[ 
https://issues.apache.org/jira/browse/IMPALA-9757?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17117005#comment-17117005
 ] 

Sahil Takiar commented on IMPALA-9757:
--------------------------------------

Did some investigation, and I don't think this is actually a bug in the code, 
just the way the tests are run.

It looks like the hs2-http connections are just timing out and getting cleaned 
up by the session maintenance threads:

{code}
logs/ee_tests/impalad.impala-ec2-centos74-m5-4xlarge-ondemand-13fb.vpc.cloudera.com.jenkins.log.INFO.20200522-193649.23127:I0523
 02:20:07.126468 13972 impala-server.cc:1079] 
28419d6d1127487e:0e10689800000000] Registered query 
query_id=28419d6d1127487e:0e10689800000000 
session_id=f94437c282d6c1c3:204ad55e51c6d6a9
logs/ee_tests/impalad.impala-ec2-centos74-m5-4xlarge-ondemand-13fb.vpc.cloudera.com.jenkins.log.INFO.20200522-193649.23127:I0523
 03:20:37.240694 24158 impala-server.cc:2229] Closing session: 
f94437c282d6c1c3:204ad55e51c6d6a9, user: jenkins, because it no longer  has any 
open connections. The last connection was closed at: 2020-05-23 02:20:37.158
logs/ee_tests/impalad.impala-ec2-centos74-m5-4xlarge-ondemand-13fb.vpc.cloudera.com.jenkins.log.INFO.20200522-193649.23127:I0523
 03:48:52.916052  5569 impala-server.cc:1403] GetSessionState(): Invalid 
session id: f94437c282d6c1c3:204ad55e51c6d6a9
logs/ee_tests/impalad.impala-ec2-centos74-m5-4xlarge-ondemand-13fb.vpc.cloudera.com.jenkins.log.INFO.20200522-193649.23127:I0523
 03:48:53.082515  5611 impala-server.cc:1334] Closing session: 
f94437c282d6c1c3:204ad55e51c6d6a9
logs/ee_tests/impalad.impala-ec2-centos74-m5-4xlarge-ondemand-13fb.vpc.cloudera.com.jenkins.log.INFO.20200522-193649.23127:I0523
 03:48:53.082520  5611 impala-server.cc:1351] CloseSessionInternal(): Invalid 
session id: f94437c282d6c1c3:204ad55e51c6d6a9
{code}

Looking through the code, the session gets invalidated if it is not used for 
more than 15 minutes. It looks like in some exhaustive tests this can happen. 
The hs2-http connection is opened when the class is created (see 
impala_test_suite.py), a few tests run that use the hs2-http, then some other 
tests run that don't use hs2-http. 15 minutes elapse and the hs2-http 
connection is closed, and then future attempts to use the hs2-http connection 
fail. It looks like this only happens in exhaustive tests because we typically 
only add the client protocol as a test dimension in exhaustive builds.

A simple fix might be to just bump the session timeout for tests.

> Test failures with HiveServer2Error: Invalid session id
> -------------------------------------------------------
>
>                 Key: IMPALA-9757
>                 URL: https://issues.apache.org/jira/browse/IMPALA-9757
>             Project: IMPALA
>          Issue Type: Test
>            Reporter: Sahil Takiar
>            Assignee: Sahil Takiar
>            Priority: Major
>              Labels: broken-build, flaky
>
> Only seen once so far on an exhaustive build. It's not clear if the 
> "HiveServer2Error: Invalid session id" error is specific to this test or not.
> {code:java}
> query_test.test_queries.TestQueries.test_inline_view[protocol: hs2-http | 
> exec_option: {'batch_size': 0, 'num_nodes': 0, 
> 'disable_codegen_rows_threshold': 0, 'disable_codegen': False, 
> 'abort_on_error': 1, 'exec_single_node_rows_threshold': 0} | table_format: 
> parquet/none] (from pytest)
> Error Message
> query_test/test_queries.py:104: in test_inline_view     
> self.run_test_case('QueryTest/inline-view', vector) 
> common/impala_test_suite.py:567: in run_test_case     table_format_info, 
> use_db, pytest.config.option.scale_factor) common/impala_test_suite.py:782: 
> in change_database     impala_client.execute(query) 
> common/impala_connection.py:331: in execute     handle = 
> self.execute_async(sql_stmt, user) common/impala_connection.py:354: in 
> execute_async     self.__cursor.execute_async(sql_stmt, 
> configuration=self.__query_options) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:375:
>  in execute_async     self._execute_async(op) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:394:
>  in _execute_async     operation_fn() 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:372:
>  in op     run_async=True) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:1096:
>  in execute     return self._operation('ExecuteStatement', req) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:1026:
>  in _operation     resp = self._rpc(kind, request) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:994:
>  in _rpc     err_if_rpc_not_ok(response) 
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:748:
>  in err_if_rpc_not_ok     raise HiveServer2Error(resp.status.errorMessage) E  
>  HiveServer2Error: Invalid session id: 3345279d9b2e75ab:3aef93f7a80d7d8a
> Stacktrace
> query_test/test_queries.py:104: in test_inline_view
>     self.run_test_case('QueryTest/inline-view', vector)
> common/impala_test_suite.py:567: in run_test_case
>     table_format_info, use_db, pytest.config.option.scale_factor)
> common/impala_test_suite.py:782: in change_database
>     impala_client.execute(query)
> common/impala_connection.py:331: in execute
>     handle = self.execute_async(sql_stmt, user)
> common/impala_connection.py:354: in execute_async
>     self.__cursor.execute_async(sql_stmt, configuration=self.__query_options)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:375:
>  in execute_async
>     self._execute_async(op)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:394:
>  in _execute_async
>     operation_fn()
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:372:
>  in op
>     run_async=True)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:1096:
>  in execute
>     return self._operation('ExecuteStatement', req)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:1026:
>  in _operation
>     resp = self._rpc(kind, request)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:994:
>  in _rpc
>     err_if_rpc_not_ok(response)
> /data/jenkins/workspace/impala-asf-master-exhaustive/repos/Impala/infra/python/env/lib/python2.7/site-packages/impala/hiveserver2.py:748:
>  in err_if_rpc_not_ok
>     raise HiveServer2Error(resp.status.errorMessage)
> E   HiveServer2Error: Invalid session id: 3345279d9b2e75ab:3aef93f7a80d7d8a 
> {code}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org
For additional commands, e-mail: issues-all-h...@impala.apache.org

Reply via email to