[jira] [Updated] (LUCENE-7025) Already open IndexSearcher sees different DVs as commits are happening

["Ishan Chattopadhyaya (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/LUCENE-7025?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ishan Chattopadhyaya updated LUCENE-7025:
-----------------------------------------
    Attachment: TestIndexSearcherStability.java

Test containing a Lucene only test demonstrating the issue I am running into.

> Already open IndexSearcher sees different DVs as commits are happening
> ----------------------------------------------------------------------
>
>                 Key: LUCENE-7025
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7025
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Ishan Chattopadhyaya
>         Attachments: TestIndexSearcherStability.java
>
>
> I'm hitting an issue in SOLR-5944 whereby I am seeing the dvGen of a reader underlying an open searcher is changing when parallel commits are happening. 
> This happens when the reader is opened from a writer, but doesn't happen when the reader is opened directly from a directory.
> Also, *I think* I am, in Solr, seeing differences in DVs when parallel threads make DV updates and do commits. In the attached test, I couldn't reproduce this scenario, though.
> Questions:
> 1) Is it a bug, or by design, that the segment readers of an open indexsearcher has their dvGens changing? (as according to the test attached).
> 2) Is it possible, as I think it is happening in Solr's 5944, that the DV values themselves could change as a result of this parallel commit happening?
> Apologies, if I should have asked this at the user mailing list, instead of opening an issue here.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

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