def _create_executor(self, executor_name):
query_options = {
'impala_beeswax': lambda: (execute_using_impala_beeswax,
BeeswaxQueryExecConfig(plugin_runner=self.config.plugin_runner,
exec_options=self.config.exec_options,
use_kerberos=self.config.use_kerberos,
user=self.config.user if self.config.password else None,
password=self.config.password,
use_ssl=self.config.use_ssl
)),
'impala_jdbc': lambda: (execute_using_jdbc,
JdbcQueryExecConfig(plugin_runner=self.config.plugin_runner)
),
'impala_hs2': lambda: (execute_using_impala_hs2,
ImpalaHS2QueryConfig(plugin_runner=self.config.plugin_runner,
use_kerberos=self.config.use_kerberos
)),
'hive_hs2': lambda: (execute_using_hive_hs2,
HiveHS2QueryConfig(hiveserver=self.config.hiveserver,
plugin_runner=self.config.plugin_runner,
exec_options=self.config.exec_options,
user=self.config.user,
use_kerberos=self.config.use_kerberos
))
} [executor_name]()
return query_options
def exec_and_compare_hive_and_impala_hs2(self, stmt, compare = lambda x, y: x == y):
"""Compare Hive and Impala results when executing the same statment over HS2"""
# execute_using_jdbc expects a Query object. Convert the query string into a Query
# object
query = Query()
query.query_str = stmt
# Run the statement targeting Hive
exec_opts = JdbcQueryExecConfig(impalad=HIVE_HS2_HOST_PORT, transport='SASL')
hive_results = execute_using_jdbc(query, exec_opts).data
# Run the statement targeting Impala
exec_opts = JdbcQueryExecConfig(impalad=IMPALAD_HS2_HOST_PORT, transport='NOSASL')
impala_results = execute_using_jdbc(query, exec_opts).data
# Compare the results
assert (impala_results is not None) and (hive_results is not None)
assert compare(impala_results, hive_results)
def exec_and_compare_hive_and_impala_hs2(self, stmt):
"""Compare Hive and Impala results when executing the same statment over HS2"""
# execute_using_jdbc expects a Query object. Convert the query string into a Query
# object
query = Query()
query.query_str = stmt
# Run the statement targeting Hive
exec_opts = JdbcQueryExecConfig(impalad=HIVE_HS2_HOST_PORT)
hive_results = execute_using_jdbc(query, exec_opts).data
# Run the statement targeting Impala
exec_opts = JdbcQueryExecConfig(impalad=IMPALAD_HS2_HOST_PORT)
impala_results = execute_using_jdbc(query, exec_opts).data
# Compare the results
assert (impala_results is not None) and (hive_results is not None)
for impala, hive in zip(impala_results, hive_results):
assert impala == hive
def _create_executor(self, executor_name):
query_options = {
'impala_beeswax': lambda: (execute_using_impala_beeswax,
BeeswaxQueryExecConfig(plugin_runner=self.config.plugin_runner,
exec_options=self.config.exec_options,
use_kerberos=self.config.use_kerberos,
)),
'jdbc': lambda: (execute_using_jdbc,
JdbcQueryExecConfig(plugin_runner=self.config.plugin_runner)
)
} [executor_name]()
return query_options