[
https://issues.apache.org/jira/browse/CASSANDRA-17891?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael Semb Wever updated CASSANDRA-17891:
-------------------------------------------
Fix Version/s: 4.1-rc1
> Fix hintedhandoff_test.py::TestHintedHandoff::test_hintedhandoff_window
> -----------------------------------------------------------------------
>
> Key: CASSANDRA-17891
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17891
> Project: Cassandra
> Issue Type: Bug
> Components: CI
> Reporter: Ekaterina Dimitrova
> Assignee: Stefan Miklosovic
> Priority: Normal
> Fix For: 4.1-rc1, 4.1-rc, 4.1, 4.2
>
>
> hintedhandoff_test.py::TestHintedHandoff::test_hintedhandoff_window is flaky
>
> {code:java}
> self = <hintedhandoff_test.TestHintedHandoff object at 0x7f2d903b8630>
> @since('4.1')
> def test_hintedhandoff_window(self):
> """
> Test that we only store at a maximum the hint window worth of
> hints.
> Prior to CASSANDRA-14309 we would store another window worth of
> hints
> if the down node was brought up and then taken back down
> immediately.
> We would also store another window of hints on a live node if the
> live
> node was restarted.
> @jira_ticket CASSANDRA-14309
> """
>
> # hint_window_persistent_enabled is set to true by default
> self.cluster.set_configuration_options({'max_hint_window_in_ms':
> 10000,
> 'hinted_handoff_enabled':
> True,
> 'max_hints_delivery_threads':
> 1,
> 'hints_flush_period_in_ms':
> 100, })
> self.cluster.populate(2).start()
> node1, node2 = self.cluster.nodelist()
> session = self.patient_cql_connection(node1)
> create_ks(session, 'ks', 2)
> create_c1c2_table(self, session)
>
> # Stop handoff until very end and take node2 down for first round of
> hints
> node1.nodetool('pausehandoff')
>
> node2.nodetool('disablebinary')
> node2.watch_log_for(["Stop listening for CQL clients"], timeout=120)
>
> node2.nodetool('disablegossip')
> node2.watch_log_for(["Announcing shutdown", "state jump to
> shutdown"], timeout=120)
> node1.watch_log_for(["state jump to shutdown"], timeout=120)
>
> log_mark_node_1 = node1.mark_log()
> log_mark_node_2 = node2.mark_log()
>
> # First round of hints. We expect these to be replayed and the only
> # hints within the window
> insert_c1c2(session, n=(0, 100), consistency=ConsistencyLevel.ONE)
>
> # Let hint window pass
> time.sleep(15)
>
> # Re-enable and disable the node. Prior to CASSANDRA-14215 this
> should make the hint window on node1 reset.
> node2.nodetool('enablegossip')
> node2.watch_log_for(["state jump to NORMAL"], timeout=120,
> from_mark=log_mark_node_2)
> node1.watch_log_for(["state jump to NORMAL"], timeout=120,
> from_mark=log_mark_node_1)
>
> log_mark_node_1 = node1.mark_log()
> log_mark_node_2 = node2.mark_log()
>
> node2.nodetool('disablegossip')
>
> node2.watch_log_for(["Announcing shutdown", "state jump to
> shutdown"], timeout=120, from_mark=log_mark_node_2)
> node1.watch_log_for(["state jump to shutdown"], timeout=120,
> from_mark=log_mark_node_1)
>
> log_mark_node_1 = node1.mark_log()
> log_mark_node_2 = node2.mark_log()
>
> def endpoint_downtime(node_to_query, node):
> mbean = make_mbean('net', type='Gossiper')
> with JolokiaAgent(node_to_query) as jmx:
> return jmx.execute_method(mbean,
> 'getEndpointDowntime(java.lang.String)', [node])
>
> while endpoint_downtime(node1, "127.0.0.2") <= 5000:
> time.sleep(1)
>
> # Second round of inserts. We do not expect hints to be stored.
> insert_c1c2(session, n=(100, 200), consistency=ConsistencyLevel.ONE)
>
> # Restart node1. Prior to CASSANDRA-14215 this would reset node1's
> hint window.
> node1.stop()
> node1.start(wait_for_binary_proto=True, wait_other_notice=False)
> session = self.patient_exclusive_cql_connection(node1)
> session.execute('USE ks')
> # Third round of inserts. We do not expect hints to be stored.
> insert_c1c2(session, n=(200, 300), consistency=ConsistencyLevel.ONE)
>
> # Enable node2 and wait for hints to be replayed
> node2.nodetool('enablegossip')
> node2.watch_log_for(["state jump to NORMAL"], timeout=120,
> from_mark=log_mark_node_2)
>
> node2.nodetool('enablebinary')
> node2.watch_log_for(["Starting listening for CQL clients"],
> timeout=120, from_mark=log_mark_node_2)
> node1.nodetool('resumehandoff')
> node1.watch_log_for('Finished hinted handoff')
> # Stop node1 so that we only query node2
> node1.stop()
>
> session = self.patient_exclusive_cql_connection(node2)
> session.execute('USE ks')
> # Ensure first dataset is present (through replayed hints)
> for x in range(0, 100):
> query_c1c2(session, x, ConsistencyLevel.ONE)
>
> # Ensure second and third datasets are not present
> for x in range(100, 300):
> > query_c1c2(session, x, ConsistencyLevel.ONE,
> > tolerate_missing=True, must_be_missing=True)
> hintedhandoff_test.py:303:
> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
> _
> tools/data.py:47: in query_c1c2
> assertions.assert_length_equal(rows, 0)
> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
> _
> object_with_length = [Row(c1='value1', c2='value2')], expected_length = 0
> def assert_length_equal(object_with_length, expected_length):
> """
> Assert an object has a specific length.
> @param object_with_length The object whose length will be checked
> @param expected_length The expected length of the object
>
> Examples:
> assert_length_equal(res, nb_counter)
> """
> assert len(object_with_length) == expected_length, \
> "Expected {} to have length {}, but instead is of length {}"\
> > .format(object_with_length, expected_length,
> > len(object_with_length))
> E AssertionError: Expected [Row(c1='value1', c2='value2')] to have
> length 0, but instead is of length 1
> tools/assertions.py:269: AssertionError
> {code}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
