mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-11-22 07:31:57 +00:00
b75963d370
This PR formats all the `*.py` files found under the `tests/integration` folder. It also reorders the imports and cleans up a bunch of unused imports. The formatting also takes care of other things like wrapping lines and fixing spaces and indents such that the tests look more readable.
50 lines
1.9 KiB
Python
50 lines
1.9 KiB
Python
import pytest
|
|
|
|
from helpers.cluster import ClickHouseCluster
|
|
|
|
cluster = ClickHouseCluster(__file__)
|
|
node1 = cluster.add_instance('node1')
|
|
node2 = cluster.add_instance('node2')
|
|
|
|
|
|
@pytest.fixture(scope="module")
|
|
def start_cluster():
|
|
try:
|
|
cluster.start()
|
|
|
|
for node in [node1, node2]:
|
|
node.query(
|
|
"create table da_memory_efficient_shard(A Int64, B Int64) Engine=MergeTree order by A partition by B % 2;")
|
|
|
|
node1.query("insert into da_memory_efficient_shard select number, number from numbers(100000);")
|
|
node2.query("insert into da_memory_efficient_shard select number + 100000, number from numbers(100000);")
|
|
|
|
yield cluster
|
|
|
|
finally:
|
|
cluster.shutdown()
|
|
|
|
|
|
def test_remote(start_cluster):
|
|
node1.query(
|
|
"set distributed_aggregation_memory_efficient = 1, group_by_two_level_threshold = 1, group_by_two_level_threshold_bytes=1")
|
|
res = node1.query(
|
|
"select sum(a) from (SELECT B, uniqExact(A) a FROM remote('node{1,2}', default.da_memory_efficient_shard) GROUP BY B)")
|
|
assert res == '200000\n'
|
|
|
|
node1.query("set distributed_aggregation_memory_efficient = 0")
|
|
res = node1.query(
|
|
"select sum(a) from (SELECT B, uniqExact(A) a FROM remote('node{1,2}', default.da_memory_efficient_shard) GROUP BY B)")
|
|
assert res == '200000\n'
|
|
|
|
node1.query(
|
|
"set distributed_aggregation_memory_efficient = 1, group_by_two_level_threshold = 1, group_by_two_level_threshold_bytes=1")
|
|
res = node1.query(
|
|
"SELECT fullHostName() AS h, uniqExact(A) AS a FROM remote('node{1,2}', default.da_memory_efficient_shard) GROUP BY h ORDER BY h;")
|
|
assert res == 'node1\t100000\nnode2\t100000\n'
|
|
|
|
node1.query("set distributed_aggregation_memory_efficient = 0")
|
|
res = node1.query(
|
|
"SELECT fullHostName() AS h, uniqExact(A) AS a FROM remote('node{1,2}', default.da_memory_efficient_shard) GROUP BY h ORDER BY h;")
|
|
assert res == 'node1\t100000\nnode2\t100000\n'
|