Hi Josh,
iiuc 17311 is a duplicate of 17140 going from what I learned working on
17140. In my experience I would start with that one. On some weird
failures processes are left around and a few other tests will then fail
on timeouts, OOM, ip collision etc. You might fix this one and get the
others to go away bc they were somehow side-effects.
#collaborating
Regards
On 27/1/22 20:20, Joshua McKenzie wrote:
Hey all. Working through our first week as Build Lead and there's a
lot of backlog to process. We had some changes to improve CPU
utilization on apache infra for ci-cassandra and infra is looking into
further optimizations; things are much more responsive from a UX
perspective at least on the site. Some key tickets and test failures
have floated to the top as being particularly high leverage /
responsible for multiple failures. For reference, see the butler
dashboards:
trunk:
https://butler.cassandra.apache.org/#/ci/upstream/compare/Cassandra-trunk/trunk
4.0:
https://butler.cassandra.apache.org/#/ci/upstream/compare/Cassandra-4.0/cassandra-4.0
3.0:
https://butler.cassandra.apache.org/#/ci/upstream/compare/Cassandra-3.0/cassandra-3.0
The tickets that show up as being responsible for multiple test failures:
trunk:
- https://issues.apache.org/jira/browse/CASSANDRA-17140 (10 total, 6
trunk, 8 4.0, 4 3.0) - I took this ticket
- https://issues.apache.org/jira/browse/CASSANDRA-17311 -
dtest-upgrade.upgrade_tests.upgrade_through_versions_test.TestUpgrade_current_3_0_x_To_indev_3_0_x.test_rolling_upgrade
(6 failures on trunk)
- https://issues.apache.org/jira/browse/CASSANDRA-17213 -
CompactStorageUpgradeTest.compactStorageUpgradeTest fails w/OOM (2
failures)
- https://issues.apache.org/jira/browse/CASSANDRA-17312 -
dtest-large.replace_address_test.TestReplaceAddress.test_restart_failed_replace
(from Cassandra dtests) (2 failures)
- https://issues.apache.org/jira/browse/CASSANDRA-17301 -
Test Failure:
org.apache.cassandra.net.ProxyHandlerConnectionsTest.suddenDisconnect-cdc
(2 failures 4.0)
Otherwise, any of the JIRAs linked to test failures that don't have
assignees can use some work to tidy up our CI.
For reference, we have 51 open JIRA tickets for test failures with no
assignees - link here:
https://issues.apache.org/jira/secure/RapidBoard.jspa?rapidView=496&quickFilter=2252
<https://issues.apache.org/jira/secure/RapidBoard.jspa?rapidView=496&quickFilter=2252>
Thanks in advance!
~Josh
