Ah, that makes sense.  I had forgotten that there was a JIRA for this:

https://issues.apache.org/jira/browse/SPARK-4021  

On October 21, 2014 at 10:08:58 AM, Patrick Wendell (pwend...@gmail.com) wrote:

Josh - the errors that broke our build indicated that JDK5 was being  
used. Somehow the upgrade caused our build to use a much older Java  
version. See the JIRA for more details.  

On Tue, Oct 21, 2014 at 10:05 AM, Josh Rosen <rosenvi...@gmail.com> wrote:  
> I find it concerning that there's a JDK version that breaks out build, since  
> we're supposed to support Java 7. Is 7u71 an upgrade or downgrade from the  
> JDK that we used before? Is there an easy way to fix our build so that it  
> compiles with 7u71's stricter settings?  
>  
> I'm not sure why the "New" PRB is failing here. It was originally created  
> as a clone of the main pull request builder job. I checked the configuration  
> history and confirmed that there aren't any settings that we've forgotten to  
> copy over (e.g. their configurations haven't diverged), so I'm not sure  
> what's causing this.  
>  
> - Josh  
>  
> On October 21, 2014 at 6:35:39 AM, Nan Zhu (zhunanmcg...@gmail.com) wrote:  
>  
> weird.....two buildings (one triggered by New, one triggered by Old) were  
> executed in the same node, amp-jenkins-slave-01, one compiles, one not...  
>  
> Best,  
>  
> --  
> Nan Zhu  
>  
>  
> On Tuesday, October 21, 2014 at 9:39 AM, Nan Zhu wrote:  
>  
>> seems that all PRs built by NewSparkPRBuilder suffers from 7u71, while  
>> SparkPRBuilder is working fine  
>>  
>> Best,  
>>  
>> --  
>> Nan Zhu  
>>  
>>  
>> On Tuesday, October 21, 2014 at 9:22 AM, Cheng Lian wrote:  
>>  
>> > It's a new pull request builder written by Josh, integrated into our  
>> > state-of-the-art PR dashboard :)  
>> >  
>> > On 10/21/14 9:33 PM, Nan Zhu wrote:  
>> > > just curious...what is this "NewSparkPullRequestBuilder"?  
>> > >  
>> > > Best,  
>> > >  
>> > > --  
>> > > Nan Zhu  
>> > >  
>> > >  
>> > > On Tuesday, October 21, 2014 at 8:30 AM, Cheng Lian wrote:  
>> > >  
>> > > >  
>> > > > Hm, seems that 7u71 comes back again. Observed similar Kinesis  
>> > > > compilation error just now:  
>> > > > https://amplab.cs.berkeley.edu/jenkins/job/NewSparkPullRequestBuilder/410/consoleFull
>> > > >   
>> > > >  
>> > > >  
>> > > > Checked Jenkins slave nodes, saw /usr/java/latest points to  
>> > > > jdk1.7.0_71. However, /usr/bin/javac -version says:  
>> > > >  
>> > > > >  
>> > > > > Eclipse Java Compiler 0.894_R34x, 3.4.2 release, Copyright IBM  
>> > > > > Corp 2000, 2008. All rights reserved.  
>> > > > >  
>> > > >  
>> > > >  
>> > > > Which JDK is actually used by Jenkins?  
>> > > >  
>> > > >  
>> > > > Cheng  
>> > > >  
>> > > >  
>> > > > On 10/21/14 8:28 AM, shane knapp wrote:  
>> > > >  
>> > > > > ok, so earlier today i installed a 2nd JDK within jenkins (7u71),  
>> > > > > which fixed the SparkR build but apparently made Spark itself quite 
>> > > > > unhappy.  
>> > > > > i removed that JDK, triggered a build (  
>> > > > > https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/21943/console),
>> > > > >   
>> > > > > and it compiled kinesis w/o dying a fiery death. apparently 7u71 is 
>> > > > > stricter  
>> > > > > when compiling. sad times. sorry about that! shane On Mon, Oct 20, 
>> > > > > 2014 at  
>> > > > > 5:16 PM, Patrick Wendell <pwend...@gmail.com> 
>> > > > > (mailto:pwend...@gmail.com)  
>> > > > > wrote:  
>> > > > > > The failure is in the Kinesis compoent, can you reproduce this  
>> > > > > > if you build with -Pkinesis-asl? - Patrick On Mon, Oct 20, 2014 at 
>> > > > > > 5:08 PM,  
>> > > > > > shane knapp <skn...@berkeley.edu> (mailto:skn...@berkeley.edu) 
>> > > > > > wrote:  
>> > > > > > > hmm, strange. i'll take a look. On Mon, Oct 20, 2014 at 5:11  
>> > > > > > > PM, Nan Zhu <zhunanmcg...@gmail.com> 
>> > > > > > > (mailto:zhunanmcg...@gmail.com) wrote:  
>> > > > > > > > yes, I can compile locally, too but it seems that Jenkins is  
>> > > > > > > > not happy now...  
>> > > > > > > > https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/
>> > > > > > > >  All  
>> > > > > > > > failed to compile Best, -- Nan Zhu On Monday, October 20, 2014 
>> > > > > > > > at 7:56 PM,  
>> > > > > > > > Ted Yu wrote:  
>> > > > > > > > > I performed build on latest master branch but didn't get  
>> > > > > > > > > compilation  
>> > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > > error.  
>> > > > > > > > > FYI On Mon, Oct 20, 2014 at 3:51 PM, Nan Zhu  
>> > > > > > > > > <zhunanmcg...@gmail.com (mailto:zhunanmcg...@gmail.com)  
>> > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > > (mailto:zhunanmcg...@gmail.com)> wrote:  
>> > > > > > > > > > Hi, I just submitted a patch  
>> > > > > > > > > >  
>> > > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > > https://github.com/apache/spark/pull/2864/files  
>> > > > > > > > > > with one line change but the Jenkins told me it's failed  
>> > > > > > > > > > to compile on the unrelated  
>> > > > > > > > > >  
>> > > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > >  
>> > > > > > >  
>> > > > > > >  
>> > > > > >  
>> > > > > > files?  
>> > > > > > > > > >  
>> > > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > >  
>> > > > > >  
>> > > > > >  
>> > > > > > https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/21935/console
>> > > > > >   
>> > > > > > > > > > Best, Nan  
>> > > > > > > > > >  
>> > > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > > >  
>> > > > > > >  
>> > > > > >  
>> > > > > >  
>> > > > > >  
>> > > > >  
>> > > > >  
>> > > > >  
>> > > >  
>> > > >  
>> > > >  
>> > > >  
>> > > >  
>> > > >  
>> > > >  
>> > >  
>> > >  
>> >  
>>  
>  

Reply via email to