[ 
https://issues.apache.org/jira/browse/IMPALA-7963?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16721649#comment-16721649
 ] 

Joe McDonnell commented on IMPALA-7963:
---------------------------------------

Added a DCHECK in stopwatch.h:
{code:java}
uint64_t RunningTime() const {
  uint64_t end = Now();
  if (time_ceiling_ > 0) {
    if (time_ceiling_ < start_) return 0;
    if (time_ceiling_ < end) end = time_ceiling_;
  }
  DCHECK_GE(end, start_); <-------
  return end - start_;
}
{code}
This DCHECK fires, indicating that time goes backwards on centos6. We are using 
CLOCK_MONOTONIC, and apparently there are some bugs where it can go backwards. 
So far, we have only seen this on centos6 (specifically 6.4).

> test_empty_build_joins failed with hdfs timeout
> -----------------------------------------------
>
>                 Key: IMPALA-7963
>                 URL: https://issues.apache.org/jira/browse/IMPALA-7963
>             Project: IMPALA
>          Issue Type: Bug
>          Components: Backend
>    Affects Versions: Impala 3.2.0
>            Reporter: Thomas Tauber-Marshall
>            Assignee: Joe McDonnell
>            Priority: Blocker
>              Labels: broken-build, flaky
>
> Seen in an exhaustive build on centos6:
> {noformat}
> 05:39:09  TestJoinQueries.test_empty_build_joins[batch_size: 1 | protocol: 
> beeswax | exec_option: {'batch_size': 0, 'num_nodes': 0, 
> 'disable_codegen_rows_threshold': 0, 'disable_codegen': False, 
> 'abort_on_error': 1, 'debug_action': None, 'exec_single_node_rows_threshold': 
> 0} | table_format: parquet/none] 
> 05:39:09 [gw3] linux2 -- Python 2.6.6 
> /data/jenkins/workspace/impala-asf-master-exhaustive-centos6/repos/Impala/bin/../infra/python/env/bin/python
> 05:39:09 query_test/test_join_queries.py:97: in test_empty_build_joins
> 05:39:09     self.run_test_case('QueryTest/empty-build-joins', new_vector)
> 05:39:09 common/impala_test_suite.py:467: in run_test_case
> 05:39:09     result = self.__execute_query(target_impalad_client, query, 
> user=user)
> 05:39:09 common/impala_test_suite.py:688: in __execute_query
> 05:39:09     return impalad_client.execute(query, user=user)
> 05:39:09 common/impala_connection.py:170: in execute
> 05:39:09     return self.__beeswax_client.execute(sql_stmt, user=user)
> 05:39:09 beeswax/impala_beeswax.py:182: in execute
> 05:39:09     handle = self.__execute_query(query_string.strip(), user=user)
> 05:39:09 beeswax/impala_beeswax.py:359: in __execute_query
> 05:39:09     self.wait_for_finished(handle)
> 05:39:09 beeswax/impala_beeswax.py:380: in wait_for_finished
> 05:39:09     raise ImpalaBeeswaxException("Query aborted:" + error_log, None)
> 05:39:09 E   ImpalaBeeswaxException: ImpalaBeeswaxException:
> 05:39:09 E    Query aborted:hdfsOpenFile() for 
> hdfs://localhost:20500/test-warehouse/alltypestiny/year=2009/month=2/090201.txt
>  failed to finish before the 300 second timeout
> 05:39:09 ----------------------------- Captured stderr call 
> -----------------------------
> 05:39:09 -- executing against localhost:21000
> 05:39:09 use functional_parquet;
> 05:39:09 
> 05:39:09 -- 2018-12-11 03:11:34,797 INFO     MainThread: Started query 
> d747763f9d663cd7:9abd4b9900000000
> 05:39:09 SET batch_size=1;
> 05:39:09 SET num_nodes=0;
> 05:39:09 SET disable_codegen_rows_threshold=0;
> 05:39:09 SET disable_codegen=False;
> 05:39:09 SET abort_on_error=1;
> 05:39:09 SET exec_single_node_rows_threshold=0;
> 05:39:09 -- executing against localhost:21000
> 05:39:09 select straight_join atp.id
> 05:39:09 from alltypes atp
> 05:39:09   inner join functional.alltypestiny att on atp.id = att.id
> 05:39:09 where att.int_col = 999;
> 05:39:09 
> 05:39:09 -- 2018-12-11 03:11:34,816 INFO     MainThread: Started query 
> 5045de8553c5843c:bdc6aa1c00000000
> 05:39:09 -- executing against localhost:21000
> 05:39:09 select straight_join atp.id
> 05:39:09 from alltypes atp
> 05:39:09   right join functional.alltypestiny att on atp.id = att.id
> 05:39:09 where att.int_col = 999;
> 05:39:09 
> 05:39:09 -- 2018-12-11 03:11:35,519 INFO     MainThread: Started query 
> 124ef451a3f65d09:f2ae4a5d00000000
> {noformat}
> Presumably caused by IMPALA-7738



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to