ClickHouse/tests/integration/test_merge_tree_s3/test.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1033 lines
35 KiB
Python
Raw Normal View History

import logging
import time
import os
import uuid
import pytest
2022-07-07 20:19:15 +00:00
from helpers.cluster import ClickHouseCluster
from helpers.mock_servers import start_s3_mock, start_mock_servers
from helpers.utility import generate_values, replace_config, SafeThread
2022-09-18 21:47:43 +00:00
from helpers.wait_for_helpers import wait_for_delete_inactive_parts
from helpers.wait_for_helpers import wait_for_delete_empty_parts
2023-05-03 01:05:38 +00:00
from helpers.wait_for_helpers import wait_for_merges
SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__))
@pytest.fixture(scope="module")
def cluster():
try:
cluster = ClickHouseCluster(__file__)
cluster.add_instance(
"node",
main_configs=[
"configs/config.xml",
"configs/config.d/storage_conf.xml",
"configs/config.d/bg_processing_pool_conf.xml",
"configs/config.d/blob_log.xml",
],
2023-05-21 10:42:28 +00:00
user_configs=[
"configs/config.d/users.xml",
],
stay_alive=True,
with_minio=True,
)
cluster.add_instance(
"node_with_limited_disk",
main_configs=[
"configs/config.d/storage_conf.xml",
"configs/config.d/bg_processing_pool_conf.xml",
"configs/config.d/blob_log.xml",
],
with_minio=True,
tmpfs=[
"/jbod1:size=2M",
],
)
2022-09-14 18:29:36 +00:00
logging.info("Starting cluster...")
cluster.start()
logging.info("Cluster started")
2021-08-04 10:27:45 +00:00
run_s3_mocks(cluster)
yield cluster
finally:
cluster.shutdown()
FILES_OVERHEAD = 1
FILES_OVERHEAD_PER_COLUMN = 2 # Data and mark files
FILES_OVERHEAD_DEFAULT_COMPRESSION_CODEC = 1
FILES_OVERHEAD_METADATA_VERSION = 1
FILES_OVERHEAD_PER_PART_WIDE = (
FILES_OVERHEAD_PER_COLUMN * 3
+ 2
+ 6
+ FILES_OVERHEAD_DEFAULT_COMPRESSION_CODEC
+ FILES_OVERHEAD_METADATA_VERSION
)
FILES_OVERHEAD_PER_PART_COMPACT = (
10 + FILES_OVERHEAD_DEFAULT_COMPRESSION_CODEC + FILES_OVERHEAD_METADATA_VERSION
)
2021-09-20 18:18:08 +00:00
def create_table(node, table_name, **additional_settings):
2021-08-04 10:27:45 +00:00
settings = {
"storage_policy": "s3",
"old_parts_lifetime": 0,
"index_granularity": 512,
"temporary_directories_lifetime": 1,
2021-08-04 10:27:45 +00:00
}
settings.update(additional_settings)
create_table_statement = f"""
CREATE TABLE {table_name} (
dt Date,
id Int64,
data String,
INDEX min_max (id) TYPE minmax GRANULARITY 3
) ENGINE=MergeTree()
PARTITION BY dt
ORDER BY (dt, id)
2021-08-04 10:27:45 +00:00
SETTINGS {",".join((k+"="+repr(v) for k, v in settings.items()))}"""
2021-09-20 18:18:08 +00:00
node.query(f"DROP TABLE IF EXISTS {table_name}")
node.query(create_table_statement)
@pytest.fixture(scope="module")
def init_broken_s3(cluster):
yield start_s3_mock(cluster, "broken_s3", "8083")
@pytest.fixture(scope="function")
def broken_s3(init_broken_s3):
init_broken_s3.reset()
yield init_broken_s3
2021-08-04 10:27:45 +00:00
def run_s3_mocks(cluster):
script_dir = os.path.join(os.path.dirname(__file__), "s3_mocks")
start_mock_servers(
cluster,
script_dir,
[
("unstable_proxy.py", "resolver", "8081"),
("no_delete_objects.py", "resolver", "8082"),
],
)
2021-08-04 10:27:45 +00:00
def list_objects(cluster, path="data/", hint="list_objects"):
minio = cluster.minio_client
objects = list(minio.list_objects(cluster.minio_bucket, path, recursive=True))
logging.info(f"{hint} ({len(objects)}): {[x.object_name for x in objects]}")
return objects
2021-04-09 19:30:42 +00:00
def wait_for_delete_s3_objects(cluster, expected, timeout=30):
while timeout > 0:
existing_objects = list_objects(cluster, "data/")
if len(existing_objects) == expected:
return existing_objects
2021-04-09 19:30:42 +00:00
timeout -= 1
time.sleep(1)
existing_objects = list_objects(cluster, "data/")
assert len(existing_objects) == expected
return existing_objects
2021-04-09 19:30:42 +00:00
2023-04-26 12:24:35 +00:00
def remove_all_s3_objects(cluster):
minio = cluster.minio_client
objects_to_delete = list_objects(cluster, "data/")
for obj in objects_to_delete:
2023-04-26 12:24:35 +00:00
minio.remove_object(cluster.minio_bucket, obj.object_name)
return objects_to_delete
2023-04-26 12:24:35 +00:00
@pytest.fixture(autouse=True, scope="function")
def clear_minio(cluster):
2023-04-26 12:24:35 +00:00
try:
# CH do some writes to the S3 at start. For example, file data/clickhouse_access_check_{server_uuid}.
# Set the timeout there as 10 sec in order to resolve the race with that file exists.
wait_for_delete_s3_objects(cluster, 0, timeout=10)
except:
# Remove extra objects to prevent tests cascade failing
remove_all_s3_objects(cluster)
yield
2021-04-09 19:30:42 +00:00
2023-04-26 12:24:35 +00:00
def check_no_objects_after_drop(cluster, table_name="s3_test", node_name="node"):
node = cluster.instances[node_name]
2023-05-03 18:06:46 +00:00
node.query(f"DROP TABLE IF EXISTS {table_name} SYNC")
return wait_for_delete_s3_objects(cluster, 0, timeout=0)
@pytest.mark.parametrize(
2021-09-20 18:18:08 +00:00
"min_rows_for_wide_part,files_per_part,node_name",
[
2021-09-20 18:18:08 +00:00
(0, FILES_OVERHEAD_PER_PART_WIDE, "node"),
(8192, FILES_OVERHEAD_PER_PART_COMPACT, "node"),
],
)
2021-09-20 18:18:08 +00:00
def test_simple_insert_select(
cluster, min_rows_for_wide_part, files_per_part, node_name
):
node = cluster.instances[node_name]
create_table(node, "s3_test", min_rows_for_wide_part=min_rows_for_wide_part)
minio = cluster.minio_client
values1 = generate_values("2020-01-03", 4096)
insert_query_id = uuid.uuid4().hex
node.query(
"INSERT INTO s3_test VALUES {}".format(values1), query_id=insert_query_id
)
assert node.query("SELECT * FROM s3_test order by dt, id FORMAT Values") == values1
assert len(list_objects(cluster, "data/")) == FILES_OVERHEAD + files_per_part
node.query("SYSTEM FLUSH LOGS")
blob_storage_log = node.query(
f"SELECT * FROM system.blob_storage_log WHERE query_id = '{insert_query_id}' FORMAT PrettyCompactMonoBlock"
)
result = node.query(
f"""SELECT
(countIf( (event_type == 'Upload' OR event_type == 'MultiPartUploadWrite') as event_match) as total_events) > 0,
countIf(event_match AND bucket == 'root') == total_events,
countIf(event_match AND remote_path != '') == total_events,
countIf(event_match AND local_path != '') == total_events,
sumIf(data_size, event_match) > 0
FROM system.blob_storage_log
WHERE query_id = '{insert_query_id}' AND error == ''
"""
)
assert result == "1\t1\t1\t1\t1\n", blob_storage_log
values2 = generate_values("2020-01-04", 4096)
node.query("INSERT INTO s3_test VALUES {}".format(values2))
assert (
node.query("SELECT * FROM s3_test ORDER BY dt, id FORMAT Values")
== values1 + "," + values2
)
assert len(list_objects(cluster, "data/")) == FILES_OVERHEAD + files_per_part * 2
2021-09-21 12:40:48 +00:00
assert (
node.query("SELECT count(*) FROM s3_test where id = 1 FORMAT Values") == "(2)"
)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
2021-09-20 18:18:08 +00:00
@pytest.mark.parametrize("merge_vertical,node_name", [(True, "node"), (False, "node")])
def test_insert_same_partition_and_merge(cluster, merge_vertical, node_name):
2021-08-04 10:27:45 +00:00
settings = {}
if merge_vertical:
2021-08-04 10:27:45 +00:00
settings["vertical_merge_algorithm_min_rows_to_activate"] = 0
settings["vertical_merge_algorithm_min_columns_to_activate"] = 0
2021-09-20 18:18:08 +00:00
node = cluster.instances[node_name]
create_table(node, "s3_test", **settings)
node.query("SYSTEM STOP MERGES s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 1024))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 2048))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 1024, -1))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 2048, -1))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096, -1))
)
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert (
node.query("SELECT count(distinct(id)) FROM s3_test FORMAT Values") == "(8192)"
)
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD_PER_PART_WIDE * 6 + FILES_OVERHEAD
)
node.query("SYSTEM START MERGES s3_test")
# Wait for merges and old parts deletion
for attempt in range(0, 60):
parts_count = node.query(
2022-05-11 22:04:54 +00:00
"SELECT COUNT(*) FROM system.parts WHERE table = 's3_test' and active = 1 FORMAT Values"
)
2022-05-11 22:04:54 +00:00
if parts_count == "(1)":
break
if attempt == 59:
assert parts_count == "(1)"
time.sleep(1)
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert (
node.query("SELECT count(distinct(id)) FROM s3_test FORMAT Values") == "(8192)"
)
2022-05-14 11:46:47 +00:00
wait_for_delete_s3_objects(
cluster, FILES_OVERHEAD_PER_PART_WIDE + FILES_OVERHEAD, timeout=45
)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
@pytest.mark.parametrize("node_name", ["node"])
2021-09-20 18:18:08 +00:00
def test_alter_table_columns(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096, -1))
)
def assert_deleted_in_log(old_objects, new_objects):
node.query("SYSTEM FLUSH LOGS")
deleted_objects = set(obj.object_name for obj in old_objects) - set(
obj.object_name for obj in new_objects
)
deleted_in_log = set(
node.query(
f"SELECT remote_path FROM system.blob_storage_log WHERE error == '' AND event_type == 'Delete'"
)
.strip()
.split()
)
# all deleted objects should be in log
assert all(obj in deleted_in_log for obj in deleted_objects), (
deleted_objects,
node.query(
f"SELECT * FROM system.blob_storage_log FORMAT PrettyCompactMonoBlock"
),
)
objects_before = list_objects(cluster, "data/")
node.query("ALTER TABLE s3_test ADD COLUMN col1 UInt64 DEFAULT 1")
# To ensure parts have merged
node.query("OPTIMIZE TABLE s3_test")
assert node.query("SELECT sum(col1) FROM s3_test FORMAT Values") == "(8192)"
assert (
node.query("SELECT sum(col1) FROM s3_test WHERE id > 0 FORMAT Values")
== "(4096)"
2021-04-09 19:30:42 +00:00
)
existing_objects = wait_for_delete_s3_objects(
2021-04-09 19:30:42 +00:00
cluster,
FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE + FILES_OVERHEAD_PER_COLUMN,
)
assert_deleted_in_log(objects_before, existing_objects)
objects_before = existing_objects
2020-03-21 08:10:44 +00:00
node.query(
"ALTER TABLE s3_test MODIFY COLUMN col1 String", settings={"mutations_sync": 2}
)
assert node.query("SELECT distinct(col1) FROM s3_test FORMAT Values") == "('1')"
2020-03-21 08:10:44 +00:00
# and file with mutation
existing_objects = wait_for_delete_s3_objects(
2021-04-09 19:30:42 +00:00
cluster,
FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE + FILES_OVERHEAD_PER_COLUMN + 1,
)
2020-03-21 08:10:44 +00:00
assert_deleted_in_log(objects_before, existing_objects)
objects_before = existing_objects
2020-03-21 08:10:44 +00:00
node.query("ALTER TABLE s3_test DROP COLUMN col1", settings={"mutations_sync": 2})
# and 2 files with mutations
existing_objects = wait_for_delete_s3_objects(
2021-04-09 19:30:42 +00:00
cluster, FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE + 2
)
assert_deleted_in_log(objects_before, existing_objects)
objects_before = existing_objects
existing_objects = check_no_objects_after_drop(cluster)
assert_deleted_in_log(objects_before, existing_objects)
objects_before = existing_objects
2023-04-26 12:24:35 +00:00
@pytest.mark.parametrize("node_name", ["node"])
2021-09-20 18:18:08 +00:00
def test_attach_detach_partition(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
node.query("ALTER TABLE s3_test DETACH PARTITION '2020-01-03'")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(4096)"
assert (
len(list_objects(cluster, "data/"))
2023-04-30 17:27:15 +00:00
== FILES_OVERHEAD
+ FILES_OVERHEAD_PER_PART_WIDE * 2
- FILES_OVERHEAD_METADATA_VERSION
)
node.query("ALTER TABLE s3_test ATTACH PARTITION '2020-01-03'")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
2023-06-29 13:26:21 +00:00
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
node.query("ALTER TABLE s3_test DROP PARTITION '2020-01-03'")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(4096)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 1
)
node.query("ALTER TABLE s3_test DETACH PARTITION '2020-01-04'")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(0)"
assert (
len(list_objects(cluster, "data/"))
2023-04-30 17:27:15 +00:00
== FILES_OVERHEAD
+ FILES_OVERHEAD_PER_PART_WIDE * 1
- FILES_OVERHEAD_METADATA_VERSION
)
node.query(
"ALTER TABLE s3_test DROP DETACHED PARTITION '2020-01-04'",
settings={"allow_drop_detached": 1},
)
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(0)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 0
)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
@pytest.mark.parametrize("node_name", ["node"])
2021-09-20 18:18:08 +00:00
def test_move_partition_to_another_disk(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
node.query("ALTER TABLE s3_test MOVE PARTITION '2020-01-04' TO DISK 'hdd'")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE
)
node.query("ALTER TABLE s3_test MOVE PARTITION '2020-01-04' TO DISK 's3'")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
2021-09-20 18:18:08 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_table_manipulations(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
node.query("RENAME TABLE s3_test TO s3_renamed")
assert node.query("SELECT count(*) FROM s3_renamed FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
node.query("RENAME TABLE s3_renamed TO s3_test")
assert node.query("CHECK TABLE s3_test FORMAT Values") == "(1)"
node.query("DETACH TABLE s3_test")
node.query("ATTACH TABLE s3_test")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert (
len(list_objects(cluster, "data/"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 2
)
node.query("TRUNCATE TABLE s3_test")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(0)"
assert len(list_objects(cluster, "data/")) == FILES_OVERHEAD
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
@pytest.mark.parametrize("node_name", ["node"])
2021-09-20 18:18:08 +00:00
def test_move_replace_partition_to_another_table(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-05", 4096, -1))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-06", 4096, -1))
)
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(16384)"
2023-04-30 16:43:56 +00:00
assert (
len(list_objects(cluster, "data/", "Objects at start"))
== FILES_OVERHEAD + FILES_OVERHEAD_PER_PART_WIDE * 4
)
2021-09-20 18:18:08 +00:00
create_table(node, "s3_clone")
node.query("ALTER TABLE s3_test MOVE PARTITION '2020-01-03' TO TABLE s3_clone")
node.query("ALTER TABLE s3_test MOVE PARTITION '2020-01-05' TO TABLE s3_clone")
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(8192)"
assert node.query("SELECT sum(id) FROM s3_clone FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_clone FORMAT Values") == "(8192)"
2023-04-30 16:43:56 +00:00
list_objects(cluster, "data/", "Object after move partition")
# Number of objects in S3 should be unchanged.
2023-05-01 11:58:10 +00:00
wait_for_delete_s3_objects(
2023-05-01 12:02:07 +00:00
cluster,
FILES_OVERHEAD * 2
2023-04-30 16:43:56 +00:00
+ FILES_OVERHEAD_PER_PART_WIDE * 4
2023-05-01 12:02:07 +00:00
- FILES_OVERHEAD_METADATA_VERSION * 2,
)
# Add new partitions to source table, but with different values and replace them from copied table.
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096, -1))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-05", 4096))
)
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(16384)"
2023-04-30 16:43:56 +00:00
list_objects(cluster, "data/", "Object after insert")
2023-05-01 11:58:10 +00:00
wait_for_delete_s3_objects(
2023-05-01 12:02:07 +00:00
cluster,
FILES_OVERHEAD * 2
2023-04-30 16:43:56 +00:00
+ FILES_OVERHEAD_PER_PART_WIDE * 6
2023-05-01 12:02:07 +00:00
- FILES_OVERHEAD_METADATA_VERSION * 2,
)
node.query("ALTER TABLE s3_test REPLACE PARTITION '2020-01-03' FROM s3_clone")
node.query("ALTER TABLE s3_test REPLACE PARTITION '2020-01-05' FROM s3_clone")
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(16384)"
assert node.query("SELECT sum(id) FROM s3_clone FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_clone FORMAT Values") == "(8192)"
# Wait for outdated partitions deletion.
2021-04-09 19:30:42 +00:00
wait_for_delete_s3_objects(
2023-04-30 17:27:15 +00:00
cluster,
FILES_OVERHEAD * 2
+ FILES_OVERHEAD_PER_PART_WIDE * 4
- FILES_OVERHEAD_METADATA_VERSION * 2,
2021-04-09 19:30:42 +00:00
)
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE s3_clone SYNC")
assert node.query("SELECT sum(id) FROM s3_test FORMAT Values") == "(0)"
assert node.query("SELECT count(*) FROM s3_test FORMAT Values") == "(16384)"
2023-05-01 11:58:10 +00:00
list_objects(cluster, "data/", "Object after drop")
2023-05-01 11:58:10 +00:00
wait_for_delete_s3_objects(
2023-05-01 12:02:07 +00:00
cluster,
FILES_OVERHEAD
2023-04-30 16:43:56 +00:00
+ FILES_OVERHEAD_PER_PART_WIDE * 4
2023-05-01 12:02:07 +00:00
- FILES_OVERHEAD_METADATA_VERSION * 2,
)
node.query("ALTER TABLE s3_test FREEZE")
# Number S3 objects should be unchanged.
list_objects(cluster, "data/", "Object after freeze")
2023-05-01 11:58:10 +00:00
wait_for_delete_s3_objects(
cluster,
FILES_OVERHEAD
2023-04-30 16:43:56 +00:00
+ FILES_OVERHEAD_PER_PART_WIDE * 4
2023-05-01 12:02:07 +00:00
- FILES_OVERHEAD_METADATA_VERSION * 2,
)
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE s3_test SYNC")
# Backup data should remain in S3.
2021-04-09 19:30:42 +00:00
2023-04-30 17:27:15 +00:00
wait_for_delete_s3_objects(
cluster, FILES_OVERHEAD_PER_PART_WIDE * 4 - FILES_OVERHEAD_METADATA_VERSION * 4
)
2023-04-26 12:24:35 +00:00
remove_all_s3_objects(cluster)
2021-02-24 14:26:46 +00:00
2021-09-20 18:18:08 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_freeze_unfreeze(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
2021-02-24 14:26:46 +00:00
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
2021-02-24 14:26:46 +00:00
node.query("ALTER TABLE s3_test FREEZE WITH NAME 'backup1'")
2021-03-02 21:10:09 +00:00
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
2021-03-02 21:10:09 +00:00
node.query("ALTER TABLE s3_test FREEZE WITH NAME 'backup2'")
2021-02-24 14:26:46 +00:00
node.query("TRUNCATE TABLE s3_test")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
2021-02-24 14:26:46 +00:00
assert (
len(list_objects(cluster, "data/"))
2023-04-30 17:27:15 +00:00
== FILES_OVERHEAD
+ (FILES_OVERHEAD_PER_PART_WIDE - FILES_OVERHEAD_METADATA_VERSION) * 2
)
2021-03-02 21:10:09 +00:00
# Unfreeze single partition from backup1.
node.query(
"ALTER TABLE s3_test UNFREEZE PARTITION '2020-01-03' WITH NAME 'backup1'"
)
# Unfreeze all partitions from backup2.
node.query("ALTER TABLE s3_test UNFREEZE WITH NAME 'backup2'")
2021-02-24 14:26:46 +00:00
2023-04-26 12:24:35 +00:00
# Data should be removed from S3.
wait_for_delete_s3_objects(cluster, FILES_OVERHEAD)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
2022-06-08 12:09:59 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_freeze_system_unfreeze(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
create_table(node, "s3_test_removed")
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096))
)
node.query("ALTER TABLE s3_test FREEZE WITH NAME 'backup3'")
node.query("ALTER TABLE s3_test_removed FREEZE WITH NAME 'backup3'")
node.query("TRUNCATE TABLE s3_test")
wait_for_delete_empty_parts(node, "s3_test")
wait_for_delete_inactive_parts(node, "s3_test")
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE s3_test_removed SYNC")
2022-06-08 12:09:59 +00:00
assert (
len(list_objects(cluster, "data/"))
2023-04-30 17:27:15 +00:00
== FILES_OVERHEAD
+ (FILES_OVERHEAD_PER_PART_WIDE - FILES_OVERHEAD_METADATA_VERSION) * 2
2022-06-08 12:09:59 +00:00
)
# Unfreeze all data from backup3.
node.query("SYSTEM UNFREEZE WITH NAME 'backup3'")
2023-04-26 12:24:35 +00:00
# Data should be removed from S3.
wait_for_delete_s3_objects(cluster, FILES_OVERHEAD)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
2022-06-08 12:09:59 +00:00
2021-09-20 18:18:08 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_s3_disk_apply_new_settings(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test")
2022-07-07 20:19:15 +00:00
config_path = os.path.join(
SCRIPT_DIR,
"./{}/node/configs/config.d/storage_conf.xml".format(
cluster.instances_dir_name
),
)
def get_s3_requests():
node.query("SYSTEM FLUSH LOGS")
return int(
node.query(
"SELECT value FROM system.events WHERE event='S3WriteRequestsCount'"
)
)
s3_requests_before = get_s3_requests()
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-03", 4096))
)
s3_requests_to_write_partition = get_s3_requests() - s3_requests_before
# Force multi-part upload mode.
replace_config(
2022-07-07 20:19:15 +00:00
config_path,
"<s3_max_single_part_upload_size>33554432</s3_max_single_part_upload_size>",
"<s3_max_single_part_upload_size>0</s3_max_single_part_upload_size>",
)
node.query("SYSTEM RELOAD CONFIG")
s3_requests_before = get_s3_requests()
node.query(
"INSERT INTO s3_test VALUES {}".format(generate_values("2020-01-04", 4096, -1))
)
# There should be 3 times more S3 requests because multi-part upload mode uses 3 requests to upload object.
assert get_s3_requests() - s3_requests_before == s3_requests_to_write_partition * 3
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
@pytest.mark.parametrize("node_name", ["node"])
def test_s3_no_delete_objects(cluster, node_name):
node = cluster.instances[node_name]
2022-07-11 14:53:05 +00:00
create_table(
node, "s3_test_no_delete_objects", storage_policy="no_delete_objects_s3"
)
2022-07-11 14:53:05 +00:00
node.query("DROP TABLE s3_test_no_delete_objects SYNC")
2023-04-26 12:24:35 +00:00
remove_all_s3_objects(cluster)
@pytest.mark.parametrize("node_name", ["node"])
2021-09-20 18:18:08 +00:00
def test_s3_disk_reads_on_unstable_connection(cluster, node_name):
node = cluster.instances[node_name]
create_table(node, "s3_test", storage_policy="unstable_s3")
2021-08-04 10:27:45 +00:00
node.query(
"INSERT INTO s3_test SELECT today(), *, toString(*) FROM system.numbers LIMIT 9000000"
)
for i in range(30):
print(f"Read sequence {i}")
assert node.query("SELECT sum(id) FROM s3_test").splitlines() == [
"40499995500000"
]
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
2021-12-16 21:29:25 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_lazy_seek_optimization_for_async_read(cluster, node_name):
node = cluster.instances[node_name]
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE IF EXISTS s3_test SYNC")
2021-12-19 18:24:52 +00:00
node.query(
"CREATE TABLE s3_test (key UInt32, value String) Engine=MergeTree() ORDER BY key SETTINGS storage_policy='s3';"
)
node.query("SYSTEM STOP MERGES s3_test")
2021-12-20 14:17:02 +00:00
node.query(
"INSERT INTO s3_test SELECT * FROM generateRandom('key UInt32, value String') LIMIT 10000000"
)
2021-12-19 18:24:52 +00:00
node.query("SELECT * FROM s3_test WHERE value LIKE '%abc%' ORDER BY value LIMIT 10")
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster)
@pytest.mark.parametrize("node_name", ["node_with_limited_disk"])
def test_cache_with_full_disk_space(cluster, node_name):
node = cluster.instances[node_name]
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE IF EXISTS s3_test SYNC")
node.query(
2023-04-11 10:32:55 +00:00
"CREATE TABLE s3_test (key UInt32, value String) Engine=MergeTree() ORDER BY value SETTINGS storage_policy='s3_with_cache_and_jbod';"
)
node.query("SYSTEM STOP MERGES s3_test")
node.query(
2023-04-11 10:32:55 +00:00
"INSERT INTO s3_test SELECT number, toString(number) FROM numbers(100000000)"
)
2023-04-11 10:32:55 +00:00
out = node.exec_in_container(
[
"/usr/bin/clickhouse",
"benchmark",
"--iterations",
"10",
"--max_threads",
"100",
"--query",
"SELECT count() FROM s3_test WHERE key < 40000000 or key > 80000000 SETTINGS max_read_buffer_size='44Ki'",
]
)
assert node.contains_in_log(
2022-07-13 09:48:39 +00:00
"Insert into cache is skipped due to insufficient disk space"
)
2023-04-26 12:24:35 +00:00
check_no_objects_after_drop(cluster, node_name=node_name)
2023-05-03 01:05:38 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_merge_canceled_by_drop(cluster, node_name):
node = cluster.instances[node_name]
node.query("DROP TABLE IF EXISTS test_merge_canceled_by_drop NO DELAY")
node.query(
"CREATE TABLE test_merge_canceled_by_drop "
" (key UInt32, value String)"
" Engine=MergeTree() "
" ORDER BY value "
" SETTINGS storage_policy='s3'"
)
node.query("SYSTEM STOP MERGES test_merge_canceled_by_drop")
node.query(
"INSERT INTO test_merge_canceled_by_drop SELECT number, toString(number) FROM numbers(100000000)"
)
node.query("SYSTEM START MERGES test_merge_canceled_by_drop")
wait_for_merges(node, "test_merge_canceled_by_drop")
check_no_objects_after_drop(
cluster, table_name="test_merge_canceled_by_drop", node_name=node_name
)
2023-05-26 23:02:48 +00:00
@pytest.mark.parametrize("storage_policy", ["broken_s3_always_multi_part", "broken_s3"])
2023-05-03 01:05:38 +00:00
@pytest.mark.parametrize("node_name", ["node"])
def test_merge_canceled_by_s3_errors(cluster, broken_s3, node_name, storage_policy):
2023-05-03 01:05:38 +00:00
node = cluster.instances[node_name]
node.query("DROP TABLE IF EXISTS test_merge_canceled_by_s3_errors NO DELAY")
node.query(
"CREATE TABLE test_merge_canceled_by_s3_errors "
" (key UInt32, value String)"
" Engine=MergeTree() "
" ORDER BY value "
2023-05-26 23:02:48 +00:00
f" SETTINGS storage_policy='{storage_policy}'"
2023-05-03 01:05:38 +00:00
)
node.query("SYSTEM STOP MERGES test_merge_canceled_by_s3_errors")
node.query(
"INSERT INTO test_merge_canceled_by_s3_errors SELECT number, toString(number) FROM numbers(10000)"
)
node.query(
"INSERT INTO test_merge_canceled_by_s3_errors SELECT 2*number, toString(number) FROM numbers(10000)"
)
min_key = node.query("SELECT min(key) FROM test_merge_canceled_by_s3_errors")
assert int(min_key) == 0, min_key
broken_s3.setup_at_object_upload()
2023-06-03 18:59:04 +00:00
broken_s3.setup_fake_multpartuploads()
broken_s3.setup_at_part_upload()
2023-05-03 01:05:38 +00:00
node.query("SYSTEM START MERGES test_merge_canceled_by_s3_errors")
error = node.query_and_get_error(
"OPTIMIZE TABLE test_merge_canceled_by_s3_errors FINAL",
)
2024-05-23 11:53:36 +00:00
assert "ExpectedError Message: mock s3 injected unretryable error" in error, error
2023-05-03 01:05:38 +00:00
2024-05-23 11:53:36 +00:00
node.wait_for_log_line("ExpectedError Message: mock s3 injected unretryable error")
2023-05-03 01:05:38 +00:00
table_uuid = node.query(
"SELECT uuid FROM system.tables WHERE database = 'default' AND name = 'test_merge_canceled_by_s3_errors' LIMIT 1"
).strip()
node.query("SYSTEM FLUSH LOGS")
error_count_in_blob_log = node.query(
2024-05-23 11:53:36 +00:00
f"SELECT count() FROM system.blob_storage_log WHERE query_id like '{table_uuid}::%' AND error like '%mock s3 injected unretryable error%'"
).strip()
assert int(error_count_in_blob_log) > 0, node.query(
f"SELECT * FROM system.blob_storage_log WHERE query_id like '{table_uuid}::%' FORMAT PrettyCompactMonoBlock"
)
2023-05-03 01:05:38 +00:00
check_no_objects_after_drop(
cluster, table_name="test_merge_canceled_by_s3_errors", node_name=node_name
)
@pytest.mark.parametrize("node_name", ["node"])
def test_merge_canceled_by_s3_errors_when_move(cluster, broken_s3, node_name):
2023-05-03 01:05:38 +00:00
node = cluster.instances[node_name]
settings = {
"storage_policy": "external_broken_s3",
"merge_with_ttl_timeout": 1,
}
create_table(node, "merge_canceled_by_s3_errors_when_move", **settings)
node.query("SYSTEM STOP MERGES merge_canceled_by_s3_errors_when_move")
node.query(
"INSERT INTO merge_canceled_by_s3_errors_when_move"
" VALUES {}".format(generate_values("2020-01-03", 1000))
)
node.query(
"INSERT INTO merge_canceled_by_s3_errors_when_move"
" VALUES {}".format(generate_values("2020-01-03", 1000, -1))
)
node.query(
"ALTER TABLE merge_canceled_by_s3_errors_when_move"
" MODIFY TTL"
" dt + INTERVAL 1 DAY "
" TO VOLUME 'external'",
settings={"materialize_ttl_after_modify": 0},
)
broken_s3.setup_at_object_upload(count=1, after=1)
2023-05-03 01:05:38 +00:00
node.query("SYSTEM START MERGES merge_canceled_by_s3_errors_when_move")
node.query("OPTIMIZE TABLE merge_canceled_by_s3_errors_when_move FINAL")
2024-05-23 11:53:36 +00:00
node.wait_for_log_line("ExpectedError Message: mock s3 injected unretryable error")
2023-05-03 01:05:38 +00:00
count = node.query("SELECT count() FROM merge_canceled_by_s3_errors_when_move")
assert int(count) == 2000, count
check_no_objects_after_drop(
cluster, table_name="merge_canceled_by_s3_errors_when_move", node_name=node_name
)
2023-05-21 10:42:28 +00:00
@pytest.mark.parametrize("node_name", ["node"])
@pytest.mark.parametrize(
"in_flight_memory", [(10, 245918115), (5, 156786752), (1, 106426187)]
)
def test_s3_engine_heavy_write_check_mem(
cluster, broken_s3, node_name, in_flight_memory
):
2023-11-21 13:03:49 +00:00
pytest.skip(
"Disabled, will be fixed after https://github.com/ClickHouse/ClickHouse/issues/51152"
)
2023-11-20 23:08:20 +00:00
2023-05-21 10:42:28 +00:00
in_flight = in_flight_memory[0]
memory = in_flight_memory[1]
2023-05-21 10:42:28 +00:00
node = cluster.instances[node_name]
2023-08-07 10:57:41 +00:00
# it's bad idea to test something related to memory with sanitizers
if node.is_built_with_sanitizer():
pytest.skip("Disabled for sanitizers")
2023-05-21 10:42:28 +00:00
node.query("DROP TABLE IF EXISTS s3_test SYNC")
node.query(
"CREATE TABLE s3_test"
" ("
" key UInt32 CODEC(NONE), value String CODEC(NONE)"
" )"
" ENGINE S3('http://resolver:8083/root/data/test-upload.csv', 'minio', 'minio123', 'CSV')",
)
2023-06-03 18:59:04 +00:00
broken_s3.setup_fake_multpartuploads()
2023-11-20 23:08:20 +00:00
slow_responces = 10
slow_timeout = 15
2023-11-21 13:03:49 +00:00
broken_s3.setup_slow_answers(
10 * 1024 * 1024, timeout=slow_timeout, count=slow_responces
)
2023-05-21 10:42:28 +00:00
query_id = f"INSERT_INTO_S3_ENGINE_QUERY_ID_{in_flight}"
node.query(
"INSERT INTO s3_test SELECT number, toString(number) FROM numbers(50000000)"
2023-11-20 23:08:20 +00:00
f" SETTINGS "
f" max_memory_usage={2*memory}"
2024-01-31 22:35:30 +00:00
", max_threads=1, optimize_trivial_insert_select=1" # ParallelFormattingOutputFormat consumption depends on it
2023-05-21 10:42:28 +00:00
f", s3_max_inflight_parts_for_one_file={in_flight}",
query_id=query_id,
)
2023-05-21 10:42:28 +00:00
node.query("SYSTEM FLUSH LOGS")
memory_usage, wait_inflight = node.query(
"SELECT memory_usage, ProfileEvents['WriteBufferFromS3WaitInflightLimitMicroseconds']"
2023-05-21 10:42:28 +00:00
" FROM system.query_log"
f" WHERE query_id='{query_id}'"
" AND type!='QueryStart'"
).split()
2023-05-28 22:37:45 +00:00
assert int(memory_usage) < 1.2 * memory
assert int(memory_usage) > 0.8 * memory
2023-11-20 23:08:20 +00:00
# The more in_flight value is the less time CH waits.
assert int(wait_inflight) / 1000 / 1000 > slow_responces * slow_timeout / in_flight
2023-05-21 10:42:28 +00:00
check_no_objects_after_drop(cluster, node_name=node_name)
@pytest.mark.parametrize("node_name", ["node"])
def test_s3_disk_heavy_write_check_mem(cluster, broken_s3, node_name):
2023-05-21 10:42:28 +00:00
memory = 2279055040
node = cluster.instances[node_name]
node.query("DROP TABLE IF EXISTS s3_test SYNC")
node.query(
"CREATE TABLE s3_test"
" ("
2023-05-21 10:42:28 +00:00
" key UInt32, value String"
" )"
2023-05-21 10:42:28 +00:00
" ENGINE=MergeTree()"
" ORDER BY key"
" SETTINGS"
" storage_policy='broken_s3'",
)
2023-05-21 10:42:28 +00:00
node.query("SYSTEM STOP MERGES s3_test")
2023-06-03 18:59:04 +00:00
broken_s3.setup_fake_multpartuploads()
broken_s3.setup_slow_answers(10 * 1024 * 1024, timeout=10, count=50)
2023-05-21 10:42:28 +00:00
query_id = f"INSERT_INTO_S3_DISK_QUERY_ID"
node.query(
"INSERT INTO s3_test SELECT number, toString(number) FROM numbers(50000000)"
2023-05-21 10:42:28 +00:00
f" SETTINGS max_memory_usage={2*memory}"
2024-01-31 22:35:30 +00:00
", max_insert_block_size=50000000"
", min_insert_block_size_rows=50000000"
", min_insert_block_size_bytes=1000000000000"
", optimize_trivial_insert_select=1",
query_id=query_id,
)
node.query("SYSTEM FLUSH LOGS")
result = node.query(
"SELECT memory_usage"
" FROM system.query_log"
f" WHERE query_id='{query_id}'"
" AND type!='QueryStart'"
)
2023-05-28 22:37:45 +00:00
assert int(result) < 1.2 * memory
assert int(result) > 0.8 * memory
check_no_objects_after_drop(cluster, node_name=node_name)