[
https://issues.apache.org/jira/browse/DRILL-5040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15853926#comment-15853926
]
Khurram Faraaz commented on DRILL-5040:
---------------------------------------
Issue a CTTAS over (~26M JSON records) and issue Ctrl-C on sqlline while the
CTTAS is under execution. And then do a select on the temporary table physical
file on disk, we see the below RuntimeException.
{noformat}
0: jdbc:drill:schema=dfs.tmp> CREATE TEMPORARY TABLE temp_tbl_1 AS select *
from `twoKeyJsn.json`;
No rows affected (1.607 seconds)
0: jdbc:drill:schema=dfs.tmp> select count(*) from temp_tbl_1;
Error: SYSTEM ERROR: RuntimeException:
maprfs:///tmp/bfe2dad0-921a-4f06-9799-494ab8a7246d/851a124c-80a1-45e3-9496-d2562007911e/0_0_0.parquet
is not a Parquet file (too small)
[Error Id: ea0d1086-13d1-416b-a1a7-165050ac209f on centos-01.qa.lab:31010]
(state=,code=0)
{noformat}
> Interrupted CTAS should not succeed & should not create physical file on disk
> -----------------------------------------------------------------------------
>
> Key: DRILL-5040
> URL: https://issues.apache.org/jira/browse/DRILL-5040
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
> Affects Versions: 1.9.0
> Reporter: Khurram Faraaz
> Assignee: Arina Ielchiieva
> Fix For: 1.10.0
>
>
> We should not allow CTAS to succeed (i.e create physical file on disk ) in
> the case where it was interrupted. (vis Ctrl-C)
> Drill 1.9.0
> git commit ID : db30854
> Consider the below CTAS that was interrupted using Ctrl-C
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> create table temp_t1 as select * from
> `twoKeyJsn.json`;
> [ issue Ctrl-C while the above CTAS is running ]
> No rows affected (7.694 seconds)
> {noformat}
> I verified that physical file was created on disk, even though the above CTAS
> was Canceled
> {noformat}
> [root@centos-01 ~]# hadoop fs -ls /tmp/temp_t1*
> -rwxr-xr-x 3 root root 36713198 2016-11-14 10:51
> /tmp/temp_t1/0_0_0.parquet
> {noformat}
> We are able to do a select on the CTAS table (above) that was Canceled.
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> select count(*) from temp_t1;
> +----------+
> | EXPR$0 |
> +----------+
> | 3747840 |
> +----------+
> 1 row selected (0.183 seconds)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
