ClickHouse/dbms/tests/integration/test_distributed_ddl/test.py

import os
import sys
import time
from contextlib import contextmanager

import pytest

sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from helpers.network import PartitionManager
from helpers.test_tools import TSV
from .cluster import ClickHouseClusterWithDDLHelpers


@pytest.fixture(scope="module", params=["configs", "configs_secure"])
def test_cluster(request):
    cluster = ClickHouseClusterWithDDLHelpers(__file__, request.param)

    try:
        cluster.prepare()

        yield cluster

        instance = cluster.instances['ch1']
        cluster.ddl_check_query(instance, "DROP DATABASE test ON CLUSTER 'cluster'")
        cluster.ddl_check_query(instance, "DROP DATABASE IF EXISTS test2 ON CLUSTER 'cluster'")

        # Check query log to ensure that DDL queries are not executed twice
        time.sleep(1.5)
        for instance in cluster.instances.values():
            cluster.ddl_check_there_are_no_dublicates(instance)

        cluster.pm_random_drops.heal_all()

    finally:
        cluster.shutdown()


def test_default_database(test_cluster):
    instance = test_cluster.instances['ch3']

    test_cluster.ddl_check_query(instance, "CREATE DATABASE IF NOT EXISTS test2 ON CLUSTER 'cluster' FORMAT TSV")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS null ON CLUSTER 'cluster' FORMAT TSV")
    test_cluster.ddl_check_query(instance, "CREATE TABLE null ON CLUSTER 'cluster2' (s String DEFAULT 'escape\t\nme') ENGINE = Null")

    contents = instance.query("SELECT hostName() AS h, database FROM all_tables WHERE name = 'null' ORDER BY h")
    assert TSV(contents) == TSV("ch1\tdefault\nch2\ttest2\nch3\tdefault\nch4\ttest2\n")

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS null ON CLUSTER cluster2")
    test_cluster.ddl_check_query(instance, "DROP DATABASE IF EXISTS test2 ON CLUSTER 'cluster'")


def test_create_view(test_cluster):
    instance = test_cluster.instances['ch3']
    test_cluster.ddl_check_query(instance, "CREATE VIEW test.super_simple_view ON CLUSTER 'cluster' AS SELECT * FROM system.numbers FORMAT TSV")
    test_cluster.ddl_check_query(instance, "CREATE MATERIALIZED VIEW test.simple_mat_view ON CLUSTER 'cluster' ENGINE = Memory AS SELECT * FROM system.numbers FORMAT TSV")
    test_cluster.ddl_check_query(instance, "DROP TABLE test.simple_mat_view ON CLUSTER 'cluster' FORMAT TSV")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test.super_simple_view2 ON CLUSTER 'cluster' FORMAT TSV")

    test_cluster.ddl_check_query(instance, "CREATE TABLE test.super_simple ON CLUSTER 'cluster' (i Int8) ENGINE = Memory")
    test_cluster.ddl_check_query(instance, "RENAME TABLE test.super_simple TO test.super_simple2 ON CLUSTER 'cluster' FORMAT TSV")
    test_cluster.ddl_check_query(instance, "DROP TABLE test.super_simple2 ON CLUSTER 'cluster'")


def test_on_server_fail(test_cluster):
    instance = test_cluster.instances['ch1']
    kill_instance = test_cluster.instances['ch2']

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test.test_server_fail ON CLUSTER 'cluster'")

    kill_instance.get_docker_handle().stop()
    request = instance.get_query_request("CREATE TABLE test.test_server_fail ON CLUSTER 'cluster' (i Int8) ENGINE=Null", timeout=30)
    kill_instance.get_docker_handle().start()

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test.__nope__ ON CLUSTER 'cluster'")

    # Check query itself
    test_cluster.check_all_hosts_successfully_executed(request.get_answer())

    # And check query artefacts
    contents = instance.query("SELECT hostName() AS h FROM all_tables WHERE database='test' AND name='test_server_fail' ORDER BY h")
    assert TSV(contents) == TSV("ch1\nch2\nch3\nch4\n")

    test_cluster.ddl_check_query(instance, "DROP TABLE test.test_server_fail ON CLUSTER 'cluster'")


def _test_on_connection_losses(test_cluster, zk_timeout):
    instance = test_cluster.instances['ch1']
    kill_instance = test_cluster.instances['ch2']

    with PartitionManager() as pm:
        pm.drop_instance_zk_connections(kill_instance)
        request = instance.get_query_request("DROP TABLE IF EXISTS test.__nope__ ON CLUSTER 'cluster'", timeout=10)
        time.sleep(zk_timeout)
        pm.restore_instance_zk_connections(kill_instance)

    test_cluster.check_all_hosts_successfully_executed(request.get_answer())


def test_on_connection_loss(test_cluster):
    _test_on_connection_losses(test_cluster, 1.5) # connection loss will occur only (3 sec ZK timeout in config)


def test_on_session_expired(test_cluster):
    _test_on_connection_losses(test_cluster, 4) # session should be expired (3 sec ZK timeout in config)


def test_simple_alters(test_cluster):
    instance = test_cluster.instances['ch2']

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS merge ON CLUSTER cluster_without_replication")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS all_merge_32 ON CLUSTER cluster_without_replication")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS all_merge_64 ON CLUSTER cluster_without_replication")

    test_cluster.ddl_check_query(instance, """
CREATE TABLE IF NOT EXISTS merge ON CLUSTER cluster_without_replication (p Date, i Int32)
ENGINE = MergeTree(p, p, 1)
""")
    test_cluster.ddl_check_query(instance, """
CREATE TABLE IF NOT EXISTS all_merge_32 ON CLUSTER cluster_without_replication (p Date, i Int32)
ENGINE = Distributed(cluster_without_replication, default, merge, i)
""")
    test_cluster.ddl_check_query(instance, """
CREATE TABLE IF NOT EXISTS all_merge_64 ON CLUSTER cluster_without_replication (p Date, i Int64, s String)
ENGINE = Distributed(cluster_without_replication, default, merge, i)
""")

    for i in xrange(4):
        k = (i / 2) * 2
        test_cluster.instances['ch{}'.format(i + 1)].query("INSERT INTO merge (i) VALUES ({})({})".format(k, k+1))

    assert TSV(instance.query("SELECT i FROM all_merge_32 ORDER BY i")) == TSV(''.join(['{}\n'.format(x) for x in xrange(4)]))


    time.sleep(5)
    test_cluster.ddl_check_query(instance, "ALTER TABLE merge ON CLUSTER cluster_without_replication MODIFY COLUMN i Int64")
    time.sleep(5)
    test_cluster.ddl_check_query(instance, "ALTER TABLE merge ON CLUSTER cluster_without_replication ADD COLUMN s DEFAULT toString(i) FORMAT TSV")

    assert TSV(instance.query("SELECT i, s FROM all_merge_64 ORDER BY i")) == TSV(''.join(['{}\t{}\n'.format(x,x) for x in xrange(4)]))


    for i in xrange(4):
        k = (i / 2) * 2 + 4
        test_cluster.instances['ch{}'.format(i + 1)].query("INSERT INTO merge (p, i) VALUES (31, {})(31, {})".format(k, k+1))

    assert TSV(instance.query("SELECT i, s FROM all_merge_64 ORDER BY i")) == TSV(''.join(['{}\t{}\n'.format(x,x) for x in xrange(8)]))


    test_cluster.ddl_check_query(instance, "ALTER TABLE merge ON CLUSTER cluster_without_replication DETACH PARTITION 197002")
    assert TSV(instance.query("SELECT i, s FROM all_merge_64 ORDER BY i")) == TSV(''.join(['{}\t{}\n'.format(x,x) for x in xrange(4)]))

    test_cluster.ddl_check_query(instance, "DROP TABLE merge ON CLUSTER cluster_without_replication")
    test_cluster.ddl_check_query(instance, "DROP TABLE all_merge_32 ON CLUSTER cluster_without_replication")
    test_cluster.ddl_check_query(instance, "DROP TABLE all_merge_64 ON CLUSTER cluster_without_replication")


def test_macro(test_cluster):
    instance = test_cluster.instances['ch2']
    test_cluster.ddl_check_query(instance, "CREATE TABLE tab ON CLUSTER '{cluster}' (value UInt8) ENGINE = Memory")

    for i in xrange(4):
        test_cluster.insert_reliable(test_cluster.instances['ch{}'.format(i + 1)], "INSERT INTO tab VALUES ({})".format(i))

    test_cluster.ddl_check_query(instance, "CREATE TABLE distr ON CLUSTER '{cluster}' (value UInt8) ENGINE = Distributed('{cluster}', 'default', 'tab', value % 4)")

    assert TSV(instance.query("SELECT value FROM distr ORDER BY value")) == TSV('0\n1\n2\n3\n')
    assert TSV(test_cluster.instances['ch3'].query("SELECT value FROM distr ORDER BY value")) == TSV('0\n1\n2\n3\n')

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS distr ON CLUSTER '{cluster}'")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS tab ON CLUSTER '{cluster}'")


def test_implicit_macros(test_cluster):
    # Temporarily disable random ZK packet drops, they might broke creation if ReplicatedMergeTree replicas
    firewall_drops_rules = test_cluster.pm_random_drops.pop_rules()

    instance = test_cluster.instances['ch2']

    test_cluster.ddl_check_query(instance, "DROP DATABASE IF EXISTS test_db ON CLUSTER '{cluster}'")
    test_cluster.ddl_check_query(instance, "CREATE DATABASE IF NOT EXISTS test_db ON CLUSTER '{cluster}'")

    test_cluster.ddl_check_query(instance, """
CREATE TABLE IF NOT EXISTS test_db.test_macro ON CLUSTER '{cluster}' (p Date, i Int32)
ENGINE = ReplicatedMergeTree('/clickhouse/tables/{database}/{layer}-{shard}/{table}', '{replica}', p, p, 1)
""")

    # Check that table was created at correct path in zookeeper
    assert test_cluster.get_kazoo_client('zoo1').exists('/clickhouse/tables/test_db/0-1/test_macro') is not None

    # Enable random ZK packet drops
    test_cluster.pm_random_drops.push_rules(firewall_drops_rules)


def test_allowed_databases(test_cluster):
    instance = test_cluster.instances['ch2']
    instance.query("CREATE DATABASE IF NOT EXISTS db1 ON CLUSTER cluster")
    instance.query("CREATE DATABASE IF NOT EXISTS db2 ON CLUSTER cluster")

    instance.query("CREATE TABLE db1.t1 ON CLUSTER cluster (i Int8) ENGINE = Memory", settings={"user" : "restricted_user"})

    with pytest.raises(Exception):
        instance.query("CREATE TABLE db2.t2 ON CLUSTER cluster (i Int8) ENGINE = Memory", settings={"user" : "restricted_user"})
    with pytest.raises(Exception):
        instance.query("CREATE TABLE t3 ON CLUSTER cluster (i Int8) ENGINE = Memory", settings={"user" : "restricted_user"})
    with pytest.raises(Exception):
        instance.query("DROP DATABASE db2 ON CLUSTER cluster", settings={"user" : "restricted_user"})

    instance.query("DROP DATABASE db1 ON CLUSTER cluster", settings={"user" : "restricted_user"})

def test_kill_query(test_cluster):
    instance = test_cluster.instances['ch3']

    test_cluster.ddl_check_query(instance, "KILL QUERY ON CLUSTER 'cluster' WHERE NOT elapsed FORMAT TSV")

def test_detach_query(test_cluster):
    instance = test_cluster.instances['ch3']

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test_attach ON CLUSTER cluster FORMAT TSV")
    test_cluster.ddl_check_query(instance, "CREATE TABLE test_attach ON CLUSTER cluster (i Int8)ENGINE = Log")
    test_cluster.ddl_check_query(instance, "DETACH TABLE test_attach ON CLUSTER cluster FORMAT TSV")
    test_cluster.ddl_check_query(instance, "ATTACH TABLE test_attach ON CLUSTER cluster")


def test_optimize_query(test_cluster):
    instance = test_cluster.instances['ch3']

    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test_optimize ON CLUSTER cluster FORMAT TSV")
    test_cluster.ddl_check_query(instance, "CREATE TABLE test_optimize ON CLUSTER cluster (p Date, i Int32) ENGINE = MergeTree(p, p, 8192)")
    test_cluster.ddl_check_query(instance, "OPTIMIZE TABLE test_optimize ON CLUSTER cluster FORMAT TSV")


def test_create_as_select(test_cluster):
    instance = test_cluster.instances['ch2']
    test_cluster.ddl_check_query(instance, "CREATE TABLE test_as_select ON CLUSTER cluster ENGINE = Memory AS (SELECT 1 AS x UNION ALL SELECT 2 AS x)")
    assert TSV(instance.query("SELECT x FROM test_as_select ORDER BY x")) == TSV("1\n2\n")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test_as_select ON CLUSTER cluster")


def test_create_reserved(test_cluster):
    instance = test_cluster.instances['ch2']
    test_cluster.ddl_check_query(instance, "CREATE TABLE test_reserved ON CLUSTER cluster (`p` Date, `image` Nullable(String), `index` Nullable(Float64), `invalidate` Nullable(Int64)) ENGINE = MergeTree(`p`, `p`, 8192)")
    test_cluster.ddl_check_query(instance, "CREATE TABLE test_as_reserved ON CLUSTER cluster ENGINE = Memory AS (SELECT * from test_reserved)")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test_reserved ON CLUSTER cluster")
    test_cluster.ddl_check_query(instance, "DROP TABLE IF EXISTS test_as_reserved ON CLUSTER cluster")


if __name__ == '__main__':
    with contextmanager(test_cluster)() as ctx_cluster:
       for name, instance in ctx_cluster.instances.items():
           print name, instance.ip_address
       raw_input("Cluster created, press any key to destroy...")