[
https://issues.apache.org/jira/browse/IMPALA-7583?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16630757#comment-16630757
]
Tim Armstrong commented on IMPALA-7583:
---------------------------------------
The plan and the data are super-simple.
{noformat}
[localhost:21000] functional> explain select f,lead(b,1,null) over (order by f)
from (select * from nulltable union all select * from nulltable) x;
Query: explain select f,lead(b,1,null) over (order by f) from (select * from
nulltable union all select * from nulltable) x
+------------------------------------------------------------------------------------+
| Explain String
|
+------------------------------------------------------------------------------------+
| Max Per-Host Resource Reservation: Memory=16.01MB Threads=3
|
| Per-Host Resource Estimates: Memory=48MB
|
| WARNING: The following tables are missing relevant table and/or column
statistics. |
| functional.nulltable
|
|
|
| PLAN-ROOT SINK
|
| |
|
| 04:ANALYTIC
|
| | functions: lead(b, 1, NULL)
|
| | order by: f ASC
|
| | window: ROWS BETWEEN UNBOUNDED PRECEDING AND 1 FOLLOWING
|
| |
|
| 05:MERGING-EXCHANGE [UNPARTITIONED]
|
| | order by: f ASC
|
| |
|
| 03:SORT
|
| | order by: f ASC
|
| |
|
| 00:UNION
|
| | pass-through-operands: all
|
| |
|
| |--02:SCAN HDFS [functional.nulltable]
|
| | partitions=1/1 files=1 size=18B
|
| |
|
| 01:SCAN HDFS [functional.nulltable]
|
| partitions=1/1 files=1 size=18B
|
+------------------------------------------------------------------------------------+
Fetched 26 row(s) in 0.01s
[localhost:21000] functional> select * from nulltable;
Query: select * from nulltable
Query submitted at: 2018-09-27 10:15:48 (Coordinator:
http://tarmstrong-box:25000)
Query progress can be monitored at:
http://tarmstrong-box:25000/query_plan?query_id=5b4e8176c27618ba:306bdf100000000
+---+---+------+------+------+----+---+
| a | b | c | d | e | f | g |
+---+---+------+------+------+----+---+
| a | | NULL | NULL | NULL | ab | |
+---+---+------+------+------+----+---+
{noformat}
Theories that come to mind are:
* There's a memory management bug in the union node or HDFS scan, e.g. related
to the union passthrough
* a bit got flipped or there was some random hardware-related bug. This seems
unlikely, but who knows
> Failure in test_analytic_fns
> ----------------------------
>
> Key: IMPALA-7583
> URL: https://issues.apache.org/jira/browse/IMPALA-7583
> Project: IMPALA
> Issue Type: Bug
> Affects Versions: Impala 3.1.0
> Reporter: Thomas Tauber-Marshall
> Assignee: Thomas Tauber-Marshall
> Priority: Critical
> Labels: broken-build, flaky
>
> Seen in a build:
> {noformat}
> 08:09:43 TestQueries.test_analytic_fns[exec_option:
> {'disable_codegen_rows_threshold': 0, 'disable_codegen': True,
> 'abort_on_error': 1, 'debug_action': None, 'exec_single_node_rows_threshold':
> '100', 'batch_size': 0, 'num_nodes': 0} | table_format: text/lzo/block]
> 08:09:43 [gw1] linux2 -- Python 2.7.5
> /data/jenkins/workspace/impala-cdh6.x-exhaustive-release/repos/Impala/bin/../infra/python/env/bin/python
> 08:09:43 query_test/test_queries.py:54: in test_analytic_fns
> 08:09:43 self.run_test_case('QueryTest/analytic-fns', vector)
> 08:09:43 common/impala_test_suite.py:437: in run_test_case
> 08:09:43 self.__verify_results_and_errors(vector, test_section, result,
> use_db)
> 08:09:43 common/impala_test_suite.py:310: in __verify_results_and_errors
> 08:09:43 replace_filenames_with_placeholder)
> 08:09:43 common/test_result_verifier.py:433: in verify_raw_results
> 08:09:43 VERIFIER_MAP[verifier](expected, actual)
> 08:09:43 common/test_result_verifier.py:260: in verify_query_result_is_equal
> 08:09:43 assert expected_results == actual_results
> 08:09:43 E assert Comparing QueryTestResults (expected vs actual):
> 08:09:43 E 'a\x00b','' != '9k\x00',''
> 08:09:43 E 'a\x00b','NULL' == 'a\x00b','NULL'
> 08:09:43 ----------------------------- Captured stderr call
> -----------------------------
> ...
> 08:09:43 -- 2018-09-12 06:49:31,694 INFO MainThread: Started query
> d14c894f7e2a20fc:c797406b00000000
> 08:09:43 -- executing against localhost:21000
> 08:09:43 select count(*) from (
> 08:09:43 select
> 08:09:43 from_unixtime(lead(bigint_col, 1) over (order by id),
> 'yyyyMMddHH:mm:ss') as a,
> 08:09:43 lead(from_unixtime(bigint_col, 'yyyyMMddHH:mm:ss'), 1) over (order
> by id) AS b
> 08:09:43 from functional.alltypes) x
> 08:09:43 where x.a = x.b;
> 08:09:43
> 08:09:43 -- 2018-09-12 06:49:31,809 INFO MainThread: Started query
> d948840775a29d70:5fce58f400000000
> 08:09:43 -- executing against localhost:21000
> 08:09:43 select count(*) from (
> 08:09:43 select
> 08:09:43 from_unixtime(lag(bigint_col, 1) over (order by id),
> 'yyyyMMddHH:mm:ss') as a,
> 08:09:43 lag(from_unixtime(bigint_col, 'yyyyMMddHH:mm:ss'), 1) over (order
> by id) AS b
> 08:09:43 from functional.alltypes) x
> 08:09:43 where x.a = x.b;
> 08:09:43
> 08:09:43 -- 2018-09-12 06:49:31,930 INFO MainThread: Started query
> 7468c485800eb48:8c6f135600000000
> 08:09:43 -- executing against localhost:21000
> 08:09:43 select f,lead(b,1,null) over (order by f)
> 08:09:43 from (select * from nulltable union all select * from nulltable) x;
> 08:09:43
> 08:09:43 -- 2018-09-12 06:49:32,178 INFO MainThread: Started query
> 3f4a69557cbfc76b:bf6a1a5800000000
> 08:09:43 -- 2018-09-12 06:49:32,260 ERROR MainThread: Comparing
> QueryTestResults (expected vs actual):
> 08:09:43 'a\x00b','' != '9k\x00',''
> 08:09:43 'a\x00b','NULL' == 'a\x00b','NULL'
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
