[
https://issues.apache.org/jira/browse/DRILL-3491?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14627808#comment-14627808
]
Carrot Hu commented on DRILL-3491:
----------------------------------
I have changed from cdh5.4.2 to hbase-1.0.1.1 and hadoop-2.7.1 with same
configurations.
The issue does not occur anymore.
My suspect this is due to some incompatibility with cdh version of hadoop and
hbase.
> SELECT COUNT(*) FROM HBASE Returns Incorrect Row Count
> ------------------------------------------------------
>
> Key: DRILL-3491
> URL: https://issues.apache.org/jira/browse/DRILL-3491
> Project: Apache Drill
> Issue Type: Bug
> Components: Storage - HBase
> Affects Versions: 1.0.0, 1.1.0
> Environment: CentOS6.5
> jdk1.8.0_45
> hadoop-2.6.0-cdh5.4.2
> hbase-1.0.0-cdh5.4.2
> IntelliJ14.1.4
> Maven3.0.5
> Reporter: Carrot Hu
> Assignee: Aditya Kishore
> Labels: hbase, sql
>
> Create a table 'test' in Hbase with 1 column family, 7 columns.
> Inserting 100,000 rows into 'test' using Java API, each column with same
> value = "value".
> SELECT COUNT(<all>) FROM hbase.test
> returns an incorrect row count.
> Verified using count 'test' in hbase shell, the row count is correct.
> SELECT COUNT(row_key) is correct,
> SELECT COUNT(<Any subset of the columns>) is also correct.
> Clear the table, and changed to inserting 1000 rows, keep the number of
> columns, Drill returns the right count. But when increasing the number of
> columns to 30. SELLECT COUNT(<all>) returns an incorrect row count (only 673).
> Use count 'test' and scan 'test' in hbase to check the result, nothing usual
> were noticed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
