siddhantsangwan commented on code in PR #8491:
URL: https://github.com/apache/ozone/pull/8491#discussion_r2102430321
##########
hadoop-hdds/server-scm/src/test/java/org/apache/hadoop/hdds/scm/container/balancer/TestMoveManager.java:
##########
@@ -496,6 +499,44 @@ public void testMoveCompleteFutureReplicasUnhealthy()
throws Exception {
.sendDeleteCommand(eq(containerInfo), eq(0), eq(src), eq(true));
}
+ @Test
+ public void testDeleteNotSentWithExpirationTimeInPast() throws Exception {
+ containerInfo = ReplicationTestUtil.createContainer(
+ HddsProtos.LifeCycleState.CLOSED, new ECReplicationConfig(3, 2));
+ setupMocks();
+
+ replicas.addAll(ReplicationTestUtil
+ .createReplicas(containerInfo.containerID(), 1, 2, 3, 4, 5));
+ Iterator<ContainerReplica> iterator = replicas.iterator();
+ ContainerReplica srcReplica = iterator.next();
+ src = srcReplica.getDatanodeDetails();
+ tgt = MockDatanodeDetails.randomDatanodeDetails();
+ nodes.put(src, NodeStatus.inServiceHealthy());
+ nodes.put(tgt, NodeStatus.inServiceHealthy());
+
+ CompletableFuture<MoveManager.MoveResult> res =
+ moveManager.move(containerInfo.containerID(), src, tgt);
+ ContainerReplicaOp op = new ContainerReplicaOp(
+ ADD, tgt, srcReplica.getReplicaIndex(), null, clock.millis() + 1000);
+ moveManager.opCompleted(op, containerInfo.containerID(), false);
+
+ ArgumentCaptor<Long> longCaptor = ArgumentCaptor.forClass(Long.class);
+ verify(replicationManager).sendDeleteCommand(
+ eq(containerInfo), eq(srcReplica.getReplicaIndex()), eq(src),
+ eq(true), longCaptor.capture());
+
+ // 6 minutes is the datanodeTimeoutOffset set for datanodeCommands sent by
replicationManager by default
+ assertTrue((Duration.ofMillis(longCaptor.getValue()).toMillis()
+ - Duration.ofMinutes(6).toMillis()) > clock.millis());
Review Comment:
Thanks for adding the test. We want to ensure it asserts the delete command
is sent with an SCM deadline of `moveStartTime + moveTimeout`, so the assertion
needs to be changed. The 6-minute `datanodeTimeoutOffset` is used later when
the Replication Manager sends the command to the datanode, so it's not relevant
here.
It'd also be good to have a test that reproduces the situation where a
delete command was being sent with a deadline in the past, and make sure that
doesn't happen with the new changes. The example I added in the jira can be a
good guide for you to reproduce the error.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
