lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Shalin Shekhar Mangar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SOLR-6290) CollectionsAPIAsyncDistributedZkTest failure
Date Mon, 28 Jul 2014 19:43:38 GMT

    [ https://issues.apache.org/jira/browse/SOLR-6290?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14076678#comment-14076678
] 

Shalin Shekhar Mangar commented on SOLR-6290:
---------------------------------------------

bq. I think these are insane timeouts though, Shalin. I mean – the tests never write tons
of data; even on heavily loaded systems there should be some detectable progress (or a sensible,
short-lived timeout after which you consider this a failure).

Yeah, perhaps I went overboard with the timeout. It can be set to a minute and that should
avoid this particular failure. Some tasks such as split or migrate require more time e.g.
I saw a simple shard split on an empty index not complete within 90 seconds and there was
nothing in the logs which showed any problems.

bq. The chances that 2 minutes or 1 minutes or 30 seconds is better seems pretty likely.

And I agreed with you and said that (twice): "That being said, 5 minutes may be too much and
we can make it something like a minute or so." and then again "we can tune it lower if it
makes people happier"

bq. Why not make it an hour with your philosophy?

Because I am not insane.

See, all the while, you've been talking as if:
# I am actually increasing the time taken by our suite by 30 minutes
# That increasing a timeout to 5 minutes here will actually slow the test for everyone by
5 minutes

Both of the above are false. The problem I am trying to solve is that our test suite fails
needlessly and often enough that many of us ignore failures on jenkins just because we can't
make a distinction quickly enough. I think there are real benefits to solving that problem
right now than an imagined problem which might manifest when you are debugging this particular
test (when you can always change the timeout if you really think it is a problem).

My quote " it won't ever take that long but we should wait for as long as possible."  and
"I'd rather have our tests take 30 minutes more" have been twisted to mean something that
I did not intend.

> CollectionsAPIAsyncDistributedZkTest failure
> --------------------------------------------
>
>                 Key: SOLR-6290
>                 URL: https://issues.apache.org/jira/browse/SOLR-6290
>             Project: Solr
>          Issue Type: Bug
>          Components: SolrCloud, Tests
>            Reporter: Shalin Shekhar Mangar
>            Assignee: Shalin Shekhar Mangar
>             Fix For: 5.0, 4.10
>
>         Attachments: SOLR-6290.patch
>
>
> Found this failure on my local jenkins instance:
> {code}
> Error Message
> CreateCollection task did not complete! expected:<[completed]> but was:<[running]>
> Stacktrace
> org.junit.ComparisonFailure: CreateCollection task did not complete! expected:<[completed]>
but was:<[running]>
> 	at __randomizedtesting.SeedInfo.seed([D4EC4F9D698903B5:550AC1851ED66389]:0)
> 	at org.junit.Assert.assertEquals(Assert.java:125)
> 	at org.apache.solr.cloud.CollectionsAPIAsyncDistributedZkTest.testSolrJAPICalls(CollectionsAPIAsyncDistributedZkTest.java:75)
> 	at org.apache.solr.cloud.CollectionsAPIAsyncDistributedZkTest.doTest(CollectionsAPIAsyncDistributedZkTest.java:61)
> 	at org.apache.solr.BaseDistributedSearchTestCase.testDistribSearch(BaseDistributedSearchTestCase.java:865)
> 	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> {code}



--
This message was sent by Atlassian JIRA
(v6.2#6252)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message