Nick Shanin created SOLR-18531:
----------------------------------
Summary: LeaderElectionIntegrationTest fails intermittently with
BindException because its restart port is taken during the restart gap
Key: SOLR-18531
URL: https://issues.apache.org/jira/browse/SOLR-18531
Project: Solr
Issue Type: Test
Components: Tests
Reporter: Nick Shanin
AI text below
h2. What happens
{\{LeaderElectionIntegrationTest.testSimpleSliceLeaderElection}} fails
intermittently in CI with a \{{BindException}} ("Address already in use") when
\{{JettySolrRunner.start}} restarts a node on its pinned port. Seen on an
apache/solr CI run for an unrelated change: [run
37991927523|[https://github.com/apache/solr/actions/runs/37991927523]], seed
\{{3E3D9FF553211ED6}}.
h2. Likely mechanism
The runner releases the port when the node stops and rebinds the same port when
it starts again. In the gap, a process outside the JVM can take the port, and
the test framework holds no reservation, so the restart fails. Any suite that
restarts a \{{JettySolrRunner}} on a fixed port is exposed, not only this test.
h2. Proposed fix
In the test framework, hold the released port across a restart: reserve it on
stop and hand it to the start, so the window where another process can take it
closes.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
