Mailing-List: contact user-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hive.apache.org
Received-SPF: pass (nike.apache.org: domain of krishnanjrao@gmail.com
 designates 209.85.223.175 as permitted sender)
MIME-Version: 1.0
From: Krishna Rao <krishnanjrao@gmail.com>
Date: Thu, 27 Feb 2014 12:44:28 +0000
Message-ID: 
 <CAPEqewJzJADfXh4zxAjZHKQCcfFNz=gNaFcdd52Q1Ka6=L88ww@mail.gmail.com>
Subject: Hive query parser bug resulting in "FAILED: NullPointerException
 null"
To: user@hive.apache.org
Content-Type: multipart/alternative; boundary=047d7bd752a8d17f6104f362af1c

--047d7bd752a8d17f6104f362af1c
Content-Type: text/plain; charset=ISO-8859-1

Hi all,

we've experienced a bug which seems to be caused by having a query
constraint involving partitioned columns. The following query results in
"FAILED: NullPointerException null" being returned nearly instantly:

EXPLAIN SELECT
  col1
FROM
  tbl1
WHERE
(part_col1 = 2014 AND part_col2 >= 2)
OR part_col1 < 2014;

The exception doesn't happen if any of the conditions are removed. The
table is defined like the following:

CREATE TABLE tbl1 (
  col1                            STRING,
  ...
  col12                           STRING
)
PARTITIONED BY (part_col1 INT, part_col2 TINYINT, part_col3 TINYINT)
STORED AS SEQUENCEFILE;


Unfortunately I cannot construct a test case to replicate this. Seen as
though it appears to be a query parser bug, I thought the following would
replicate it:

CREATE TABLE tbl2 LIKE tbl1;
EXPLAIN SELECT
  col1
FROM
  tbl2
WHERE
(part_col1 = 2014 AND part_col2 >= 2)
OR part_col1 < 2014;

But it does not. Could it somehow be data specific? Does the query parser
use partition information?

Are there any logs I could see to investigate this further? Or is this a
known bug?

We're using hive 0.10.0-cdh4.4.0.


Cheers,

Krishna

--047d7bd752a8d17f6104f362af1c
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi all,<div><br></div><div>we&#39;ve experienced a bug whi=
ch seems to be caused by having a query constraint involving partitioned co=
lumns. The following query results in &quot;FAILED: NullPointerException nu=
ll&quot; being returned nearly instantly:</div>


<div><br></div><div><div>EXPLAIN SELECT</div><div>=A0 col1</div><div>FROM</=
div><div>=A0 tbl1</div><div>WHERE</div><div>(part_col1 =3D 2014 AND part_co=
l2 &gt;=3D 2)</div><div>OR=A0part_col1=A0&lt; 2014;</div></div><div><br></d=
iv><div>


The exception doesn&#39;t happen if any of the conditions are removed. The =
table is defined like the following:</div><div><br></div><div><div>CREATE T=
ABLE=A0tbl1=A0(</div><div>=A0=A0col1=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
 =A0 =A0 =A0 =A0 STRING,</div><div>

=A0 ...</div><div>=A0=A0col12=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 =A0STRING</div><div>)</div><div>PARTITIONED BY (part_col1=A0INT,=A0=
part_col2=A0TINYINT,=A0part_col3=A0TINYINT)</div><div>STORED AS SEQUENCEFIL=
E;</div></div><div><br></div><div><br></div>

<div>Unfortunately I cannot construct a test case to replicate this. Seen a=
s though it appears to be a query parser bug, I thought the following would=
 replicate it:<br>
</div><div><br></div><div>CREATE TABLE tbl2 LIKE tbl1;</div><div><div>EXPLA=
IN SELECT</div><div>=A0 col1</div><div>FROM</div><div>=A0 tbl2</div><div>WH=
ERE</div><div>(part_col1 =3D 2014 AND part_col2 &gt;=3D 2)</div><div>OR=A0p=
art_col1=A0&lt; 2014;</div>


</div><div><br></div><div>But it does not. Could it somehow be data specifi=
c? Does the query parser use partition information?</div><div><br></div><di=
v>Are there any logs I could see to investigate this further? Or is this a =
known bug?</div>

<div><br></div><div>We&#39;re using hive 0.10.0-cdh4.4.0.</div><div><br></d=
iv><div><br></div><div>Cheers,</div><div><br></div><div>Krishna</div></div>

--047d7bd752a8d17f6104f362af1c--