Apache
/
hadoop
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

YARN-8579. Recover NMToken of previous attempted component data. 

Contributed by Gour Saha
This commit is contained in:
Eric Yang 2018-07-31 18:01:02 -04:00
parent 5583711419
commit 7640d62716
4 changed files with 24 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-services/hadoop-yarn-services-core/src/main/java/org/apache/hadoop/yarn/service/ServiceScheduler.java
View File

@ -649,6 +649,7 @@ public class ServiceScheduler extends CompositeService {
@Override
public void onContainersReceivedFromPreviousAttempts(
List<Container> containers) {
LOG.info("Containers recovered after AM registered: {}", containers);
if (containers == null || containers.isEmpty()) {
return;
}

1
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/SchedulerApplicationAttempt.java
View File

@ -785,6 +785,7 @@ public class SchedulerApplicationAttempt implements SchedulableEntity {
List<Container> returnContainerList = new ArrayList<>
(recoveredPreviousAttemptContainers);
recoveredPreviousAttemptContainers.clear();
updateNMTokens(returnContainerList);
return returnContainerList;
} finally {
writeLock.unlock();

9
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/fair/FairScheduler.java
View File

@ -32,6 +32,7 @@ import org.apache.hadoop.yarn.api.records.ApplicationId;
import org.apache.hadoop.yarn.api.records.Container;
import org.apache.hadoop.yarn.api.records.ContainerId;
import org.apache.hadoop.yarn.api.records.ContainerStatus;
import org.apache.hadoop.yarn.api.records.NMToken;
import org.apache.hadoop.yarn.api.records.NodeId;
import org.apache.hadoop.yarn.api.records.QueueACL;
import org.apache.hadoop.yarn.api.records.QueueInfo;
@ -912,12 +913,16 @@ public class FairScheduler extends
Resource headroom = application.getHeadroom();
application.setApplicationHeadroomForMetrics(headroom);
List<Container> previousAttemptContainers = application
.pullPreviousAttemptContainers();
List<NMToken> updatedNMTokens = application.pullUpdatedNMTokens();
return new Allocation(newlyAllocatedContainers, headroom,
preemptionContainerIds, null, null,
application.pullUpdatedNMTokens(), null, null,
updatedNMTokens, null, null,
application.pullNewlyPromotedContainers(),
application.pullNewlyDemotedContainers(),
application.pullPreviousAttemptContainers());
previousAttemptContainers);
}
@Override

18
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/applicationsmanager/TestAMRestart.java
View File

@ -1048,12 +1048,12 @@ public class TestAMRestart extends ParameterizedSchedulerTestBase {
rm1.start();
YarnScheduler scheduler = rm1.getResourceScheduler();
MockNM nm1 = new MockNM("127.0.0.1:1234", 10240,
rm1.getResourceTrackerService());
String nm1Address = "127.0.0.1:1234";
MockNM nm1 = new MockNM(nm1Address, 10240, rm1.getResourceTrackerService());
nm1.registerNode();
MockNM nm2 = new MockNM("127.0.0.1:2351", 4089,
rm1.getResourceTrackerService());
String nm2Address = "127.0.0.1:2351";
MockNM nm2 = new MockNM(nm2Address, 4089, rm1.getResourceTrackerService());
nm2.registerNode();
RMApp app1 = rm1.submitApp(200, "name", "user",
@ -1120,6 +1120,11 @@ public class TestAMRestart extends ParameterizedSchedulerTestBase {
registerResponse.getContainersFromPreviousAttempts().size());
Assert.assertEquals("container 2", containerId2,
registerResponse.getContainersFromPreviousAttempts().get(0).getId());
List<NMToken> prevNMTokens = registerResponse
.getNMTokensFromPreviousAttempts();
Assert.assertEquals(1, prevNMTokens.size());
// container 2 is running on node 1
Assert.assertEquals(nm1Address, prevNMTokens.get(0).getNodeId().toString());
rm2.waitForState(app1.getApplicationId(), RMAppState.RUNNING);
@ -1145,6 +1150,11 @@ public class TestAMRestart extends ParameterizedSchedulerTestBase {
allocateResponse.getContainersFromPreviousAttempts());
Assert.assertEquals("new containers should not be allocated",
0, allocateResponse.getAllocatedContainers().size());
List<NMToken> nmTokens = allocateResponse.getNMTokens();
Assert.assertEquals(1, nmTokens.size());
// container 3 is running on node 2
Assert.assertEquals(nm2Address,
nmTokens.get(0).getNodeId().toString());
return true;
}
} catch (Exception e) {