[jira] [Commented] (DRILL-4173) Query did not return all documents if collection using a hashed shard key

["B Anil Kumar (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-4173?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15081057#comment-15081057 ] 

B Anil Kumar commented on DRILL-4173:
-------------------------------------

[~ITDonald] Thanks for reporting this issue.

We are also able to reproduce issue, the root cause is, in sharded env, we construct a mapping b/w chunk and Drillbit and assigned Drillbit will read the chunk based on minKey and maxKey range.

In the case of hashed sharding the minKey and maxKey is not matching with actual shardKey values. That's why there is a data loss through Drill.

We are looking into this issue.

> Query did not return all documents if collection using a hashed shard key
> -------------------------------------------------------------------------
>
>                 Key: DRILL-4173
>                 URL: https://issues.apache.org/jira/browse/DRILL-4173
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Storage - MongoDB
>    Affects Versions: 1.3.0
>         Environment: Windows 2012
>            Reporter: Yuqing Tang
>            Assignee: Jason Altekruse
>
> MongoDB 3.0.6
> If a collection used a hashed shard key ({ "shardkey": "hashed" }), queries like "select * from ..." may not return all documents which should be returned from the collection.
> Test Case:
> Create 3 mongos, 3 config, 3 replicaset each with 3 mongod.
> Create one collection with hashed shard key.
> Insert 6 documents to this collection with shard key value 1,2,3,4,5,6
> Do a query select * from ....
> Only 2,3,4 will be returned.



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