This is an automated email from the ASF dual-hosted git repository.
prozsa pushed a commit to branch branch-4.5.0
in repository https://gitbox.apache.org/repos/asf/impala.git
commit e4b5f36003dffb73256c5e614f88ceb5b64e1972
Author: Riza Suminto <riza.sumi...@cloudera.com>
AuthorDate: Mon Feb 10 18:23:11 2025 -0800
IMPALA-13747: Use fresh HS2 client for unique_database fixture
Just like IMPALA-13634 bug, unique_database fixture may taint
ImpalaTestSuite.client by setting 'sync_ddl=True' option and never
cleaning it up for test method usage.
This patch fixes unique_database fixture by always creating fresh HS2
client during CREATE DATABASE and DROP DATABASE.
Testing:
- Caught an assert error at test_set_fallback_db_for_functions
Expected: "default.fn() unknown for database default"
Actual: "functional.fn() unknown for database functional"
In this case, shared ImpalaTestSuite.client may have changed database
via ImpalaTestSuite.run_test_case() or
ImpalaTestSuite.execute_wrapper() in other test method.
Removing unused 'vector' argument somehow fixed the issue.
- For both query_test/test_udfs.py and metadata/test_ddl.py:
- Fixed flake8 issues and remove unused pytest fixture.
- Run and pass the tests in exhaustive exploration.
Change-Id: Ib503e829552d436035c57b489ffda0d0299f8405
Reviewed-on: http://gerrit.cloudera.org:8080/22471
Reviewed-by: Riza Suminto <riza.sumi...@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenk...@cloudera.com>
---
tests/conftest.py | 42 +++++++++++++++----------
tests/metadata/test_ddl.py | 72 +++++++++++++++++++++++--------------------
tests/query_test/test_udfs.py | 41 +++++++++++++-----------
3 files changed, 88 insertions(+), 67 deletions(-)
diff --git a/tests/conftest.py b/tests/conftest.py
index 9366e021c..59d2ebbd6 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -83,7 +83,7 @@ def configure_logging():
# Verify that the logging level is set to the correct value.
rootLoggerLevel = logging.getLogger().getEffectiveLevel()
print("rootLoggerLevel = {0}".format(logging.getLevelName(rootLoggerLevel)))
- assert(rootLoggerLevel == logging.INFO)
+ assert (rootLoggerLevel == logging.INFO)
def pytest_addoption(parser):
@@ -192,6 +192,12 @@ def pytest_addoption(parser):
help="Impala protocol to run test if test does not specify
any.")
+def assert_impala_test_suite(request):
+ assert issubclass(request.cls,
tests.common.impala_test_suite.ImpalaTestSuite), (
+ "This fixture is only applicable to ImpalaTestSuite subclasses."
+ )
+
+
def pytest_assertrepr_compare(op, left, right):
"""
Provides a hook for outputting type-specific assertion messages
@@ -353,6 +359,7 @@ def unique_database(request, testid_checksum):
assert 'function' == request.scope, ('This fixture must have scope
"function" since '
'the fixture must guarantee unique
per-test '
'databases.')
+ assert_impala_test_suite(request)
db_name_prefix = request.function.__name__
sync_ddl = False
@@ -376,11 +383,10 @@ def unique_database(request, testid_checksum):
' test function name or any prefixes for long length or
invalid '
'characters.'.format(db_name))
- def cleanup_database(db_name, must_exist):
- request.instance.execute_query_expect_success(request.instance.client,
- 'DROP DATABASE {0} `{1}` CASCADE'.format(
- "" if must_exist else "IF EXISTS", db_name),
- {'sync_ddl': sync_ddl})
+ def cleanup_database(client, db_name, must_exist):
+ result = client.execute('DROP DATABASE {0} `{1}` CASCADE'.format(
+ "" if must_exist else "IF EXISTS", db_name))
+ assert result.success
# The database directory may not be removed if there are external tables
in the
# database when it is dropped. The external locations are not removed by
cascade.
# These preexisting files/directories can cause errors when tests run
repeatedly or
@@ -391,21 +397,25 @@ def unique_database(request, testid_checksum):
def cleanup():
# Make sure we don't try to drop the current session database
+ # TODO: clean this up via IMPALA-13758.
request.instance.execute_query_expect_success(request.instance.client,
"use default")
- for db_name in db_names:
- cleanup_database(db_name, True)
- LOG.info('Dropped database "{0}" for test ID "{1}"'.format(
+ with request.cls.create_impala_client(protocol=HS2) as client:
+ client.set_configuration({'sync_ddl': sync_ddl})
+ for db_name in db_names:
+ cleanup_database(client, db_name, True)
+ LOG.info('Dropped database "{0}" for test ID "{1}"'.format(
db_name, str(request.node.nodeid)))
request.addfinalizer(cleanup)
- for db_name in db_names:
- cleanup_database(db_name, False)
- request.instance.execute_query_expect_success(
- request.instance.client, 'CREATE DATABASE `{0}`'.format(db_name),
- {'sync_ddl': sync_ddl})
- LOG.info('Created database "{0}" for test ID "{1}"'.format(db_name,
-
str(request.node.nodeid)))
+ with request.cls.create_impala_client(protocol=HS2) as client:
+ client.set_configuration({'sync_ddl': sync_ddl})
+ for db_name in db_names:
+ cleanup_database(client, db_name, False)
+ result = client.execute('CREATE DATABASE `{0}`'.format(db_name))
+ assert result.success
+ LOG.info('Created database "{0}" for test ID "{1}"'.format(
+ db_name, str(request.node.nodeid)))
return first_db_name
diff --git a/tests/metadata/test_ddl.py b/tests/metadata/test_ddl.py
index 782629ac3..c6e64ab22 100644
--- a/tests/metadata/test_ddl.py
+++ b/tests/metadata/test_ddl.py
@@ -31,8 +31,12 @@ from tests.common.environ import (HIVE_MAJOR_VERSION)
from tests.common.file_utils import create_table_from_orc
from tests.common.impala_test_suite import LOG
from tests.common.parametrize import UniqueDatabase
-from tests.common.skip import (SkipIf, SkipIfFS, SkipIfKudu, SkipIfLocal,
- SkipIfCatalogV2, SkipIfHive2,
SkipIfDockerizedCluster)
+from tests.common.skip import (
+ SkipIfDockerizedCluster,
+ SkipIfFS,
+ SkipIfHive2,
+ SkipIfKudu,
+ SkipIfLocal)
from tests.common.test_dimensions import create_single_exec_option_dimension
from tests.common.test_dimensions import (create_exec_option_dimension,
create_client_protocol_dimension, create_exec_option_dimension_from_dict)
@@ -56,6 +60,7 @@ def get_trash_path(bucket, path):
getpass.getuser(), WAREHOUSE_PREFIX, path))
return '/user/{0}/.Trash/Current/{1}/{2}'.format(getpass.getuser(), bucket,
path)
+
# Validates DDL statements (create, drop)
class TestDdlStatements(TestDdlBase):
@SkipIfLocal.hdfs_client
@@ -129,7 +134,7 @@ class TestDdlStatements(TestDdlBase):
self._create_db(unique_database)
self.client.execute("create table {0}.t1(i int)".format(unique_database))
self.client.execute("create table {0}.t2(i int)".format(unique_database))
- result = self.client.execute("create external table {0}.t3(i int) "
+ self.client.execute("create external table {0}.t3(i int) "
"location '{1}/{0}/t3/'".format(unique_database, WAREHOUSE))
self.client.execute("drop database {0} cascade".format(unique_database))
assert not self.filesystem_client.exists(
@@ -184,7 +189,8 @@ class TestDdlStatements(TestDdlBase):
assert len(self.filesystem_client.ls(
"{1}/{0}.db/t2/p=1".format(unique_database, WAREHOUSE))) == 1
- # Truncating the table removes the data files and preserves the
partition's directory
+ # Truncating the table removes the data files and preserves the
partition's
+ # directory
self.client.execute("truncate table {0}.t2".format(unique_database))
assert self.filesystem_client.exists(
"{1}/{0}.db/t2/p=1".format(unique_database, WAREHOUSE))
@@ -208,7 +214,7 @@ class TestDdlStatements(TestDdlBase):
self.run_test_case('QueryTest/create-database', vector,
use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
- def test_comment_on_database(self, vector, unique_database):
+ def test_comment_on_database(self, unique_database):
comment = self._get_db_comment(unique_database)
assert '' == comment
@@ -220,7 +226,8 @@ class TestDdlStatements(TestDdlBase):
comment = self._get_db_comment(unique_database)
assert '' == comment
- self.client.execute("comment on database {0} is
'\\'comment\\''".format(unique_database))
+ self.client.execute("comment on database {0} is '\\'comment\\''".format(
+ unique_database))
comment = self._get_db_comment(unique_database)
assert "\\'comment\\'" == comment
@@ -228,7 +235,7 @@ class TestDdlStatements(TestDdlBase):
comment = self._get_db_comment(unique_database)
assert '' == comment
- def test_alter_database_set_owner(self, vector, unique_database):
+ def test_alter_database_set_owner(self, unique_database):
self.client.execute("alter database {0} set owner user foo_user".format(
unique_database))
properties = self._get_db_owner_properties(unique_database)
@@ -241,7 +248,7 @@ class TestDdlStatements(TestDdlBase):
assert len(properties) == 1
assert {'foo_role': 'ROLE'} == properties
- def test_metadata_after_alter_database(self, vector, unique_database):
+ def test_metadata_after_alter_database(self, unique_database):
self.client.execute("create table {0}.tbl (i int)".format(unique_database))
self.client.execute("create function {0}.f() returns int "
"location '{1}/libTestUdfs.so' symbol='NoArgs'"
@@ -255,7 +262,7 @@ class TestDdlStatements(TestDdlBase):
unique_database)).get_data()
assert "INT\tf()\tNATIVE\ttrue" == func_names
- def test_alter_table_set_owner(self, vector, unique_database):
+ def test_alter_table_set_owner(self, unique_database):
table_name = "{0}.test_owner_tbl".format(unique_database)
self.client.execute("create table {0}(i int)".format(table_name))
self.client.execute("alter table {0} set owner user
foo_user".format(table_name))
@@ -266,7 +273,7 @@ class TestDdlStatements(TestDdlBase):
owner = self._get_table_or_view_owner(table_name)
assert ('foo_role', 'ROLE') == owner
- def test_alter_view_set_owner(self, vector, unique_database):
+ def test_alter_view_set_owner(self, unique_database):
view_name = "{0}.test_owner_tbl".format(unique_database)
self.client.execute("create view {0} as select 1".format(view_name))
self.client.execute("alter view {0} set owner user
foo_user".format(view_name))
@@ -333,7 +340,7 @@ class TestDdlStatements(TestDdlBase):
self.run_test_case('QueryTest/kudu_create', vector, use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
- def test_comment_on_table(self, vector, unique_database):
+ def test_comment_on_table(self, unique_database):
table = '{0}.comment_table'.format(unique_database)
self.client.execute("create table {0} (i int)".format(table))
@@ -356,7 +363,7 @@ class TestDdlStatements(TestDdlBase):
comment = self._get_table_or_view_comment(table)
assert comment is None
- def test_comment_on_view(self, vector, unique_database):
+ def test_comment_on_view(self, unique_database):
view = '{0}.comment_view'.format(unique_database)
self.client.execute("create view {0} as select 1".format(view))
@@ -379,7 +386,7 @@ class TestDdlStatements(TestDdlBase):
comment = self._get_table_or_view_comment(view)
assert comment is None
- def test_comment_on_column(self, vector, unique_database):
+ def test_comment_on_column(self, unique_database):
table = "{0}.comment_table".format(unique_database)
self.client.execute("create table {0} (i int) partitioned by (j
int)".format(table))
@@ -431,7 +438,7 @@ class TestDdlStatements(TestDdlBase):
assert "" == comment
@UniqueDatabase.parametrize(sync_ddl=True)
- def test_sync_ddl_drop(self, vector, unique_database):
+ def test_sync_ddl_drop(self, unique_database):
"""Verifies the catalog gets updated properly when dropping objects with
sync_ddl
enabled"""
self.client.set_configuration({'sync_ddl': 1})
@@ -479,7 +486,7 @@ class TestDdlStatements(TestDdlBase):
use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
@SkipIfLocal.hdfs_client
- def test_drop_partition_with_purge(self, vector, unique_database):
+ def test_drop_partition_with_purge(self, unique_database):
"""Verfies whether alter <tbl> drop partition purge actually skips trash"""
self.client.execute(
"create table {0}.t1(i int) partitioned by (j
int)".format(unique_database))
@@ -497,8 +504,8 @@ class TestDdlStatements(TestDdlBase):
assert self.filesystem_client.exists('{}/t1/j=1/j1.txt'.format(trash))
assert self.filesystem_client.exists('{}/t1/j=1'.format(trash))
# Drop the partition (with purge) and make sure it doesn't exist in trash
- self.client.execute("alter table {0}.t1 drop partition(j=2) purge".\
- format(unique_database));
+ self.client.execute("alter table {0}.t1 drop partition(j=2) purge".format(
+ unique_database))
if not IS_S3 and not IS_ADLS:
# In S3, deletes are eventual. So even though we dropped the partition,
the files
# belonging to this partition may still be visible for some unbounded
time. This
@@ -518,7 +525,7 @@ class TestDdlStatements(TestDdlBase):
multiple_impalad=self._use_multiple_impalad(vector))
@UniqueDatabase.parametrize()
- def test_view_hints(self, vector, unique_database):
+ def test_view_hints(self, unique_database):
# Test that plan hints are stored in the view's comment field; this should
work
# regardless of how Hive formats the output. Getting this to work with the
# automated test case runner is rather difficult, so verify directly.
There
@@ -530,7 +537,7 @@ class TestDdlStatements(TestDdlBase):
inner join /* +shuffle */ functional.alltypessmall c on b.id = c.id
""".format(unique_database))
results = self.execute_query("describe formatted %s.hints_test" %
unique_database)
- sj, bc, shuf = 0,0,0
+ sj, bc, shuf = 0, 0, 0
for row in results.data:
sj += '-- +straight_join' in row
bc += '-- +broadcast' in row
@@ -547,7 +554,7 @@ class TestDdlStatements(TestDdlBase):
# Test the plan to make sure hints were applied correctly
plan = self.execute_query("explain select * from %s.hints_test" %
unique_database,
- query_options={'explain_level':0})
+ query_options={'explain_level': 0})
plan_match = """PLAN-ROOT SINK
08:EXCHANGE [UNPARTITIONED]
04:HASH JOIN [INNER JOIN, PARTITIONED]
@@ -580,7 +587,7 @@ class TestDdlStatements(TestDdlBase):
try:
result = self.execute_query_expect_success(
client, "describe formatted %s" % view_name)
- exp_line = [l for l in result.data if 'View Expanded' in l][0]
+ exp_line = [line for line in result.data if 'View Expanded' in
line][0]
except ImpalaBeeswaxException as e:
# In non-SYNC_DDL tests, it's OK to get a "missing view" type error
# until the metadata propagates.
@@ -591,7 +598,6 @@ class TestDdlStatements(TestDdlBase):
finally:
client.close()
-
def test_views_describe(self, vector, unique_database):
# IMPALA-6896: Tests that altered views can be described by all impalads.
impala_cluster = ImpalaCluster.get_e2e_test_cluster()
@@ -617,14 +623,13 @@ class TestDdlStatements(TestDdlBase):
finally:
first_client.close()
-
@UniqueDatabase.parametrize(sync_ddl=True)
def test_functions_ddl(self, vector, unique_database):
self.run_test_case('QueryTest/functions-ddl', vector,
use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
@SkipIfLocal.hdfs_client
- def test_create_alter_bulk_partition(self, vector, unique_database):
+ def test_create_alter_bulk_partition(self, unique_database):
# Change the scale depending on the exploration strategy, with 50
partitions this
# test runs a few minutes, with 10 partitions it takes ~50s for two
configurations.
num_parts = 50 if self.exploration_strategy() == 'exhaustive' else 10
@@ -667,7 +672,7 @@ class TestDdlStatements(TestDdlBase):
assert '1' == self.execute_scalar("select count(*) from
{0}".format(fq_tbl_name))
@SkipIfLocal.hdfs_client
- def test_alter_table_set_fileformat(self, vector, unique_database):
+ def test_alter_table_set_fileformat(self, unique_database):
# Tests that SET FILEFORMAT clause is set for ALTER TABLE ADD PARTITION
statement
fq_tbl_name = unique_database + ".p_fileformat"
self.client.execute(
@@ -689,7 +694,7 @@ class TestDdlStatements(TestDdlBase):
assert 1 == len([line for line in result.data if line.find("PARQUET") !=
-1])
assert 2 == len([line for line in result.data if line.find("ORC") != -1])
- def test_alter_table_create_many_partitions(self, vector, unique_database):
+ def test_alter_table_create_many_partitions(self, unique_database):
"""
Checks that creating more partitions than the MAX_PARTITION_UPDATES_PER_RPC
batch size works, in that it creates all the underlying partitions.
@@ -709,7 +714,7 @@ class TestDdlStatements(TestDdlBase):
assert list(map(int, PARTITION_RE.findall(str(partitions)))) == \
list(range(MAX_PARTITION_UPDATES_PER_RPC + 2))
- def test_create_alter_tbl_properties(self, vector, unique_database):
+ def test_create_alter_tbl_properties(self, unique_database):
fq_tbl_name = unique_database + ".test_alter_tbl"
# Specify TBLPROPERTIES and SERDEPROPERTIES at CREATE time
@@ -756,7 +761,7 @@ class TestDdlStatements(TestDdlBase):
assert properties[''] == ''
@SkipIfHive2.acid
- def test_create_insertonly_tbl(self, vector, unique_database):
+ def test_create_insertonly_tbl(self, unique_database):
insertonly_tbl = unique_database + ".test_insertonly"
self.client.execute("""create table {0} (coli int) stored as parquet
tblproperties(
'transactional'='true', 'transactional_properties'='insert_only')"""
@@ -764,7 +769,7 @@ class TestDdlStatements(TestDdlBase):
properties = self._get_tbl_properties(insertonly_tbl)
assert properties['OBJCAPABILITIES'] ==
'HIVEMANAGEDINSERTREAD,HIVEMANAGEDINSERTWRITE'
- def test_alter_tbl_properties_reload(self, vector, unique_database):
+ def test_alter_tbl_properties_reload(self, unique_database):
# IMPALA-8734: Force a table schema reload when setting table properties.
tbl_name = "test_tbl"
self.execute_query_expect_success(self.client, "create table {0}.{1} (c1
string)"
@@ -792,7 +797,7 @@ class TestDdlStatements(TestDdlBase):
self.run_test_case('QueryTest/partition-ddl-predicates-hdfs-only',
vector,
use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
- def test_create_table_file_format(self, vector, unique_database):
+ def test_create_table_file_format(self, unique_database):
# When default_file_format query option is not specified, the default
table file
# format is TEXT.
text_table = "{0}.text_tbl".format(unique_database)
@@ -826,7 +831,7 @@ class TestDdlStatements(TestDdlBase):
assert any("ORC" in x for x in result.data)
@SkipIfHive2.acid
- def test_create_table_transactional_type(self, vector, unique_database):
+ def test_create_table_transactional_type(self, unique_database):
# When default_transactional_type query option is not specified, the
transaction
# related table properties are not set.
non_acid_table = "{0}.non_acid_tbl".format(unique_database)
@@ -876,7 +881,7 @@ class TestDdlStatements(TestDdlBase):
assert "transactional" not in props
assert "transactional_properties" not in props
- def test_kudu_column_comment(self, vector, unique_database):
+ def test_kudu_column_comment(self, unique_database):
table = "{0}.kudu_table0".format(unique_database)
self.client.execute("create table {0}(x int comment 'x' primary key) \
stored as kudu".format(table))
@@ -925,6 +930,7 @@ class TestDdlStatements(TestDdlBase):
self.run_test_case('QueryTest/describe-materialized-view', vector,
use_db=unique_database,
multiple_impalad=self._use_multiple_impalad(vector))
+
# IMPALA-10811: RPC to submit query getting stuck for AWS NLB forever
# Test HS2, Beeswax and HS2-HTTP three clients.
class TestAsyncDDL(TestDdlBase):
@@ -942,7 +948,7 @@ class TestAsyncDDL(TestDdlBase):
def test_async_ddl(self, vector, unique_database):
self.run_test_case('QueryTest/async_ddl', vector, use_db=unique_database)
- def test_async_ddl_with_JDBC(self, vector, unique_database):
+ def test_async_ddl_with_JDBC(self, unique_database):
self.exec_with_jdbc("drop table if exists
{0}.test_table".format(unique_database))
self.exec_with_jdbc_and_compare_result(
"create table {0}.test_table(a int)".format(unique_database),
diff --git a/tests/query_test/test_udfs.py b/tests/query_test/test_udfs.py
index a7e93fcbe..31bcb5873 100644
--- a/tests/query_test/test_udfs.py
+++ b/tests/query_test/test_udfs.py
@@ -28,13 +28,13 @@ from tests.common.impala_cluster import ImpalaCluster
from tests.common.impala_test_suite import ImpalaTestSuite
from tests.common.skip import SkipIfLocal
from tests.common.test_dimensions import (
- create_exec_option_dimension,
create_exec_option_dimension_from_dict,
create_uncompressed_text_dimension)
from tests.util.calculation_util import get_random_id
from tests.util.filesystem_utils import get_fs_path, WAREHOUSE
from tests.verifiers.metric_verifier import MetricVerifier
+
class TestUdfBase(ImpalaTestSuite):
"""
Base class with utility functions for testing UDFs.
@@ -109,12 +109,14 @@ returns date intermediate date location '{location}'
init_fn='AggDateIntermediateInit' update_fn='AggDateIntermediateUpdate'
merge_fn='AggDateIntermediateMerge' finalize_fn='AggDateIntermediateFinalize';
-create aggregate function {database}.agg_string_intermediate(decimal(20,10),
bigint, string)
+create aggregate function {database}.agg_string_intermediate(
+ decimal(20,10), bigint, string)
returns decimal(20,0) intermediate string location '{location}'
init_fn='AggStringIntermediateInit' update_fn='AggStringIntermediateUpdate'
merge_fn='AggStringIntermediateMerge'
finalize_fn='AggStringIntermediateFinalize';
-create aggregate function {database}.agg_binary_intermediate(decimal(20,10),
bigint, binary)
+create aggregate function {database}.agg_binary_intermediate(
+ decimal(20,10), bigint, binary)
returns decimal(20,0) intermediate binary location '{location}'
init_fn='AggStringIntermediateInit' update_fn='AggStringIntermediateUpdate'
merge_fn='AggStringIntermediateMerge'
finalize_fn='AggStringIntermediateFinalize';
@@ -234,7 +236,8 @@ create function {database}.count_rows() returns bigint
location '{location}' symbol='Count' prepare_fn='CountPrepare'
close_fn='CountClose';
create function {database}.constant_arg(int) returns int
-location '{location}' symbol='ConstantArg' prepare_fn='ConstantArgPrepare'
close_fn='ConstantArgClose';
+location '{location}' symbol='ConstantArg' prepare_fn='ConstantArgPrepare'
+close_fn='ConstantArgClose';
create function {database}.validate_open(int) returns boolean
location '{location}' symbol='ValidateOpen'
@@ -271,11 +274,12 @@ create function {database}.twenty_args(int, int, int,
int, int, int, int, int, i
int, int, int, int, int, int, int, int, int, int) returns int
location '{location}' symbol='TwentyArgs';
-create function {database}.twenty_one_args(int, int, int, int, int, int, int,
int, int, int,
- int, int, int, int, int, int, int, int, int, int, int) returns int
+create function {database}.twenty_one_args(int, int, int, int, int, int, int,
int, int,
+ int, int, int, int, int, int, int, int, int, int, int, int) returns int
location '{location}' symbol='TwentyOneArgs';
"""
+
class TestUdfExecution(TestUdfBase):
"""Test execution of UDFs with a combination of different query options."""
@classmethod
@@ -286,10 +290,10 @@ class TestUdfExecution(TestUdfBase):
def add_test_dimensions(cls):
super(TestUdfExecution, cls).add_test_dimensions()
cls.ImpalaTestMatrix.add_dimension(
- create_exec_option_dimension_from_dict({"disable_codegen" : [False,
True],
- "disable_codegen_rows_threshold" : [0],
- "exec_single_node_rows_threshold" : [0,100],
- "enable_expr_rewrites" : [False, True]}))
+ create_exec_option_dimension_from_dict({"disable_codegen": [False,
True],
+ "disable_codegen_rows_threshold": [0],
+ "exec_single_node_rows_threshold": [0, 100],
+ "enable_expr_rewrites": [False, True]}))
# There is no reason to run these tests using all dimensions.
cls.ImpalaTestMatrix.add_dimension(
create_uncompressed_text_dimension(cls.get_workload()))
@@ -448,7 +452,7 @@ class TestUdfTargeted(TestUdfBase):
cls.ImpalaTestMatrix.add_dimension(
create_uncompressed_text_dimension(cls.get_workload()))
- def test_udf_invalid_symbol(self, vector, unique_database):
+ def test_udf_invalid_symbol(self, unique_database):
""" IMPALA-1642: Impala crashes if the symbol for a Hive UDF doesn't exist
Invalid symbols are checked at UDF creation time."""
src_udf_path = os.path.join(
@@ -467,7 +471,7 @@ class TestUdfTargeted(TestUdfBase):
ex = self.execute_query_expect_failure(self.client, create_fn_stmt)
assert "ClassNotFoundException" in str(ex)
- def test_hidden_symbol(self, vector, unique_database):
+ def test_hidden_symbol(self, unique_database):
"""Test that symbols in the test UDFs are hidden by default and that
therefore
they cannot be used as a UDF entry point."""
symbol = "_Z16UnexportedSymbolPN10impala_udf15FunctionContextE"
@@ -478,7 +482,7 @@ class TestUdfTargeted(TestUdfBase):
assert "Could not find symbol '{0}'".format(symbol) in str(ex), str(ex)
# IMPALA-8196: IR UDFs ignore whether symbol is hidden or not. Exercise
the current
# behaviour, where the UDF can be created and executed.
- result = self.execute_query_expect_success(self.client, """
+ self.execute_query_expect_success(self.client, """
create function `{0}`.unexported() returns BIGINT LOCATION '{1}'
SYMBOL='{2}'""".format(
unique_database, get_fs_path('/test-warehouse/test-udfs.ll'), symbol))
@@ -528,7 +532,8 @@ class TestUdfTargeted(TestUdfBase):
assert "Failed to get file info" in str(e)
def test_libs_with_same_filenames(self, vector, unique_database):
- self.run_test_case('QueryTest/libs_with_same_filenames', vector,
use_db=unique_database)
+ self.run_test_case('QueryTest/libs_with_same_filenames', vector,
+ use_db=unique_database)
def test_udf_update_via_drop(self, vector, unique_database):
"""Test updating the UDF binary without restarting Impala. Dropping
@@ -638,7 +643,7 @@ class TestUdfTargeted(TestUdfBase):
assert results.success
assert len(results.data) == 9999
- def test_udf_profile(self, vector, unique_database):
+ def test_udf_profile(self, unique_database):
"""Test to validate that explain plans and runtime profiles contain
information about
any custom UDFs used in an Impala query."""
self.client.execute(
@@ -649,13 +654,13 @@ class TestUdfTargeted(TestUdfBase):
"select {0}.hive_substring(string_col, 1),
{0}.hive_substring(string_col, 2) "
"from functional.alltypes limit
10".format(unique_database)).runtime_profile
- assert re.search("output exprs.*hive_substring.*/\* JAVA UDF \*/", profile)
+ assert re.search(r"output exprs.*hive_substring.*/\* JAVA UDF \*/",
profile)
# Ensure that hive_substring only shows up once in the list of UDFs.
assert re.search(
- "User Defined Functions \(UDFs\): {0}\.hive_substring\s*[\r\n]".format(
+ r"User Defined Functions \(UDFs\):
{0}\.hive_substring\s*[\r\n]".format(
unique_database), profile)
- def test_set_fallback_db_for_functions(self, vector, unique_database):
+ def test_set_fallback_db_for_functions(self, unique_database):
"""IMPALA-11728: Set fallback database for functions."""
create_function_stmt = "create function `{0}`.fn() returns int "\
"location '{1}/libTestUdfs.so'
symbol='NoArgs'".format(unique_database,
