Jim Witschey created CASSANDRA-10553:
----------------------------------------
Summary: Investigate likely Windows OOMs on CassCI
Key: CASSANDRA-10553
URL: https://issues.apache.org/jira/browse/CASSANDRA-10553
Project: Cassandra
Issue Type: Bug
Reporter: Jim Witschey
Assignee: Joshua McKenzie
[~philipthompson] had a look at a few Windows timeouts on CassCI:
https://cassci.datastax.com/view/cassandra-3.0/job/cassandra-3.0_dtest_win32/86/console
https://cassci.datastax.com/view/cassandra-3.0/job/cassandra-3.0_dtest_win32/87/console
https://cassci.datastax.com/view/cassandra-3.0/job/cassandra-3.0_dtest_win32/88/console
The offending tests were
{{bootstrap_test.py:TestBootstrap.killed_wiped_node_cannot_join_test}} in two
runs and
{{consistency_test.py:TestAccuracy.test_network_topology_strategy_counters}} in
one. In the latter case, {{killed_wiped_node_cannot_join_test}} passed, so this
doesn't happen 100% of the time.
Philip has reproduced this locally
({{bootstrap_test.py:TestBootstrap.decommissioned_wiped_node_can_join_test}}),
so it's not purely a CassCI problem. The cause there seems to be an OOM -- when
running the tests, he sees 8GB "committed" in the process explorer, but only 4
actually used.
I'm opening this ticket to track work on configuring the Windows dtests so this
doesn't happen on CassCI anymore. Assigning [~JoshuaMcKenzie] for the moment,
but [~pauloricardomg] might be able to make short work of this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
