|
This looks great. Can you go ahead and list these tests into a Junit test suite class? I think that is the best way for now to capture them into a single suite that is run by the commit build.
I spoke with Nigel and, as discussed in
I've yet another conversation with Jacob wrt
On the other hand, a separate JUnit test suite won't give us any extra benefits compare to a flat test file. +1 on Changes to build.xml for the run-commit-tests target. The changes are similar to run-commit-tests target of HDFS.
It seems that indentation isn't consistent across the changes.
<fileset dir="${test.src.dir}/mapred" excludes="**/${test.exclude}.java"> <patternset> <includesfile name="@{test.file}"/> </patternset> </fileset> or <target name="run-test-mapred" depends="compile-mapred-test" description="Run mapred unit tests"> <macro-test-runner test.file="${test.mapred.all.tests.file}" /> </target> Looks good otherwise Patch fixing the indentation issue pointed out by Konstantin.
A few additional points: Code coverage (for the mapred package) of all-tests list is 76%. I have left out two tests (TestJobTrackerRestart and TestQueueManager) out of this list as these take about 7 and 6 minutes respectively. Separate effort is underway to refactor these to become unit tests and when completed, these tests will be added to the commit-tests list. The current tests run in less than 9 minutes, so adding these two tests, after they have been refactored, should still keep the run time down to 10 minutes. Code coverage (for the mapred package) of commit-tests + TestJobTrackerRestart + TestQueueManager (as they exist now) is 63% I'm still seeing two more misaligned spots:
1. Everything below <sequential> is a body of this element and suppose to be indented. + <macrodef name="macro-test-runner">
+ <attribute name="test.file" />
+ <sequential>
<delete dir="${test.build.data}"/>
<mkdir dir="${test.build.data}"/>
<delete dir="${test.log.dir}"/>
2. Same is here: body of <patternset> isn't indented. + <patternset>
+ <includesfile name="@{test.file}"/>
+ </patternset>
Looks good otherwise. Fixing the indentation (again !) The original source also had this mismatch and was carried over in the previous patch too
 -1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12415093/mapred-670-v2.patch  against trunk revision 799551. +1 @author. The patch does not contain any @author tags. +1 tests included. The patch appears to include 48 new or modified tests. +1 javadoc. The javadoc tool did not generate any warning messages. +1 javac. The applied patch does not increase the total number of javac compiler warnings. +1 findbugs. The patch does not introduce any new Findbugs warnings. +1 release audit. The applied patch does not increase the total number of release audit warnings. +1 core tests. The patch passed core unit tests. -1 contrib tests. The patch failed contrib unit tests. Test results: http://hudson.zones.apache.org/hudson/job/Mapreduce-Patch-vesta.apache.org/432/testReport/ This message is automatically generated. Streaming test failures are unrelated to the patch
I just committed this, Thanks Jothi
The patch shows that there should be a file called commit-tests as part of the patch, but when I go to the tip of the trunk, there is no such file commited.
Looks like 'svn add' has been omitted.
I added & committed the missing files.
Integrated in Hadoop-Mapreduce-trunk #38 (See http://hudson.zones.apache.org/hudson/job/Hadoop-Mapreduce-trunk/38/
). Adding the 'new' files which got missed in the original commit. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
Test
Resolved
Major
Create target for 10 minute patch test build for mapreduce
Attaching a spread sheet as a first step in this identification/classification of test suites into Fast Tests. The run time of these tests is marginally over 10 minutes (The entire test suite runs in about 2 hours). Effort is still on to improve a few more test cases that would hopefully bring this run time down even further.
The attached file has three sheets –
Clover was used for measuring code coverage.
Please share your thoughts/suggestions/ideas