Python Node.watch_log_for方法代码示例

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_with_reset_resume_state_test(self):
        """Test replace with resetting bootstrap progress"""

        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        node1.stress(['write', 'n=100K', 'no-warmup', '-schema', 'replication(factor=3)'])

        session = self.patient_cql_connection(node1)
        stress_table = 'keyspace1.standard1'
        query = SimpleStatement('select * from %s LIMIT 1' % stress_table, consistency_level=ConsistencyLevel.THREE)
        initial_data = rows_to_list(session.execute(query))

        node3.stop(gently=False)

        # kill node1 in the middle of streaming to let it fail
        t = InterruptBootstrap(node1)
        t.start()
        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")
        node4 = Node('node4', cluster=cluster, auto_bootstrap=True, thrift_interface=('127.0.0.4', 9160),
                     storage_interface=('127.0.0.4', 7000), jmx_port='7400', remote_debug_port='0',
                     initial_token=None, binary_interface=('127.0.0.4', 9042))

        # keep timeout low so that test won't hang
        node4.set_configuration_options(values={'streaming_socket_timeout_in_ms': 1000})
        cluster.add(node4, False)
        try:
            node4.start(jvm_args=["-Dcassandra.replace_address_first_boot=127.0.0.3"], wait_other_notice=False)
        except NodeError:
            pass  # node doesn't start as expected
        t.join()
        node1.start()

        # restart node4 bootstrap with resetting bootstrap state
        node4.stop()
        mark = node4.mark_log()
        node4.start(jvm_args=[
                    "-Dcassandra.replace_address_first_boot=127.0.0.3",
                    "-Dcassandra.reset_bootstrap_progress=true"
                    ])
        # check if we reset bootstrap state
        node4.watch_log_for("Resetting bootstrap progress to start fresh", from_mark=mark)
        # wait for node3 ready to query
        node4.watch_log_for("Listening for thrift clients...", from_mark=mark)

        # check if 2nd bootstrap succeeded
        assert_bootstrap_state(self, node4, 'COMPLETED')

        # query should work again
        debug("Stopping old nodes")
        node1.stop(gently=False, wait_other_notice=True)
        node2.stop(gently=False, wait_other_notice=True)

        debug("Verifying data on new node.")
        session = self.patient_exclusive_cql_connection(node4)
        assert_all(session, 'SELECT * from {} LIMIT 1'.format(stress_table),
                   expected=initial_data,
                   cl=ConsistencyLevel.ONE)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def resumable_replace_test(self):
        """
        Test resumable bootstrap while replacing node. Feature introduced in
        2.2 with ticket https://issues.apache.org/jira/browse/CASSANDRA-8838

        @jira_ticket https://issues.apache.org/jira/browse/CASSANDRA-8838
        """

        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        node1.stress(['write', 'n=100K', 'no-warmup', '-schema', 'replication(factor=3)'])

        session = self.patient_cql_connection(node1)
        stress_table = 'keyspace1.standard1'
        query = SimpleStatement('select * from %s LIMIT 1' % stress_table, consistency_level=ConsistencyLevel.THREE)
        initial_data = rows_to_list(session.execute(query))

        node3.stop(gently=False)

        # kill node1 in the middle of streaming to let it fail
        t = InterruptBootstrap(node1)
        t.start()
        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")
        node4 = Node('node4', cluster=cluster, auto_bootstrap=True, thrift_interface=('127.0.0.4', 9160),
                     storage_interface=('127.0.0.4', 7000), jmx_port='7400', remote_debug_port='0',
                     initial_token=None, binary_interface=('127.0.0.4', 9042))
        # keep timeout low so that test won't hang
        node4.set_configuration_options(values={'streaming_socket_timeout_in_ms': 1000})
        cluster.add(node4, False)
        try:
            node4.start(jvm_args=["-Dcassandra.replace_address_first_boot=127.0.0.3"], wait_other_notice=False)
        except NodeError:
            pass  # node doesn't start as expected
        t.join()

        # bring back node1 and invoke nodetool bootstrap to resume bootstrapping
        node1.start()
        node4.nodetool('bootstrap resume')
        # check if we skipped already retrieved ranges
        node4.watch_log_for("already available. Skipping streaming.")
        # wait for node3 ready to query
        node4.watch_log_for("Listening for thrift clients...")

        # check if 2nd bootstrap succeeded
        assert_bootstrap_state(self, node4, 'COMPLETED')

        # query should work again
        debug("Stopping old nodes")
        node1.stop(gently=False, wait_other_notice=True)
        node2.stop(gently=False, wait_other_notice=True)

        debug("Verifying data on new node.")
        session = self.patient_exclusive_cql_connection(node4)
        assert_all(session, 'SELECT * from {} LIMIT 1'.format(stress_table),
                   expected=initial_data,
                   cl=ConsistencyLevel.ONE)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_with_reset_resume_state_test(self):
        """Test replace with resetting bootstrap progress"""

        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        node1.stress(['write', 'n=100000', '-schema', 'replication(factor=3)'])

        session = self.patient_cql_connection(node1)
        stress_table = 'keyspace1.standard1'
        query = SimpleStatement('select * from %s LIMIT 1' % stress_table, consistency_level=ConsistencyLevel.THREE)
        initialData = list(session.execute(query))

        node3.stop(gently=False)

        # kill node1 in the middle of streaming to let it fail
        t = InterruptBootstrap(node1)
        t.start()
        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")
        node4 = Node('node4', cluster, True, ('127.0.0.4', 9160), ('127.0.0.4', 7000), '7400', '0', None, binary_interface=('127.0.0.4', 9042))
        # keep timeout low so that test won't hang
        node4.set_configuration_options(values={'streaming_socket_timeout_in_ms': 1000})
        cluster.add(node4, False)
        try:
            node4.start(jvm_args=["-Dcassandra.replace_address_first_boot=127.0.0.3"], wait_other_notice=False)
        except NodeError:
            pass  # node doesn't start as expected
        t.join()
        node1.start()

        # restart node4 bootstrap with resetting bootstrap state
        node4.stop()
        mark = node4.mark_log()
        node4.start(jvm_args=[
                    "-Dcassandra.replace_address_first_boot=127.0.0.3",
                    "-Dcassandra.reset_bootstrap_progress=true"
                    ])
        # check if we reset bootstrap state
        node4.watch_log_for("Resetting bootstrap progress to start fresh", from_mark=mark)
        # wait for node3 ready to query
        node4.watch_log_for("Listening for thrift clients...", from_mark=mark)

        # check if 2nd bootstrap succeeded
        session = self.exclusive_cql_connection(node4)
        rows = list(session.execute("SELECT bootstrapped FROM system.local WHERE key='local'"))
        assert len(rows) == 1
        assert rows[0][0] == 'COMPLETED', rows[0][0]

        # query should work again
        debug("Verifying querying works again.")
        finalData = list(session.execute(query))
        self.assertListEqual(initialData, finalData)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def resumable_replace_test(self):
        """
        Test resumable bootstrap while replacing node. Feature introduced in
        2.2 with ticket https://issues.apache.org/jira/browse/CASSANDRA-8838

        @jira_ticket https://issues.apache.org/jira/browse/CASSANDRA-8838
        """

        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        node1.stress(['write', 'n=100K', '-schema', 'replication(factor=3)'])

        session = self.patient_cql_connection(node1)
        stress_table = 'keyspace1.standard1'
        query = SimpleStatement('select * from %s LIMIT 1' % stress_table, consistency_level=ConsistencyLevel.THREE)
        initialData = list(session.execute(query))

        node3.stop(gently=False)

        # kill node1 in the middle of streaming to let it fail
        t = InterruptBootstrap(node1)
        t.start()
        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")
        node4 = Node('node4', cluster, True, ('127.0.0.4', 9160), ('127.0.0.4', 7000), '7400', '0', None, binary_interface=('127.0.0.4', 9042))
        # keep timeout low so that test won't hang
        node4.set_configuration_options(values={'streaming_socket_timeout_in_ms': 1000})
        cluster.add(node4, False)
        try:
            node4.start(jvm_args=["-Dcassandra.replace_address_first_boot=127.0.0.3"], wait_other_notice=False)
        except NodeError:
            pass  # node doesn't start as expected
        t.join()

        # bring back node1 and invoke nodetool bootstrap to resume bootstrapping
        node1.start()
        node4.nodetool('bootstrap resume')
        # check if we skipped already retrieved ranges
        node4.watch_log_for("already available. Skipping streaming.")
        # wait for node3 ready to query
        node4.watch_log_for("Listening for thrift clients...")

        # check if 2nd bootstrap succeeded
        session = self.exclusive_cql_connection(node4)
        rows = list(session.execute("SELECT bootstrapped FROM system.local WHERE key='local'"))
        assert len(rows) == 1
        assert rows[0][0] == 'COMPLETED', rows[0][0]

        # query should work again
        debug("Verifying querying works again.")
        finalData = list(session.execute(query))
        self.assertListEqual(initialData, finalData)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_nonexistent_node_test(self):
        debug("Starting cluster with 3 nodes.")
        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        debug('Start node 4 and replace an address with no node')
        node4 = Node('node4', cluster, True, ('127.0.0.4', 9160), ('127.0.0.4', 7000), '7400', '0', None, binary_interface=('127.0.0.4', 9042))
        cluster.add(node4, False)

        # try to replace an unassigned ip address
        mark = node4.mark_log()
        node4.start(replace_address='127.0.0.5', wait_other_notice=False)
        node4.watch_log_for("java.lang.RuntimeException: Cannot replace_address /127.0.0.5 because it doesn't exist in gossip", from_mark=mark)
        self.check_not_running(node4)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_active_node_test(self):

        debug("Starting cluster with 3 nodes.")
        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        # replace active node 3 with node 4
        debug("Starting node 4 to replace active node 3")
        node4 = Node('node4', cluster, True, ('127.0.0.4', 9160), ('127.0.0.4', 7000), '7400', '0', None, binary_interface=('127.0.0.4', 9042))
        cluster.add(node4, False)

        mark = node4.mark_log()
        node4.start(replace_address='127.0.0.3', wait_other_notice=False)
        node4.watch_log_for("java.lang.UnsupportedOperationException: Cannot replace a live node...", from_mark=mark)
        self.check_not_running(node4)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def resumable_replace_test(self):
        """Test resumable bootstrap while replacing node"""

        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        node1.stress(['write', 'n=100000', '-schema', 'replication(factor=3)'])

        session = self.patient_cql_connection(node1)
        stress_table = 'keyspace1.standard1' if self.cluster.version() >= '2.1' else '"Keyspace1"."Standard1"'
        query = SimpleStatement('select * from %s LIMIT 1' % stress_table, consistency_level=ConsistencyLevel.THREE)
        initialData = session.execute(query)

        node3.stop(gently=False)

        # kill node1 in the middle of streaming to let it fail
        t = InterruptBootstrap(node1)
        t.start()
        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")
        node4 = Node('node4', cluster, True, ('127.0.0.4', 9160), ('127.0.0.4', 7000), '7400', '0', None, ('127.0.0.4', 9042))
        cluster.add(node4, False)
        try:
            node4.start(jvm_args=["-Dcassandra.replace_address_first_boot=127.0.0.3"])
        except NodeError:
            pass  # node doesn't start as expected
        t.join()

        # bring back node1 and invoke nodetool bootstrap to resume bootstrapping
        node1.start()
        node4.nodetool('bootstrap resume')
        # check if we skipped already retrieved ranges
        node4.watch_log_for("already available. Skipping streaming.")
        # wait for node3 ready to query
        node4.watch_log_for("Listening for thrift clients...")

        # check if 2nd bootstrap succeeded
        session = self.exclusive_cql_connection(node4)
        rows = session.execute("SELECT bootstrapped FROM system.local WHERE key='local'")
        assert len(rows) == 1
        assert rows[0][0] == 'COMPLETED', rows[0][0]

        #query should work again
        debug("Verifying querying works again.")
        finalData = session.execute(query)
        self.assertListEqual(initialData, finalData)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_nonexistent_node_test(self):
        debug("Starting cluster with 3 nodes.")
        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        debug('Start node 4 and replace an address with no node')
        node4 = Node('node4', cluster=cluster, auto_bootstrap=True, thrift_interface=('127.0.0.4', 9160),
                     storage_interface=('127.0.0.4', 7000), jmx_port='7400', remote_debug_port='0',
                     initial_token=None, binary_interface=('127.0.0.4', 9042))
        cluster.add(node4, False)

        # try to replace an unassigned ip address
        mark = node4.mark_log()
        node4.start(replace_address='127.0.0.5', wait_other_notice=False)
        node4.watch_log_for("java.lang.RuntimeException: Cannot replace_address /127.0.0.5 because it doesn't exist in gossip", from_mark=mark)
        assert_not_running(node4)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_active_node_test(self):
        debug("Starting cluster with 3 nodes.")
        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        # replace active node 3 with node 4
        debug("Starting node 4 to replace active node 3")
        node4 = Node('node4', cluster=cluster, auto_bootstrap=True, thrift_interface=('127.0.0.4', 9160),
                     storage_interface=('127.0.0.4', 7000), jmx_port='7400', remote_debug_port='0',
                     initial_token=None, binary_interface=('127.0.0.4', 9042))
        cluster.add(node4, False)

        mark = node4.mark_log()
        node4.start(replace_address='127.0.0.3', wait_other_notice=False)
        node4.watch_log_for("java.lang.UnsupportedOperationException: Cannot replace a live node...", from_mark=mark)
        assert_not_running(node4)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def replace_with_insufficient_replicas_test(self):
        """
        Test that replace fails when there are insufficient replicas
        @jira_ticket CASSANDRA-11848
        """
        debug("Starting cluster with 3 nodes.")
        cluster = self.cluster
        cluster.populate(3).start()
        node1, node2, node3 = cluster.nodelist()

        if DISABLE_VNODES:
            num_tokens = 1
        else:
            # a little hacky but grep_log returns the whole line...
            num_tokens = int(node3.get_conf_option('num_tokens'))

        debug("testing with num_tokens: {}".format(num_tokens))

        debug("Inserting Data...")
        node1.stress(['write', 'n=10K', 'no-warmup', '-schema', 'replication(factor=2)'])

        # stop node to replace
        debug("Stopping node to replace.")
        node3.stop(wait_other_notice=True)

        # stop other replica
        debug("Stopping node2 (other replica)")
        node2.stop(wait_other_notice=True)

        # replace node 3 with node 4
        debug("Starting node 4 to replace node 3")

        node4 = Node('node4', cluster=cluster, auto_bootstrap=True, thrift_interface=('127.0.0.4', 9160),
                     storage_interface=('127.0.0.4', 7000), jmx_port='7400', remote_debug_port='0',
                     initial_token=None, binary_interface=('127.0.0.4', 9042))
        cluster.add(node4, False)
        node4.start(replace_address='127.0.0.3', wait_for_binary_proto=False, wait_other_notice=False)

        # replace should fail due to insufficient replicas
        node4.watch_log_for("Unable to find sufficient sources for streaming range")
        assert_not_running(node4)

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
    def test_upgrade_legacy_table(self):
        """
        Upgrade with bringing up the legacy tables after the newer nodes (without legacy tables)
        were started.

        @jira_ticket CASSANDRA-12813
        """

        cluster = self.cluster

        # Forcing cluster version on purpose
        cluster.set_install_dir(version="2.1.16")
        cluster.populate(3).start()

        node1, node2, node3 = cluster.nodelist()

        # Wait for default user to get created on one of the nodes
        time.sleep(15)

        # Upgrade to current version
        for node in [node1, node2, node3]:
            node.drain()
            node.watch_log_for("DRAINED")
            node.stop(gently=True)
            self.set_node_to_current_version(node)

        cluster.start()

        # Make sure the system_auth table will get replicated to the node that we're going to replace
        session = self.patient_cql_connection(node1, user='cassandra', password='cassandra')
        session.execute("ALTER KEYSPACE system_auth WITH REPLICATION = { 'class' : 'SimpleStrategy', 'replication_factor' : 3 };")
        cluster.repair()

        cluster.stop()

        # Replace the node
        cluster.seeds.remove(node1)
        cluster.remove(node1)

        replacement_address = node1.address()
        replacement_node = Node('replacement', cluster=self.cluster, auto_bootstrap=True,
                                thrift_interface=(replacement_address, 9160), storage_interface=(replacement_address, 7000),
                                jmx_port='7400', remote_debug_port='0', initial_token=None, binary_interface=(replacement_address, 9042))
        self.set_node_to_current_version(replacement_node)

        cluster.add(replacement_node, True)
        replacement_node.start(wait_for_binary_proto=True)

        node2.start(wait_for_binary_proto=True)
        node3.start(wait_for_binary_proto=True)

        replacement_node.watch_log_for('Initializing system_auth.credentials')
        replacement_node.watch_log_for('Initializing system_auth.permissions')
        replacement_node.watch_log_for('Initializing system_auth.users')

        cluster.repair()
        replacement_node.watch_log_for('Repair command')

        # Should succeed. Will throw an NPE on pre-12813 code.
        self.patient_cql_connection(replacement_node, user='cassandra', password='cassandra')

# 需要导入模块: from ccmlib.node import Node [as 别名]
# 或者: from ccmlib.node.Node import watch_log_for [as 别名]
class BaseReplaceAddressTest(Tester):

    @pytest.fixture(autouse=True)
    def fixture_add_additional_log_patterns(self, fixture_dtest_setup):
        fixture_dtest_setup.ignore_log_patterns = (
            # This one occurs when trying to send the migration to a
            # node that hasn't started yet, and when it does, it gets
            # replayed and everything is fine.
            r'Can\'t send migration request: node.*is down',
            r'Migration task failed to complete',  # 10978
            # ignore streaming error during bootstrap
            r'Streaming error occurred',
            r'failed stream session',
            r'Failed to properly handshake with peer'
        )

    def _setup(self, n=3, opts=None, enable_byteman=False, mixed_versions=False):
        logger.debug("Starting cluster with {} nodes.".format(n))
        self.cluster.populate(n)
        if opts is not None:
            logger.debug("Setting cluster options: {}".format(opts))
            self.cluster.set_configuration_options(opts)

        self.cluster.set_batch_commitlog(enabled=True)
        self.query_node = self.cluster.nodelist()[0]
        self.replaced_node = self.cluster.nodelist()[-1]

        self.cluster.seeds.remove(self.replaced_node)
        NUM_TOKENS = os.environ.get('NUM_TOKENS', '256')
        if not self.dtest_config.use_vnodes:
            self.cluster.set_configuration_options(values={'initial_token': None, 'num_tokens': 1})
        else:
            self.cluster.set_configuration_options(values={'initial_token': None, 'num_tokens': NUM_TOKENS})

        if enable_byteman:
            # set up byteman
            self.query_node.byteman_port = '8100'
            self.query_node.import_config_files()

        if mixed_versions:
            logger.debug("Starting nodes on version 2.2.4")
            self.cluster.set_install_dir(version="2.2.4")

        self.cluster.start()

        if self.cluster.cassandra_version() >= '2.2.0':
            session = self.patient_cql_connection(self.query_node)
            # Change system_auth keyspace replication factor to 2, otherwise replace will fail
            session.execute("""ALTER KEYSPACE system_auth
                            WITH replication = {'class':'SimpleStrategy',
                            'replication_factor':2};""")

    def _do_replace(self, same_address=False, jvm_option='replace_address',
                    wait_other_notice=False, wait_for_binary_proto=True,
                    replace_address=None, opts=None, data_center=None,
                    extra_jvm_args=None):
        if replace_address is None:
            replace_address = self.replaced_node.address()

        # only create node if it's not yet created
        if self.replacement_node is None:
            replacement_address = '127.0.0.4'
            if same_address:
                replacement_address = self.replaced_node.address()
                self.cluster.remove(self.replaced_node)

            logger.debug("Starting replacement node {} with jvm_option '{}={}'".format(replacement_address, jvm_option, replace_address))
            self.replacement_node = Node('replacement', cluster=self.cluster, auto_bootstrap=True,
                                         thrift_interface=None, storage_interface=(replacement_address, 7000),
                                         jmx_port='7400', remote_debug_port='0', initial_token=None, binary_interface=(replacement_address, 9042))
            if opts is not None:
                logger.debug("Setting options on replacement node: {}".format(opts))
                self.replacement_node.set_configuration_options(opts)
            self.cluster.add(self.replacement_node, False, data_center=data_center)

        if extra_jvm_args is None:
            extra_jvm_args = []
        extra_jvm_args.extend(["-Dcassandra.{}={}".format(jvm_option, replace_address),
                               "-Dcassandra.ring_delay_ms=10000",
                               "-Dcassandra.broadcast_interval_ms=10000"])

        self.replacement_node.start(jvm_args=extra_jvm_args,
                                    wait_for_binary_proto=wait_for_binary_proto, wait_other_notice=wait_other_notice)

        if self.cluster.cassandra_version() >= '2.2.8' and same_address:
            self.replacement_node.watch_log_for("Writes will not be forwarded to this node during replacement",
                                                timeout=60)

    def _stop_node_to_replace(self, gently=False, table='keyspace1.standard1', cl=ConsistencyLevel.THREE):
        if self.replaced_node.is_running():
            logger.debug("Stopping {}".format(self.replaced_node.name))
            self.replaced_node.stop(gently=gently, wait_other_notice=True)

        logger.debug("Testing node stoppage (query should fail).")
        with pytest.raises((Unavailable, ReadTimeout)):
            session = self.patient_cql_connection(self.query_node)
            query = SimpleStatement('select * from {}'.format(table), consistency_level=cl)
            session.execute(query)

    def _insert_data(self, n='1k', rf=3, whitelist=False):
#.........这里部分代码省略.........