mirror of
https://github.com/apache/impala.git
synced 2026-01-01 09:00:42 -05:00
99879aad47
We used to check error_log empty with error_log.empty(), but this may return false even when some of its member has empty messages, thus crashes impala. This commit: 1. Remove error_log from HdfsScanNode::ProcessSplit() since the logs may contain irrelevant errors. This also prevents a potential race condition when error_log is checked and enters if clause, but later is changed by other threads. Also prevent crashes when error_log has a cleared entry. 2. Hold locks and return a copy of the RuntimeState::error_log_, fixing a race in the coordinator. Change-Id: I3a7e3d22e26147ada780aae5aed1f2e25a515afc Reviewed-on: http://gerrit.cloudera.org:8080/2829 Reviewed-by: Dan Hecht <dhecht@cloudera.com> Reviewed-by: Huaisi Xu <hxu@cloudera.com> Tested-by: Internal Jenkins
54 lines
2.3 KiB
Python
54 lines
2.3 KiB
Python
# Copyright (c) 2012 Cloudera, Inc. All rights reserved.
|
|
# Injects failures at specific locations in each of the plan nodes. Currently supports
|
|
# two types of failures - cancellation of the query and a failure test hook.
|
|
#
|
|
from tests.beeswax.impala_beeswax import ImpalaBeeswaxException
|
|
from tests.common.impala_test_suite import ImpalaTestSuite
|
|
from tests.common.test_dimensions import create_exec_option_dimension
|
|
from time import sleep
|
|
|
|
# Test injecting error logs in prepare phase and status::OK(). This tests one of race
|
|
# conditions in error reporting (IMPALA-3385).
|
|
class TestErrorLogs(ImpalaTestSuite):
|
|
@classmethod
|
|
def get_workload(cls):
|
|
return 'functional-query'
|
|
|
|
@classmethod
|
|
def add_test_dimensions(cls):
|
|
super(TestErrorLogs, cls).add_test_dimensions()
|
|
cls.TestMatrix.add_constraint(lambda v:\
|
|
v.get_value('table_format').file_format == 'text')
|
|
cls.TestMatrix.add_dimension(create_exec_option_dimension(
|
|
cluster_sizes=[0], disable_codegen_options=[False], batch_sizes=[0]))
|
|
|
|
def test_errorlog(self, vector):
|
|
query = 'select count(*) from tpch.lineitem;'
|
|
action = 'INJECT_ERROR_LOG'
|
|
location = 'PREPARE'
|
|
# scan node has id=0
|
|
node_id = 0
|
|
# Without restarting impala, reporting thread reports every 5 seconds.
|
|
# Having a lower scan range can make the query run more than 2*5 seconds
|
|
# Select location in prepare to guarantee at least one cleared error maps will
|
|
# be sent to coordinator.
|
|
debug_action = '%d:%s:%s' % (node_id, location, action)
|
|
vector.get_value('exec_option')['debug_action'] = debug_action
|
|
vector.get_value('exec_option')['MAX_SCAN_RANGE_LENGTH'] = '1000'
|
|
self.__execute_inject_error_log_action(query, vector)
|
|
|
|
def __execute_inject_error_log_action(self, query, vector):
|
|
try:
|
|
handle = self.execute_query_async(query, vector.get_value('exec_option'))
|
|
# sleep() is used to terminate the query; otherwise it runs for a long time. It is
|
|
# large enough to further guarantee at least one cleared error maps to be sent to
|
|
# coordinator.
|
|
sleep(30)
|
|
cancel_result = self.client.cancel(handle)
|
|
self.client.close_query(handle)
|
|
assert cancel_result.status_code == 0,\
|
|
'Unexpected status code from cancel request: %s' % cancel_result
|
|
# As long as impala did not crash we are good.
|
|
except ImpalaBeeswaxException:
|
|
return
|