mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-11-24 08:32:02 +00:00
97f2a2213e
* Move some code outside dbms/src folder * Fix paths
98 lines
4.0 KiB
Python
98 lines
4.0 KiB
Python
import time
|
|
import pytest
|
|
|
|
from helpers.cluster import ClickHouseCluster
|
|
from multiprocessing.dummy import Pool
|
|
from helpers.client import QueryRuntimeException, QueryTimeoutExceedException
|
|
|
|
from helpers.test_tools import assert_eq_with_retry
|
|
|
|
cluster = ClickHouseCluster(__file__)
|
|
|
|
node1 = cluster.add_instance('node1', with_zookeeper=True)
|
|
node2 = cluster.add_instance('node2', with_zookeeper=True)
|
|
node3 = cluster.add_instance('node3', with_zookeeper=True, image='yandex/clickhouse-server:19.1.14', with_installed_binary=True)
|
|
|
|
@pytest.fixture(scope="module")
|
|
def start_cluster():
|
|
try:
|
|
cluster.start()
|
|
time.sleep(1)
|
|
|
|
yield cluster
|
|
finally:
|
|
cluster.shutdown()
|
|
|
|
def test_creating_table_different_setting(start_cluster):
|
|
node1.query("CREATE TABLE t1 (c1 String, c2 String) ENGINE=ReplicatedMergeTree('/clickhouse/t1', '1') ORDER BY tuple(c1) SETTINGS index_granularity_bytes = 0")
|
|
node1.query("INSERT INTO t1 VALUES('x', 'y')")
|
|
|
|
node2.query("CREATE TABLE t1 (c1 String, c2 String) ENGINE=ReplicatedMergeTree('/clickhouse/t1', '2') ORDER BY tuple(c1)")
|
|
|
|
node1.query("INSERT INTO t1 VALUES('a', 'b')")
|
|
node2.query("SYSTEM SYNC REPLICA t1", timeout=5)
|
|
|
|
node1.query("SELECT count() FROM t1") == "2\n"
|
|
node2.query("SELECT count() FROM t1") == "1\n"
|
|
|
|
node2.query("INSERT INTO t1 VALUES('c', 'd')")
|
|
node1.query("SYSTEM SYNC REPLICA t1", timeout=5)
|
|
|
|
# replication works
|
|
node1.query("SELECT count() FROM t1") == "3\n"
|
|
node2.query("SELECT count() FROM t1") == "2\n"
|
|
|
|
# OPTIMIZE also works correctly
|
|
node2.query("OPTIMIZE TABLE t1 FINAL") == "3\n"
|
|
node1.query("SYSTEM SYNC REPLICA t1", timeout=5)
|
|
|
|
node1.query("SELECT count() FROM t1") == "3\n"
|
|
node2.query("SELECT count() FROM t1") == "2\n"
|
|
|
|
path_part = node1.query("SELECT path FROM system.parts WHERE table = 't1' AND active=1 ORDER BY partition DESC LIMIT 1").strip()
|
|
|
|
with pytest.raises(Exception): # check that we have no adaptive files
|
|
node1.exec_in_container(["bash", "-c", "find {p} -name '*.mrk2' | grep '.*'".format(p=path_part)])
|
|
|
|
path_part = node2.query("SELECT path FROM system.parts WHERE table = 't1' AND active=1 ORDER BY partition DESC LIMIT 1").strip()
|
|
|
|
with pytest.raises(Exception): # check that we have no adaptive files
|
|
node2.exec_in_container(["bash", "-c", "find {p} -name '*.mrk2' | grep '.*'".format(p=path_part)])
|
|
|
|
|
|
def test_old_node_with_new_node(start_cluster):
|
|
node3.query("CREATE TABLE t2 (c1 String, c2 String) ENGINE=ReplicatedMergeTree('/clickhouse/t2', '3') ORDER BY tuple(c1)")
|
|
node3.query("INSERT INTO t2 VALUES('x', 'y')")
|
|
|
|
node2.query("CREATE TABLE t2 (c1 String, c2 String) ENGINE=ReplicatedMergeTree('/clickhouse/t2', '2') ORDER BY tuple(c1)")
|
|
|
|
node3.query("INSERT INTO t2 VALUES('a', 'b')")
|
|
node2.query("SYSTEM SYNC REPLICA t2", timeout=5)
|
|
|
|
node3.query("SELECT count() FROM t2") == "2\n"
|
|
node2.query("SELECT count() FROM t2") == "1\n"
|
|
|
|
node2.query("INSERT INTO t2 VALUES('c', 'd')")
|
|
node3.query("SYSTEM SYNC REPLICA t2", timeout=5)
|
|
|
|
# replication works
|
|
node3.query("SELECT count() FROM t2") == "3\n"
|
|
node2.query("SELECT count() FROM t2") == "2\n"
|
|
|
|
# OPTIMIZE also works correctly
|
|
node3.query("OPTIMIZE table t2 FINAL")
|
|
node2.query("SYSTEM SYNC REPLICA t2", timeout=5)
|
|
|
|
node3.query("SELECT count() FROM t2") == "3\n"
|
|
node2.query("SELECT count() FROM t2") == "2\n"
|
|
|
|
path_part = node3.query("SELECT path FROM system.parts WHERE table = 't2' AND active=1 ORDER BY partition DESC LIMIT 1").strip()
|
|
|
|
with pytest.raises(Exception): # check that we have no adaptive files
|
|
node3.exec_in_container(["bash", "-c", "find {p} -name '*.mrk2' | grep '.*'".format(p=path_part)])
|
|
|
|
path_part = node2.query("SELECT path FROM system.parts WHERE table = 't2' AND active=1 ORDER BY partition DESC LIMIT 1").strip()
|
|
|
|
with pytest.raises(Exception): # check that we have no adaptive files
|
|
node2.exec_in_container(["bash", "-c", "find {p} -name '*.mrk2' | grep '.*'".format(p=path_part)])
|