder if something about an exception
propagating out of JUnit4ZKTestRunner causes console output that the
Jenkins reporting doesn't understand.
Chris Nauroth
Hortonworks
http://hortonworks.com/
On 4/20/15, 2:18 PM, "Patrick Hunt" <ph...@apache.org> wrote:
>Look near the bottom of the console output.
>
>It failed for two reasons - no tests as part of the patch, and the c
>client
>tests failed. Jenkins doesn't know how to report the c client tests, just
>the java.
>
>Patrick
>
>
>On Mon, Apr 20, 2015 at 2:04 PM, Flavio Junqueira <
>fpjunqueira@yahoo.com.invalid> wrote:
>
>> While looking at ZK-2124, I checked the report #2640 and it says no test
>> failures (https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/
>><
>> https://builds.apache.org/job/PreCommit-ZOOKEEPER-Build/2640/>). The
>> jenkins summary on the jira correctly reports that there has been core
>>test
>> failures, though. Any clue of what needs to be fixed?
>>
>> -Flavio