You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-commits@hadoop.apache.org by aa...@apache.org on 2022/03/10 07:39:03 UTC
[hadoop] branch branch-3.2.3 updated: YARN-10538: Add RECOMMISSIONING nodes to the list of updated nodes returned to the AM (#2564)
This is an automated email from the ASF dual-hosted git repository.
aajisaka pushed a commit to branch branch-3.2.3
in repository https://gitbox.apache.org/repos/asf/hadoop.git
The following commit(s) were added to refs/heads/branch-3.2.3 by this push:
new e765f9f YARN-10538: Add RECOMMISSIONING nodes to the list of updated nodes returned to the AM (#2564)
e765f9f is described below
commit e765f9f85ee8163eb630bc14ab45a71d2d688771
Author: srinivasst <40...@users.noreply.github.com>
AuthorDate: Fri Jan 8 10:52:52 2021 +0530
YARN-10538: Add RECOMMISSIONING nodes to the list of updated nodes returned to the AM (#2564)
Contributed by Srinivas S T
(cherry picked from commit 1b1791075a8a45ff8c7d25d52cb014d0ce858cc2)
(cherry picked from commit 98565b6c60c4c1cefaabe4a1fa5fe92158b19b97)
(cherry picked from commit c64eda080e6ed0d47fbc91a22050612e1d4c80d3)
---
.../server/resourcemanager/rmnode/RMNodeImpl.java | 7 +++
.../TestAMRMRPCNodeUpdates.java | 54 ++++++++++++++++++++++
2 files changed, 61 insertions(+)
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeImpl.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeImpl.java
index 0594207..d296854 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeImpl.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeImpl.java
@@ -1302,6 +1302,13 @@ public class RMNodeImpl implements RMNode, EventHandler<RMNodeEvent> {
.handle(
new NodeResourceUpdateSchedulerEvent(rmNode, ResourceOption
.newInstance(rmNode.totalCapability, 0)));
+
+ // Notify NodesListManager to notify all RMApp that this node has been
+ // recommissioned so that each Application Master can take any required
+ // actions.
+ rmNode.context.getDispatcher().getEventHandler().handle(
+ new NodesListManagerEvent(
+ NodesListManagerEventType.NODE_USABLE, rmNode));
}
}
diff --git a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/applicationsmanager/TestAMRMRPCNodeUpdates.java b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/applicationsmanager/TestAMRMRPCNodeUpdates.java
index dd9bcd0..c846da1 100644
--- a/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/applicationsmanager/TestAMRMRPCNodeUpdates.java
+++ b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/applicationsmanager/TestAMRMRPCNodeUpdates.java
@@ -39,6 +39,7 @@ import org.apache.hadoop.yarn.server.resourcemanager.MockNM;
import org.apache.hadoop.yarn.server.resourcemanager.MockRM;
import org.apache.hadoop.yarn.server.resourcemanager.rmapp.RMApp;
import org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt.RMAppAttempt;
+import org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNodeEventType;
import org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacitySchedulerConfiguration;
import org.junit.After;
import org.junit.Before;
@@ -90,6 +91,12 @@ public class TestAMRMRPCNodeUpdates {
rm.drainEvents();
}
+ private void syncNodeRecommissioning(MockNM nm) throws Exception {
+ rm.sendNodeEvent(nm, RMNodeEventType.RECOMMISSION);
+ rm.waitForState(nm.getNodeId(), NodeState.RUNNING);
+ rm.drainEvents();
+ }
+
private AllocateResponse allocate(final ApplicationAttemptId attemptId,
final AllocateRequest req) throws Exception {
UserGroupInformation ugi =
@@ -140,6 +147,53 @@ public class TestAMRMRPCNodeUpdates {
}
@Test
+ public void testAMRMRecommissioningNodes() throws Exception {
+ MockNM nm1 = rm.registerNode("127.0.0.1:1234", 10000);
+ MockNM nm2 = rm.registerNode("127.0.0.2:1234", 10000);
+ rm.drainEvents();
+
+ RMApp app1 = rm.submitApp(2000);
+
+ // Trigger the scheduling so the AM gets 'launched' on nm1
+ nm1.nodeHeartbeat(true);
+
+ RMAppAttempt attempt1 = app1.getCurrentAppAttempt();
+ MockAM am1 = rm.sendAMLaunched(attempt1.getAppAttemptId());
+
+ // register AM returns no unusable node
+ am1.registerAppAttempt();
+
+ // DECOMMISSION nm2
+ Integer decommissioningTimeout = 600;
+ syncNodeGracefulDecommission(nm2, decommissioningTimeout);
+
+ AllocateRequest allocateRequest1 =
+ AllocateRequest.newInstance(0, 0F, null, null, null);
+ AllocateResponse response1 =
+ allocate(attempt1.getAppAttemptId(), allocateRequest1);
+ List<NodeReport> updatedNodes = response1.getUpdatedNodes();
+ Assert.assertEquals(1, updatedNodes.size());
+ NodeReport nr = updatedNodes.iterator().next();
+ Assert.assertEquals(
+ decommissioningTimeout, nr.getDecommissioningTimeout());
+ Assert.assertEquals(
+ NodeUpdateType.NODE_DECOMMISSIONING, nr.getNodeUpdateType());
+
+ // Wait for nm2 to RECOMMISSION
+ syncNodeRecommissioning(nm2);
+
+ AllocateRequest allocateRequest2 = AllocateRequest
+ .newInstance(response1.getResponseId(), 0F, null, null, null);
+ AllocateResponse response2 =
+ allocate(attempt1.getAppAttemptId(), allocateRequest2);
+ List<NodeReport> updatedNodes2 = response2.getUpdatedNodes();
+ Assert.assertEquals(1, updatedNodes2.size());
+ NodeReport nr2 = updatedNodes2.iterator().next();
+ Assert.assertEquals(
+ NodeUpdateType.NODE_USABLE, nr2.getNodeUpdateType());
+ }
+
+ @Test
public void testAMRMUnusableNodes() throws Exception {
MockNM nm1 = rm.registerNode("127.0.0.1:1234", 10000);
---------------------------------------------------------------------
To unsubscribe, e-mail: common-commits-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-commits-help@hadoop.apache.org