[
https://issues.apache.org/jira/browse/DRILL-3865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14937450#comment-14937450
]
Victoria Markman commented on DRILL-3865:
-----------------------------------------
Khurram,
Important detail that is missing here is that if you cast CONVERT_FROM
expression to time, you will get correct result:
SELECT CONVERT_FROM(BYTE_SUBSTR(row_key,1,8),'time_epoch_be') AS RK from dt_Tbl
T WHERE {color:red}CAST({color}
CONVERT_FROM(BYTE_SUBSTR(row_key,1,8),'time_epoch_be') {color:red}as TIME
){color} >= TIME '00:00:00' ...
Vicky.
> Difference in results - query parquet table vs query HBase table
> ----------------------------------------------------------------
>
> Key: DRILL-3865
> URL: https://issues.apache.org/jira/browse/DRILL-3865
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
> Affects Versions: 1.2.0
> Environment: 4 node cluster CentOS
> Reporter: Khurram Faraaz
> Assignee: Aditya Kishore
>
> Querying HBase from Drill returns no results. Query from Drill over same data
> in parquet format from Drill returns expected results.
> {code}
> 0: jdbc:drill:schema=dfs.tmp> describe dt_Tbl;
> +--------------+------------+--------------+
> | COLUMN_NAME | DATA_TYPE | IS_NULLABLE |
> +--------------+------------+--------------+
> | row_key | ANY | NO |
> | colfam1 | MAP | NO |
> +--------------+------------+--------------+
> 2 rows selected (1.494 seconds)
> 0: jdbc:drill:schema=dfs.tmp> select RK from dfs.tmp.t20 WHERE RK >= TIME
> '00:00:00' AND RK <= TIME '04:14:46.094';
> +---------------+
> | RK |
> +---------------+
> | 00:00:00 |
> | 04:14:46.094 |
> | 04:14:46.094 |
> | 01:50:46.094 |
> | 04:14:46.094 |
> +---------------+
> 5 rows selected (0.242 seconds)
> {code}
> {code}
> root@centos-01 ~]# parquet-tools/parquet-schema 0_0_0.parquet
> message root {
> required int32 RK (TIME_MILLIS);
> }
> 0: jdbc:drill:schema=dfs.tmp> SELECT
> CONVERT_FROM(BYTE_SUBSTR(row_key,1,8),'time_epoch_be') AS RK from dt_Tbl T
> WHERE CONVERT_FROM(BYTE_SUBSTR(row_key,1,8),'time_epoch_be') >= TIME
> '00:00:00' AND CONVERT_FROM(BYTE_SUBSTR(row_key,1,8),'time_epoch_be') <= TIME
> '04:14:46.094';
> +--+
> | |
> +--+
> +--+
> No rows selected (0.902 seconds)
> explain plan for above query
> 00-01 Project(RK=[CONVERT_FROMTIME_EPOCH_BE(BYTE_SUBSTR($0, 1, 8))])
> 00-02 Scan(groupscan=[HBaseGroupScan [HBaseScanSpec=HBaseScanSpec
> [tableName=dt_Tbl, startRow=\x00\x00\x00\x00\x00\x00\x00\x00,
> stopRow=\x00\x00\x00\x00\x00\xE9?O, filter=null], columns=[`row_key`]]])
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
