[
https://issues.apache.org/jira/browse/MAHOUT-822?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tom White updated MAHOUT-822:
-----------------------------
Attachment: MAHOUT-822.patch
> The fact that 0.23 is alpha is probably precluding Mahout release with
> dependencies on it.
Correct, the hadoop-0.23 profile is just to allow folks to build Mahout against
0.23 in development at the moment.
> On top of it, we of course need to fix compatibility issues caught up by the
> tests.
Of course, however there is no reason they have to all be done in this JIRA is
there?
Having said that, I've gone through the test failures and fixed the following
(see the latest patch):
* ConditionalEntropyTest, EntropyTest, InformationGainRatioTest,
InformationGainTest are fixed by MAPREDUCE-3433
* TestMapReduce and TestDistributedRowMatrix now pass (there was a change in
the semantics of FileSystem#listStatus between 0.20 and 0.23)
The following tests are still failing (help fixing them gratefully received!):
* TestMeanShift no longer produces an error, but fails because count is 5 not 3.
* TestCanopyCreation and TestKmeansClustering are failing because the data is
coming back in a different order. Can the tests be relaxed?
* LocalSSVDSolverDenseTest and LocalSSVDSolverSparseSequentialTest are failing
but it's hard to say why since there is no debug output. The failure is likely
due to the gymnastics BtJob goes through to run jobs with multiple outputs.
> Mahout needs to be made compatible with Hadoop .23 releases
> -----------------------------------------------------------
>
> Key: MAHOUT-822
> URL: https://issues.apache.org/jira/browse/MAHOUT-822
> Project: Mahout
> Issue Type: Improvement
> Components: build
> Affects Versions: 0.6
> Reporter: Roman Shaposhnik
> Attachments: MAHOUT-822-build.patch.txt, MAHOUT-822.patch,
> MAHOUT-822.patch, MAHOUT-822.patch, MAHOUT-822.patch
>
>
> As part of the Hadoop stack integration project (Apache Bigtop) we are now
> trying to compile Mahout's upcoming 0.6 release against Hadoop 0.22 and 0.23.
> I'm attaching the patch to Mahout's Maven build system that made it possible.
> I would also like to request help in solving the real issues that poped
> up when we tried to compile Mahout:
> http://bigtop01.cloudera.org:8080/job/Bigtop-hadoop22/COMPONENT=mahout,label=centos5/6/console
> {noformat}
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/Step0JobTest.java:[182,33]
> org.apache.hadoop.mapreduce.TaskAttemptContext is abstract; cannot be
> instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/Step0JobTest.java:[218,9]
> org.apache.mahout.df.mapreduce.partial.Step0JobTest.Step0Context is not
> abstract and does not override abstract method getInputSplit() in
> org.apache.hadoop.mapreduce.MapContext
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/Step0JobTest.java:[229,12]
> cannot find symbol
> [ERROR] symbol : constructor
> Context(org.apache.hadoop.conf.Configuration,org.apache.hadoop.mapreduce.TaskAttemptID,<nulltype>,<nulltype>,<nulltype>,<nulltype>,<nulltype>)
> [ERROR] location: class org.apache.hadoop.mapreduce.Mapper.Context
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/common/DummyRecordWriter.java:[68,18]
> org.apache.hadoop.mapreduce.Mapper.Context is abstract; cannot be
> instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/common/DummyRecordWriter.java:[77,19]
> org.apache.hadoop.mapreduce.Reducer.Context is abstract; cannot be
> instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/PartialSequentialBuilder.java:[110,30]
> org.apache.hadoop.mapreduce.TaskAttemptContext is abstract; cannot be
> instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/PartialSequentialBuilder.java:[206,28]
> org.apache.hadoop.mapreduce.JobContext is abstract; cannot be instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/PartialSequentialBuilder.java:[227,30]
> org.apache.hadoop.mapreduce.TaskAttemptContext is abstract; cannot be
> instantiated
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/MockContext.java:[30,6]
> org.apache.mahout.df.mapreduce.partial.MockContext is not abstract and does
> not override abstract method getInputSplit() in
> org.apache.hadoop.mapreduce.MapContext
> [ERROR]
> /mnt/jenkins/workspace/workspace/Bigtop-hadoop22/COMPONENT/mahout/label/centos5/build/mahout/rpm/BUILD/apache-mahout-c298f70/core/src/test/java/org/apache/mahout/df/mapreduce/partial/MockContext.java:[38,10]
> cannot find symbol
> {noformat}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
