drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From rahul challapalli <challapallira...@gmail.com>
Subject Re: Column ordering is incorrect when ORDER BY is used with LIMIT clause in query over parquet data
Date Tue, 07 Feb 2017 18:34:56 GMT
I don't think a "select *......." query is guaranteed to maintain order.

A similar scenario : https://issues.apache.org/jira/browse/DRILL-1259

On Tue, Feb 7, 2017 at 8:36 AM, Khurram Faraaz <kfaraaz@mapr.com> wrote:

> Can someone please look at this. Is this a bug ?
>
>
> Thanks,
>
> Khurram
>
> ________________________________
> From: Khurram Faraaz <kfaraaz@mapr.com>
> Sent: Monday, February 6, 2017 2:52:25 PM
> To: dev@drill.apache.org
> Subject: Column ordering is incorrect when ORDER BY is used with LIMIT
> clause in query over parquet data
>
> All,
>
>
> This looks incorrect.
>
>
> Query with order by + limit clause, the ordering of the columns returned
> in the query results is NOT the same as the column ordering in the parquet
> file.
>
>
> {noformat}
>
> 0: jdbc:drill:schema=dfs.tmp> SELECT * FROM typeall_l ORDER BY col_int
> limit 1;
> +----------+----------+-------------+------------+----------
> +-----------------+----------------+-----------+------------
> ------------+----------------+-------------+
> | col_bln  | col_chr  |   col_dt    |  col_flt   | col_int  |
> col_intrvl_day  | col_intrvl_yr  |  col_tim  |       col_tmstmp       |
>  col_vrchr1   | col_vrchr2  |
> +----------+----------+-------------+------------+----------
> +-----------------+----------------+-----------+------------
> ------------+----------------+-------------+
> | false    | MI       | 1967-05-01  | 32.901897  | 0        | P12DT20775S
>    | P196M          | 19:50:17  | 2004-10-15 17:49:36.0  | Felecia Gourd  |
> NLBQMg9     |
> +----------+----------+-------------+------------+----------
> +-----------------+----------------+-----------+------------
> ------------+----------------+-------------+
> 1 row selected (0.279 seconds)
>
> {noformat}
>
> Without the ORDER BY clause the columns are returned in correct order,
> same as the ordering in the parquet file.
>
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> SELECT * FROM typeall_l limit 1;
> +----------+----------+----------------+--------------------
> ------------------------------------------------------------
> -------------------------------------+-------------+--------
> ---+------------------------+------------+----------------+-
> ----------------+----------+
> | col_int  | col_chr  |   col_vrchr1   |
>                    col_vrchr2
>         |   col_dt    |  col_tim  |       col_tmstmp       |  col_flt   |
> col_intrvl_yr  | col_intrvl_day  | col_bln  |
> +----------+----------+----------------+--------------------
> ------------------------------------------------------------
> -------------------------------------+-------------+--------
> ---+------------------------+------------+----------------+-
> ----------------+----------+
> | 45436    | WV       | John Mcginity  | Rhbf6VFLJguvH9ejrWNkY1CDO8Qqum
> TZAGjwa9cHfjBnLmNIWvo9YfcGObxbeXwa1NkemW9ULxsq5293wEA2v5FFCduwt03D7ysI3RlH8b4B0XAPKY
> | 2011-11-04  | 18:02:26  | 1988-09-23 16:58:42.0  | 10.193293  | P314M
>       | P26DT27386S     | false    |
> +----------+----------+----------------+--------------------
> ------------------------------------------------------------
> -------------------------------------+-------------+--------
> ---+------------------------+------------+----------------+-
> ----------------+----------+
> 1 row selected (0.22 seconds)
>
>
> {noformat}
>
>
> Thanks,
>
> Khurram
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message