[ 
https://issues.apache.org/jira/browse/SOLR-4169?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13530080#comment-13530080
 ] 

Mark Miller commented on SOLR-4169:
-----------------------------------

For to add a bit of history:

Most of these fails never would have been fails when most of this code was 
written. Back then, we didnt check for directories not being closed, we didnt 
run tests in parallel, tests ran once a day, we didn't have an awesome test 
framework that failed based on all these things.

So before, most of what randomly fails now, would not even have failed before.

The original coders for a lot of the tests and code failing had moved on to 
other things. It's taken time to slowly harden everything. SolrCloud and its 
jetty based tests introduced a new layer.

Meanwhile, I do a lot of solr development and i find the tests very useful. If 
you need some help, please, ask me and I can lend you hand. 

The tests have made great strides - I and a couple others are reducing the 
failures all the time. Sometimes they flair up due to some change or mix of 
changes - currently that is happening due to a couple issues that are well 
documented and attended to on the list. Real issues that need to be dealt with.

The random fails are not usually test issues - but nor are they always critical 
issues. In fact, most involve shutdown - an area we just have not cared a great 
deal about in Solr in the past. So what if shutdown wasnt perfect. SolrCloud 
changes that a bit.

Anyhow, I put a lot of effort into improving the tests, and I know I've made a 
lot of ground, and will continue to do so. There are a lot of challenges. Like 
I said, Ive been in the Lucene world too, and the nature of it being a library 
without jetty or large integration tests really made it an breeze in 
comparison. Lucene also is lucky enough to have more people that are test 
fanatics at the moment.
                
> The test suite should consistently pass
> ---------------------------------------
>
>                 Key: SOLR-4169
>                 URL: https://issues.apache.org/jira/browse/SOLR-4169
>             Project: Solr
>          Issue Type: Bug
>          Components: Tests
>            Reporter: Robert Muir
>            Priority: Blocker
>             Fix For: 4.1
>
>
> I'm at the end of my rope. 
> Last week I tried to help with this: but its already wednesday and solr tests 
> are 20x worse.
> Users are complaining that they cannot get snapshots and test them, because 
> the tests fail all the time.
> Maven dependencies are wrong, but we never knew, because the tests never 
> succeeded so the validation task hasn't actually run in months.
> I think fixing this situation is a blocker to release: I can't imagine a 
> bigger blocker actually.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

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

Reply via email to