This is an automated email from the ASF dual-hosted git repository.
devmadhuu pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/ozone.git
The following commit(s) were added to refs/heads/master by this push:
new ec96e716fba HDDS-15223. Stabilize
TestReconAndAdminContainerCLI.testMissingContainer (#10258)
ec96e716fba is described below
commit ec96e716fbada274b8119de1b502cd2f0457cc70
Author: Arun Sarin <[email protected]>
AuthorDate: Thu Jun 4 12:56:27 2026 +0530
HDDS-15223. Stabilize TestReconAndAdminContainerCLI.testMissingContainer
(#10258)
---
.../ozone/recon/TestReconAndAdminContainerCLI.java | 30 ++++++++++++++++------
1 file changed, 22 insertions(+), 8 deletions(-)
diff --git
a/hadoop-ozone/integration-test-recon/src/test/java/org/apache/hadoop/ozone/recon/TestReconAndAdminContainerCLI.java
b/hadoop-ozone/integration-test-recon/src/test/java/org/apache/hadoop/ozone/recon/TestReconAndAdminContainerCLI.java
index a41aeac735e..9514c05d6b2 100644
---
a/hadoop-ozone/integration-test-recon/src/test/java/org/apache/hadoop/ozone/recon/TestReconAndAdminContainerCLI.java
+++
b/hadoop-ozone/integration-test-recon/src/test/java/org/apache/hadoop/ozone/recon/TestReconAndAdminContainerCLI.java
@@ -43,6 +43,7 @@
import java.util.Collections;
import java.util.List;
import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicInteger;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import org.apache.hadoop.hdds.HddsConfigKeys;
@@ -104,6 +105,16 @@ class TestReconAndAdminContainerCLI {
private static final Logger LOG =
LoggerFactory.getLogger(TestReconAndAdminContainerCLI.class);
+ /** Pause between SCM/Recon checks while waiting for matching reports. */
+ private static final int RM_RECON_COMPARE_POLL_INTERVAL_MS = 1000;
+ /** Max wait (Recon can trail SCM briefly). */
+ private static final int RM_RECON_COMPARE_WAIT_MS = 90_000;
+ /**
+ * Two matches in a row on purpose. A single agreeing poll can be luck while
RM and Recon counts
+ * are still drifting past each other (HDDS-15223).
+ */
+ private static final int RM_RECON_COMPARE_STABLE_POLLS = 2;
+
private static final OzoneConfiguration CONF = new OzoneConfiguration();
private static ScmClient scmClient;
private static MiniOzoneCluster cluster;
@@ -186,7 +197,6 @@ static void shutdown() {
* but it's easier to test with Ratis ONE.
*/
@Test
- @Flaky("HDDS-15223")
void testMissingContainer() throws Exception {
String keyNameR1 = "key2";
long containerID = setupRatisKey(recon, keyNameR1,
@@ -311,18 +321,22 @@ void testNodesInDecommissionOrMaintenance(
}
/**
- * The purpose of this method, isn't to validate the numbers
- * but to make sure that they are consistent between
- * Recon and the ReplicationManager.
+ * Checks that SCM's replication manager and Recon show the same unhealthy
stats
+ * (counts and RM sample IDs in Recon's list). Waits until that lines up for
a short
+ * stretch of time so a one-off tick does not hide a real mismatch
(HDDS-15223).
*/
private static void compareRMReportToReconResponse(UnHealthyContainerStates
containerState)
throws Exception {
assertNotNull(containerState);
- // Both threads are running every 1 second.
- // Wait until all values are equal.
- GenericTestUtils.waitFor(() -> assertReportsMatch(containerState),
- 1000, 40000);
+ AtomicInteger stablePolls = new AtomicInteger(0);
+ GenericTestUtils.waitFor(() -> {
+ if (assertReportsMatch(containerState)) {
+ return stablePolls.incrementAndGet() >= RM_RECON_COMPARE_STABLE_POLLS;
+ }
+ stablePolls.set(0);
+ return false;
+ }, RM_RECON_COMPARE_POLL_INTERVAL_MS, RM_RECON_COMPARE_WAIT_MS);
}
private static boolean assertReportsMatch(UnHealthyContainerStates state) {
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
