drill-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mehant Baid (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (DRILL-1039) Queries against hbase table with multiple regions return null
Date Sat, 16 Aug 2014 00:42:18 GMT

     [ https://issues.apache.org/jira/browse/DRILL-1039?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Mehant Baid updated DRILL-1039:
-------------------------------

    Attachment: DRILL-1039.patch

Problem was FilterRecordBatch was filtering out all the records in the first batch, so RemovingRecordBatch
was not copying any records in its output container. The output container had an incomplete
schema (only the top level Map vector was copied and not the primitive vectors underneath
it). When filter did receive a batch where not all records where filtered out, and added the
primitive vectors to the Map vector in its output container, Project had no way to detect
the schema change which was causing problems and output NULLS for all the columns in the Map.

Fix is to preserve the schema in RemovingRecordBatch even if no records are copied.  

> Queries against hbase table with multiple regions return null 
> --------------------------------------------------------------
>
>                 Key: DRILL-1039
>                 URL: https://issues.apache.org/jira/browse/DRILL-1039
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Storage - HBase
>         Environment: 
>            Reporter: Krystal
>            Assignee: Mehant Baid
>             Fix For: 0.5.0
>
>         Attachments: DRILL-1039.patch, votertab
>
>
> git.commit.id.abbrev=894037a
> This is a regression. Ran the following query in sqlline:
> 0: jdbc:drill:schema=dfs> select cast(row_key as integer) voter_id, cast(onecf['name']
as varchar(30)) name, cast(twocf['age'] as integer) age, cast(twocf['registration'] as varchar(20))
registration, cast(threecf['contributions'] as decimal(6,2)) contributions, cast(threecf['voterzone']
as integer) voterzone,cast(fourcf['create_date'] as timestamp) create_date from hbase.voter
where row_key = 5;
> +------------+------------+------------+--------------+---------------+------------+-------------+
> |  voter_id  |    name    |    age     | registration | contributions | voterzone  |
create_date |
> +------------+------------+------------+--------------+---------------+------------+-------------+
> | 5          | null       | null       | null         | null          | null       |
1970-01-01 00:00:00.0 |
> +------------+------------+------------+--------------+---------------+------------+-------------+
> The same query returns expected data against an hbase table that resides in single partition.




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

Mime
View raw message