ClickHouse/tests/integration/test_merge_tree_s3_with_cache/test.py

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

130 lines
3.9 KiB
Python
Raw Normal View History

2020-07-30 14:49:56 +00:00
import logging
import pytest
from helpers.cluster import ClickHouseCluster
@pytest.fixture(scope="module")
def cluster():
try:
cluster = ClickHouseCluster(__file__)
cluster.add_instance(
"node",
main_configs=[
"configs/config.d/storage_conf.xml",
"configs/config.d/ssl_conf.xml",
"configs/config.d/query_log.xml",
],
user_configs=["configs/config.d/users.xml"],
with_minio=True,
)
2020-07-30 14:49:56 +00:00
logging.info("Starting cluster...")
cluster.start()
logging.info("Cluster started")
yield cluster
finally:
cluster.shutdown()
def get_query_stat(instance, hint):
result = {}
instance.query("SYSTEM FLUSH LOGS")
events = instance.query(
"""
2021-01-21 06:55:13 +00:00
SELECT ProfileEvents.keys, ProfileEvents.values
2020-07-30 14:49:56 +00:00
FROM system.query_log
ARRAY JOIN ProfileEvents
WHERE type != 1 AND query LIKE '%{}%'
""".format(
hint.replace("'", "\\'")
)
).split("\n")
2020-07-30 14:49:56 +00:00
for event in events:
ev = event.split("\t")
if len(ev) == 2:
if ev[0].startswith("S3"):
result[ev[0]] = int(ev[1])
return result
2020-09-14 12:49:04 +00:00
@pytest.mark.parametrize("min_rows_for_wide_part,read_requests", [(0, 2), (8192, 1)])
def test_write_is_cached(cluster, min_rows_for_wide_part, read_requests):
2020-07-30 14:49:56 +00:00
node = cluster.instances["node"]
node.query(
"""
CREATE TABLE s3_test (
id Int64,
data String
) ENGINE=MergeTree()
ORDER BY id
2020-09-14 12:49:04 +00:00
SETTINGS storage_policy='s3', min_rows_for_wide_part={}
""".format(
min_rows_for_wide_part
)
2020-07-30 14:49:56 +00:00
)
2020-07-31 09:24:57 +00:00
node.query("SYSTEM FLUSH LOGS")
2020-07-30 14:49:56 +00:00
node.query("TRUNCATE TABLE system.query_log")
node.query("INSERT INTO s3_test VALUES (0,'data'),(1,'data')")
select_query = "SELECT * FROM s3_test order by id FORMAT Values"
assert node.query(select_query) == "(0,'data'),(1,'data')"
# With async reads profile events are not updated because reads are done in a separate thread.
# stat = get_query_stat(node, select_query)
# assert stat["S3ReadRequestsCount"] == read_requests # Only .bin files should be accessed from S3.
2020-07-30 14:49:56 +00:00
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE IF EXISTS s3_test SYNC")
2020-07-30 14:49:56 +00:00
@pytest.mark.parametrize(
"min_rows_for_wide_part,all_files,bin_files", [(0, 4, 2), (8192, 2, 1)]
)
def test_read_after_cache_is_wiped(
cluster, min_rows_for_wide_part, all_files, bin_files
):
2020-07-30 14:49:56 +00:00
node = cluster.instances["node"]
node.query(
"""
CREATE TABLE s3_test (
id Int64,
data String
) ENGINE=MergeTree()
ORDER BY id
2020-09-14 12:49:04 +00:00
SETTINGS storage_policy='s3', min_rows_for_wide_part={}
""".format(
min_rows_for_wide_part
)
2020-07-30 14:49:56 +00:00
)
2020-07-31 09:24:57 +00:00
node.query("SYSTEM FLUSH LOGS")
2020-07-30 14:49:56 +00:00
node.query("TRUNCATE TABLE system.query_log")
node.query("INSERT INTO s3_test VALUES (0,'data'),(1,'data')")
# Wipe cache
cluster.exec_in_container(
cluster.get_container_id("node"),
["rm", "-rf", "/var/lib/clickhouse/disks/s3/cache/"],
)
2020-07-30 14:49:56 +00:00
select_query = "SELECT * FROM s3_test"
node.query(select_query)
# With async reads profile events are not updated because reads are done in a separate thread.
# stat = get_query_stat(node, select_query)
# assert stat["S3ReadRequestsCount"] == all_files # .mrk and .bin files should be accessed from S3.
2020-07-30 14:49:56 +00:00
# After cache is populated again, only .bin files should be accessed from S3.
select_query = "SELECT * FROM s3_test order by id FORMAT Values"
assert node.query(select_query) == "(0,'data'),(1,'data')"
# With async reads profile events are not updated because reads are done in a separate thread.
# stat = get_query_stat(node, select_query)
# assert stat["S3ReadRequestsCount"] == bin_files
2020-07-30 14:49:56 +00:00
2023-05-03 18:06:46 +00:00
node.query("DROP TABLE IF EXISTS s3_test SYNC")