This is an automated email from the ASF dual-hosted git repository.
jxue pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/helix.git
The following commit(s) were added to refs/heads/master by this push:
new d0a531387 Extend the timeout for flaky as slowness of the machine and
add logs for TestRoutingTableProviderFromCurrentStates
d0a531387 is described below
commit d0a5313877ca53703bc76cf11ee4b115add523f4
Author: Junkai Xue <[email protected]>
AuthorDate: Wed Feb 25 22:49:59 2026 -0800
Extend the timeout for flaky as slowness of the machine and add logs for
TestRoutingTableProviderFromCurrentStates
1. After cluster verification - logs when rebalance completes
2. Waiting for routing table update - logs when waiting for MASTER
assignment
3. If update fails - logs what instances are actually present
4. After participant shutdown - logs which node was stopped
5. After cluster stabilization - logs when verify passes
6. Before validation steps - logs which step is running
7. If routing tables don't match - logs ideal state partitions
Now when it fails in CI, you'll see detailed logs like:
INFO: Cluster verified after rebalance
INFO: Waiting for routing table to update for partition TestDB-0
INFO: Target node with MASTER: localhost_12918
INFO: Shutdown participant: localhost_12918
INFO: Waiting for cluster to stabilize after shutdown
INFO: Cluster verified after shutdown
INFO: Validate after step 4
ERROR: Routing tables did not match between EV and CS providers
ERROR: IdealState partitions: [TestDB-0, TestDB-1, ...]
---
.../TestRoutingTableProviderFromCurrentStates.java | 41 ++++++++++++++++++----
1 file changed, 34 insertions(+), 7 deletions(-)
diff --git
a/helix-core/src/test/java/org/apache/helix/integration/spectator/TestRoutingTableProviderFromCurrentStates.java
b/helix-core/src/test/java/org/apache/helix/integration/spectator/TestRoutingTableProviderFromCurrentStates.java
index cbf299860..6e5a3357d 100644
---
a/helix-core/src/test/java/org/apache/helix/integration/spectator/TestRoutingTableProviderFromCurrentStates.java
+++
b/helix-core/src/test/java/org/apache/helix/integration/spectator/TestRoutingTableProviderFromCurrentStates.java
@@ -30,6 +30,9 @@ import javax.management.MBeanServer;
import javax.management.MalformedObjectNameException;
import javax.management.ObjectName;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
import com.google.common.collect.ImmutableMap;
import org.apache.helix.ConfigAccessor;
import org.apache.helix.HelixException;
@@ -72,6 +75,7 @@ import org.testng.annotations.BeforeClass;
import org.testng.annotations.Test;
public class TestRoutingTableProviderFromCurrentStates extends ZkTestBase {
+ private static final Logger LOG =
LoggerFactory.getLogger(TestRoutingTableProviderFromCurrentStates.class);
private HelixManager _manager;
private final int NUM_NODES = 10;
protected int NUM_PARTITIONS = 20;
@@ -283,7 +287,7 @@ public class TestRoutingTableProviderFromCurrentStates
extends ZkTestBase {
.getResourcesInCluster(CLUSTER_NAME)) {
IdealState idealState = _gSetupTool.getClusterManagementTool()
.getResourceIdealState(CLUSTER_NAME, resourceName);
- validate(idealState, routingTableEV, routingTableCS, 3000);
+ validate(idealState, routingTableEV, routingTableCS, 10000);
}
// Blocking the current state event processing
@@ -299,40 +303,63 @@ public class TestRoutingTableProviderFromCurrentStates
extends ZkTestBase {
new
BestPossibleExternalViewVerifier.Builder(CLUSTER_NAME).setZkClient(_gZkClient)
.setWaitTillVerify(TestHelper.DEFAULT_REBALANCE_PROCESSING_WAIT_TIME)
.build();
- Assert.assertTrue(clusterVerifier.verifyByPolling(5000, 500));
+ Assert.assertTrue(clusterVerifier.verifyByPolling(30000, 500));
+ LOG.info("Cluster verified after rebalance");
// 2. Process one event, so the current state will be refreshed with the
new DB partitions
routingTableCS.proceedNewEventHandling();
IdealState idealState =
_gSetupTool.getClusterManagementTool().getResourceIdealState(CLUSTER_NAME, db);
String targetPartitionName =
idealState.getPartitionSet().iterator().next();
+ LOG.info("Waiting for routing table to update for partition {}",
targetPartitionName);
// Wait until the routing table is updated.
BlockingCurrentStateRoutingTableProvider finalRoutingTableCS =
routingTableCS;
- Assert.assertTrue(TestHelper.verify(
+ boolean hasMaster = TestHelper.verify(
() -> !finalRoutingTableCS.getInstances(db, targetPartitionName,
"MASTER").isEmpty(),
- 2000));
+ 10000);
+ if (!hasMaster) {
+ LOG.error("Routing table did not update - no MASTER found for
partition {} in resource {}", targetPartitionName, db);
+ LOG.error("Current instances: {}",
finalRoutingTableCS.getInstances(db, targetPartitionName, "MASTER"));
+ }
+ Assert.assertTrue(hasMaster);
String targetNodeName =
routingTableCS.getInstances(db, targetPartitionName,
"MASTER").get(0).getInstanceName();
+ LOG.info("Target node with MASTER: {}", targetNodeName);
// 3. Shutdown one of the instance that contains a master partition
for (int i = 0; i < _participants.length; i++) {
if (_participants[i].getInstanceName().equals(targetNodeName)) {
shutdownParticipantIndex = i;
_participants[i].syncStop();
+ LOG.info("Shutdown participant: {}", targetNodeName);
}
}
+ LOG.info("Waiting for cluster to stabilize after shutdown");
Assert.assertTrue(clusterVerifier.verifyByPolling());
+ LOG.info("Cluster verified after shutdown");
// 4. Process one of the stale current state event.
// The expectation is that, the provider will refresh with all the
latest data including the
// the live instance change. Even the corresponding ZK event has not
been processed yet.
routingTableCS.proceedNewEventHandling();
+ LOG.info("Validate after step 4");
validate(idealState, routingTableEV, routingTableCS, 3000);
// 5. Unblock the event processing and let the provider to process all
events.
// The expectation is that, the eventual routing tables are still the
same.
routingTableCS.setBlocking(false);
routingTableCS.proceedNewEventHandling();
- // Wait for a short while so the router will process at least one event.
- Thread.sleep(500);
+ LOG.info("Validate after unblock (step 5)");
+ // Wait for routing table to converge using polling instead of hardcoded
sleep
+ // Use copies for lambda since routingTableCS is modified
+ RoutingTableProvider routingTableEVCopy = routingTableEV;
+ RoutingTableProvider routingTableCSCopy = routingTableCS;
+ boolean matched = TestHelper.verify(
+ () -> compare(idealState, routingTableEVCopy, routingTableCSCopy),
+ 10000);
+ if (!matched) {
+ LOG.error("Routing tables did not match between EV and CS providers");
+ LOG.error("IdealState partitions: {}", idealState.getPartitionSet());
+ }
+ Assert.assertTrue(matched);
// Confirm that 2 providers match eventually
- validate(idealState, routingTableEV, routingTableCS, 2000);
+ validate(idealState, routingTableEV, routingTableCS, 10000);
} finally {
if (routingTableCS != null) {
routingTableCS.setBlocking(false);
