On Jan 27, 2012, at 4:50 AM, Simon Willnauer wrote:

> mark | yonik, can you open issues for all these failures.
> 
> I see lots of failures for tests like:
> 
> ChaosMonkeySafeLeaderTest
> RecoveryZkTest
> TermVectorComponentTest
> LeaderElectionTest
> FullSolrCloudDistribCmdsTest
> SpatialFilterTest

It appears to be one issue that is falling into other tests rather than an 
issue for each of these tests.

> 
> 
> The last two days are crazy and I don't see any issues ie. reactions
> to these failures. Please open issues for these test.

I've been in reaction mode pretty much every moment I've been awake - take a 
look at my commit history. I haven't sent additional email because I assumed a 
concerned committer might notice my action trail.

> I am consider
> disabling the tests since the apparently rarely pass. Almost every
> build fails see the jenkins stats:
> 
> this is the first build with solr cloud:
> https://builds.apache.org/job/Lucene-Solr-tests-only-trunk/12258/
> 
> since this commit not a single run passed:
> 
> https://builds.apache.org/job/Lucene-Solr-tests-only-trunk/
> 
> maybe it was too early to merge this in? Should we back it out and
> create a branch jenkins build until this stabelizes. I mean trunk you
> should be able to go crazy with features but this seems ridiculous. I
> really fear that this stuff is moving us 6 month back with respect to
> Lucene / Solr 4.0 :(

These tests pass over an over on a ton of machines - there is some issue with 
hudson that we have not yet been able to suss out but that we are working on.

Looking at the lucene tests lately I see a lot of red too. Perhaps stuff is too 
unstable there? They are passing like 50% at best on Lucene-Trunk most of the 
time it looks. Or perhaps good random tests can pass a lot in some envs but not 
in others without some investigation and hardening.

Having been looking into this issue non stop for the past many many hours, I 
can say I don't think its a stability issue with the cloud code at all - its 
very likely a test/env issue. And we are working non stop to figure out what it 
is. The slow feedback loop with hudson makes it so I can only work so fast - 
like I said, I don't see this failure on multiple other machines.

I only ask for a little patience - every failure is being looked at, and every 
attempt is being made to get hudson happy. If we don't make progress we will 
obviously take another route.



> 
> simon
> On Fri, Jan 27, 2012 at 10:08 AM, Apache Jenkins Server
> <[email protected]> wrote:
>> Build: https://builds.apache.org/job/Solr-trunk/1746/
>> 
>> 3 tests failed.
>> FAILED:  
>> junit.framework.TestSuite.org.apache.solr.cloud.ChaosMonkeySafeLeaderTest
>> 
>> Error Message:
>> ERROR: SolrIndexSearcher opens=166 closes=159
>> 
>> Stack Trace:
>> junit.framework.AssertionFailedError: ERROR: SolrIndexSearcher opens=166 
>> closes=159
>>        at 
>> org.apache.solr.SolrTestCaseJ4.endTrackingSearchers(SolrTestCaseJ4.java:151)
>>        at 
>> org.apache.solr.SolrTestCaseJ4.afterClassSolrTestCase(SolrTestCaseJ4.java:76)
>> 
>> 
>> FAILED:  junit.framework.TestSuite.org.apache.solr.cloud.RecoveryZkTest
>> 
>> Error Message:
>> ERROR: SolrIndexSearcher opens=13 closes=12
>> 
>> Stack Trace:
>> junit.framework.AssertionFailedError: ERROR: SolrIndexSearcher opens=13 
>> closes=12
>>        at 
>> org.apache.solr.SolrTestCaseJ4.endTrackingSearchers(SolrTestCaseJ4.java:151)
>>        at 
>> org.apache.solr.SolrTestCaseJ4.afterClassSolrTestCase(SolrTestCaseJ4.java:76)
>> 
>> 
>> FAILED:  
>> junit.framework.TestSuite.org.apache.solr.handler.component.TermVectorComponentTest
>> 
>> Error Message:
>> ERROR: SolrIndexSearcher opens=2 closes=3
>> 
>> Stack Trace:
>> junit.framework.AssertionFailedError: ERROR: SolrIndexSearcher opens=2 
>> closes=3
>>        at 
>> org.apache.solr.SolrTestCaseJ4.endTrackingSearchers(SolrTestCaseJ4.java:151)
>>        at 
>> org.apache.solr.SolrTestCaseJ4.afterClassSolrTestCase(SolrTestCaseJ4.java:76)
>> 
>> 
>> 
>> 
>> Build Log (for compile errors):
>> [...truncated 49188 lines...]
>> 
>> 
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [email protected]
>> For additional commands, e-mail: [email protected]
>> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
> 

- Mark Miller
lucidimagination.com












---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to