[
https://issues.apache.org/jira/browse/IMPALA-9267?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17004354#comment-17004354
]
Joe McDonnell commented on IMPALA-9267:
---------------------------------------
This seems to happen during hs2 tests. First symptom:
{noformat}
06:09:57 [gw2] PASSED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
06:09:58
hs2/test_fetch_timeout.py::TestFetchTimeoutWithResultSpooling::test_multiple_fetch_multiple_batches_timeout
<- hs2/hs2_test_suite.py
06:09:59 [gw3] PASSED hs2/test_hs2.py::TestHS2::test_get_tables <-
hs2/hs2_test_suite.py
06:09:59 hs2/test_hs2.py::TestHS2::test_open_session_unsupported_protocol
06:09:59 [gw3] PASSED
hs2/test_hs2.py::TestHS2::test_open_session_unsupported_protocol
06:10:03 hs2/test_hs2.py::TestHS2::test_close_connection
06:10:03 [gw0] PASSED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_materialization_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
06:14:37
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_before_finished_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
06:14:37 [gw2] FAILED
hs2/test_fetch_timeout.py::TestFetchTimeoutWithResultSpooling::test_multiple_fetch_multiple_batches_timeout
<- hs2/hs2_test_suite.py
06:14:37 [gw1] FAILED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_timeout <-
hs2/hs2_test_suite.py
06:14:37 [gw0] FAILED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_before_finished_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
06:14:37 [gw3] FAILED hs2/test_hs2.py::TestHS2::test_close_connection {noformat}
Second symptom:
{noformat}
23:01:13 [gw6] PASSED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_finished_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
23:01:13
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_before_finished_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
23:01:13 [gw1] PASSED
hs2/test_fetch_timeout.py::TestFetchTimeoutWithResultSpooling::test_multiple_fetch_multiple_batches_timeout
<- hs2/hs2_test_suite.py
23:01:14 hs2/test_hs2.py::TestHS2::test_close_connection <- hs2/test_hs2.py
23:01:14 [gw7] PASSED hs2/test_hs2.py::TestHS2::test_get_log <-
hs2/hs2_test_suite.py
23:01:16 hs2/test_hs2.py::TestHS2::test_session_options_via_set <-
hs2/hs2_test_suite.py
23:01:16 [gw0] ERROR hs2/test_hs2.py::TestHS2::test_get_primary_keys <-
hs2/hs2_test_suite.py
23:01:16 [gw5] FAILED
metadata/test_ddl.py::TestDdlStatements::test_drop_cleans_hdfs_dirs
23:01:16 [gw2] FAILED
metadata/test_ddl.py::TestDdlStatements::test_truncate_cleans_hdfs_files
23:01:16 [gw6] FAILED
hs2/test_fetch_timeout.py::TestFetchTimeout::test_fetch_before_finished_timeout_with_result_spooling
<- hs2/hs2_test_suite.py
23:01:16 hs2/test_hs2.py::TestHS2::test_open_session_empty_user <-
hs2/test_hs2.py {noformat}
Uploaded a backtrace.
> Impala hits DCHECK in ClientRequestState due to invalid state transition
> ------------------------------------------------------------------------
>
> Key: IMPALA-9267
> URL: https://issues.apache.org/jira/browse/IMPALA-9267
> Project: IMPALA
> Issue Type: Bug
> Components: Backend
> Affects Versions: Impala 3.4.0
> Reporter: Joe McDonnell
> Priority: Blocker
> Labels: broken-build
> Attachments: backtrace.txt
>
>
> Some test runs see the coordinator Impalad hit this DCHECK:
> {code:java}
> case ExecState::FINISHED:
> // A query can transition from PENDING to FINISHED if it is cancelled by the
> // client.
> DCHECK(old_state == ExecState::PENDING || old_state == ExecState::RUNNING)
> << Substitute(
> error_msg, ExecStateToString(old_state),
> ExecStateToString(new_state));
> UpdateExecState(new_state);
> break;{code}
> The error message is:
> {noformat}
> F1222 23:01:14.020613 10494 client-request-state.cc:951] Check failed:
> old_state == ExecState::PENDING || old_state == ExecState::RUNNING Illegal
> state transition: ERROR -> FINISHED{noformat}
> This happened on one exhaustive run on Centos6 and an s3 run. There doesn't
> seem to be a pattern, so I'm assuming this is a generic issue.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
