(druid) branch master updated: minor: reduce flakiness in CI (#19706)

[tu...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

jtuglu1 pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/druid.git


The following commit(s) were added to refs/heads/master by this push:
     new 48647470fec minor: reduce flakiness in CI (#19706)
48647470fec is described below

commit 48647470fec00f00d9fbf21c52d46bc0fae072ac
Author: jtuglu1 <jt...@netflix.com>
AuthorDate: Mon Jul 20 14:34:49 2026 -0700

    minor: reduce flakiness in CI (#19706)
    
    We have a few flaky tests in our own CI environment that fail (I've also seen failures in OSS).
    
    1. I've observed that under a heavily-subscribed CI machine the KafkaIndexFaultToleranceTest.test_supervisorRecovers_afterChangeInTopicPartitions test can sometimes timeout under the current 120s limit. Increasing this to allow up to 240s has removed these failures for us.
    
    2. Adjust the TLS tests to select ports that are not actually in-use. This should prevent any races that might occur from running tests at a higher concurrency in the future.
---
 .../testing/embedded/indexing/StreamIndexFaultToleranceTest.java | 2 +-
 .../apache/druid/server/initialization/JettyCertRenewTest.java   | 9 +++++++--
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/embedded-tests/src/test/java/org/apache/druid/testing/embedded/indexing/StreamIndexFaultToleranceTest.java b/embedded-tests/src/test/java/org/apache/druid/testing/embedded/indexing/StreamIndexFaultToleranceTest.java
index 89219c58317..c315969e5f5 100644
--- a/embedded-tests/src/test/java/org/apache/druid/testing/embedded/indexing/StreamIndexFaultToleranceTest.java
+++ b/embedded-tests/src/test/java/org/apache/druid/testing/embedded/indexing/StreamIndexFaultToleranceTest.java
@@ -53,7 +53,7 @@ public abstract class StreamIndexFaultToleranceTest extends StreamIndexTestBase
   @Override
   protected EmbeddedDruidCluster createCluster()
   {
-    return super.createCluster().useDefaultTimeoutForLatchableEmitter(120);
+    return super.createCluster().useDefaultTimeoutForLatchableEmitter(240);
   }
 
   @BeforeEach
diff --git a/server/src/test/java/org/apache/druid/server/initialization/JettyCertRenewTest.java b/server/src/test/java/org/apache/druid/server/initialization/JettyCertRenewTest.java
index 6f29d1c0036..2f48f1fcf6a 100644
--- a/server/src/test/java/org/apache/druid/server/initialization/JettyCertRenewTest.java
+++ b/server/src/test/java/org/apache/druid/server/initialization/JettyCertRenewTest.java
@@ -27,6 +27,7 @@ import com.google.inject.Key;
 import com.google.inject.Module;
 import com.google.inject.multibindings.Multibinder;
 import org.apache.commons.io.IOUtils;
+import org.apache.druid.common.utils.SocketUtil;
 import org.apache.druid.guice.GuiceInjectors;
 import org.apache.druid.guice.Jerseys;
 import org.apache.druid.guice.JsonConfigProvider;
@@ -135,7 +136,11 @@ public class JettyCertRenewTest extends BaseJettyTest
       throw new RuntimeException(e);
     }
 
-    final int ephemeralPort = ThreadLocalRandom.current().nextInt(49152, 65535);
+    // Pick ports that are actually bindable rather than guessing a random one: with reused forks and
+    // many concurrent test shards a blind random port frequently collides (BindException). Verify the
+    // plaintext and TLS ports independently since both are enabled below.
+    final int ephemeralPort = SocketUtil.findOpenPortFrom(ThreadLocalRandom.current().nextInt(49152, 60000));
+    final int tlsEphemeralPort = SocketUtil.findOpenPortFrom(ephemeralPort + 1);
 
     latchedRequestState = new LatchedRequestStateHolder();
     injector = Initialization.makeInjectorWithModules(
@@ -149,7 +154,7 @@ public class JettyCertRenewTest extends BaseJettyTest
                 JsonConfigProvider.bindInstance(
                     binder,
                     Key.get(DruidNode.class, Self.class),
-                    new DruidNode("test", "localhost", false, ephemeralPort, ephemeralPort + 1, true, true)
+                    new DruidNode("test", "localhost", false, ephemeralPort, tlsEphemeralPort, true, true)
                 );
                 binder.bind(TLSServerConfig.class).toInstance(tlsConfig);
                 binder.bind(JettyServerInitializer.class).to(JettyServerInit.class).in(LazySingleton.class);


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org