Thanks mark. Okay, that clarifies.

Iirc this kind of error only happens if mahout version is compiled for 1x.  I 
can run a test to see if bigtop 0.7.0 has the same issue. Right now I'm going 
to test that the issue goes away when yum installing mahout from cdh . 

That will confir it as a bug, then your idea to regression test will confirm it 
as a regression or not.

> On Sep 20, 2014, at 10:53 AM, "Mark Grover (JIRA)" <[email protected]> wrote:
> 
> 
>    [ 
> https://issues.apache.org/jira/browse/BIGTOP-1453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14142017#comment-14142017
>  ] 
> 
> Mark Grover commented on BIGTOP-1453:
> -------------------------------------
> 
> Jay. That "-1" is the package release identifier. It's just a naming thing, 
> it's used for all our packages.
> 
> I don't know much about #2, but I am assuming the error only happens with 
> Bigtop 0.8 RC and not Bigtop 0.7?
> 
>> 0.8 RC:  Looks like it is building Mahout1x ?
>> ---------------------------------------------
>> 
>>                Key: BIGTOP-1453
>>                URL: https://issues.apache.org/jira/browse/BIGTOP-1453
>>            Project: Bigtop
>>         Issue Type: Bug
>>   Affects Versions: 0.8.0
>>           Reporter: jay vyas
>>           Priority: Critical
>> 
>> Hi folks.  
>> Looks like rc08 is using the old mahout, not the new one.  I.e. its build 
>> for *hadoop 0-1x*.. not for *hadoop 2x*   .Poking aroudn on RC testing 
>> today, I found this. 
>> 1) yum list output shows:
>> {noformat}
>> mahout.noarch                                       0.9.6-1.el6              
>>                          @Bigtop
>> {noformat}
>> The 0.9.6-*-1* seems suspicious.
>> 2) And I ran the mahout tests on a clean RC08 repo, and I got some errors 
>> indicating that it was built for hadoop 1 {{class found but interface 
>> expected}}... indicating that it is an older mahout api which is running the 
>> job.
>> (a side note : i also  saw that pom.xml has references to hadoop 0.23, but 
>> looking at other mahout packages from vendors, its clear that some how that 
>> remains in the mahout pom for some reason or another , so thats not an 
>> issue.  the main evidence pointing to bad mahout version is that #2 error i 
>> sited above )
> 
> 
> 
> --
> This message was sent by Atlassian JIRA
> (v6.3.4#6332)

Reply via email to