Re: Solr nightly build failure

[Yonik Seeley]

On 2/2/07, Mike Klaas <[EMAIL PROTECTED]> wrote:

On 2/2/07, Yonik Seeley <[EMAIL PROTECTED]> wrote:

> > Does anyone know how to access the build logs?  I don't know from
> > apache infra (except people./minotaur.)
>
> It's on the lucene zone with restricted access.
> It might be possible to set something up to copy some stuff to
> minotaur on a failure though.
>
> I think this is the relevant portion:
>
>   <testcase classname="org.apache.solr.update.AutoCommitTest"
> name="testMaxDocs" time="16.867">
>     <failure message="expected:&lt;0&gt; but was:&lt;1&gt;"
> 
type="junit.framework.AssertionFailedError">junit.framework.AssertionFailedError:
> expected:&lt;0&gt;
>  but was:&lt;1&gt;
>         at 
org.apache.solr.update.AutoCommitTest.testMaxDocs(AutoCommitTest.java:89)
> </failure>
>   </testcase>
>   <testcase classname="org.apache.solr.update.AutoCommitTest" 
name="testMaxTime"
>  time="9.633"></testcase>
>
> Let me know if you need the whole thing or any other files.

OK, this time:
 BUILD SUCCESSFUL
 Total time: 2 minutes 44 seconds

Well, the cause is pretty much as I expected.  The autocommit test
expects to be able to add fifteen small documents in less than a
seconds (ie. without triggering autocommit).  Not sure how to make
this robust against severe machine load save increasing the time
drastically.

Does this happen due to many test suites running at a time?  Is there
any way to modify the scheduling?  It would be nice to not add too
much time to the test suite run--the longer it is, the less
religiously devs will run it.

I don't think there is a lucene scheduling problem, but a zone is a
virtual server, so who knows what else is going on at the time.

http://www.apache.org/dev/solaris-zones.html

lucene runs at 00:03, hadoop at 00:11, nutch at 00:17, solr at 31 after.

But looking at the dates on the log files,
lucene ended at 00:30,
hadoop ended at 00:44,
nutch ended at 00:41

Ouch!  Something is going on...

-Yonik