big-data/hadoop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

HDFS-16320. Datanode retrieve slownode information from NameNode (#3654)

Browse Source
This commit is contained in:
Symious 2021-11-22 18:04:32 +08:00 committed by GitHub
parent 20b78c8f97
commit c88640c4ad
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
9 changed files with 129 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/protocolPB/DatanodeProtocolClientSideTranslatorPB.java
View File

@ -183,7 +183,8 @@ public HeartbeatResponse sendHeartbeat(DatanodeRegistration registration,
rollingUpdateStatus = PBHelperClient.convert(resp.getRollingUpgradeStatus());
}
return new HeartbeatResponse(cmds, PBHelper.convert(resp.getHaStatus()),
rollingUpdateStatus, resp.getFullBlockReportLeaseId());
rollingUpdateStatus, resp.getFullBlockReportLeaseId(),
resp.getIsSlownode());
}
@Override

8
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BPOfferService.java
View File

@ -853,4 +853,12 @@ boolean shouldRetryInit() {
return isAlive();
}
boolean isSlownode() {
for (BPServiceActor actor : bpServices) {
if (actor.isSlownode()) {
return true;
}
}
return false;
}
}

7
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BPServiceActor.java
View File

@ -112,6 +112,7 @@ enum RunningState {
private String nnId = null;
private volatile RunningState runningState = RunningState.CONNECTING;
private volatile boolean shouldServiceRun = true;
private volatile boolean isSlownode = false;
private final DataNode dn;
private final DNConf dnConf;
private long prevBlockReportId;
@ -205,6 +206,7 @@ Map<String, String> getActorInfoMap() {
String.valueOf(getScheduler().getLastBlockReportTime()));
info.put("maxBlockReportSize", String.valueOf(getMaxBlockReportSize()));
info.put("maxDataLength", String.valueOf(maxDataLength));
info.put("isSlownode", String.valueOf(isSlownode));
return info;
}
@ -729,6 +731,7 @@ private void offerService() throws Exception {
handleRollingUpgradeStatus(resp);
}
commandProcessingThread.enqueue(resp.getCommands());
isSlownode = resp.getIsSlownode();
}
}
@ -1474,4 +1477,8 @@ void stopCommandProcessingThread() {
commandProcessingThread.interrupt();
}
}
boolean isSlownode() {
return isSlownode;
}
}

23
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockPoolManager.java
View File

@ -307,4 +307,27 @@ protected BPOfferService createBPOS(
Map<String, BPOfferService> getBpByNameserviceId() {
return bpByNameserviceId;
}
boolean isSlownodeByNameserviceId(String nsId) {
if (bpByNameserviceId.containsKey(nsId)) {
return bpByNameserviceId.get(nsId).isSlownode();
}
return false;
}
boolean isSlownodeByBlockPoolId(String bpId) {
if (bpByBlockPoolId.containsKey(bpId)) {
return bpByBlockPoolId.get(bpId).isSlownode();
}
return false;
}
boolean isSlownode() {
for (BPOfferService bpOfferService : bpByBlockPoolId.values()) {
if (bpOfferService.isSlownode()) {
return true;
}
}
return false;
}
}

12
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/DataNode.java
View File

@ -3814,4 +3814,16 @@ private static boolean isWrite(BlockConstructionStage stage) {
return (stage == PIPELINE_SETUP_STREAMING_RECOVERY
|| stage == PIPELINE_SETUP_APPEND_RECOVERY);
}
boolean isSlownodeByNameserviceId(String nsId) {
return blockPoolManager.isSlownodeByNameserviceId(nsId);
}
boolean isSlownodeByBlockPoolId(String bpId) {
return blockPoolManager.isSlownodeByBlockPoolId(bpId);
}
boolean isSlownode() {
return blockPoolManager.isSlownode();
}
}

5
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSNamesystem.java
View File

@ -4421,8 +4421,11 @@ nodeReg, reports, getBlockPoolId(), cacheCapacity, cacheUsed,
haContext.getState().getServiceState(),
getFSImage().getCorrectLastAppliedOrWrittenTxId());
Set<String> slownodes = DatanodeManager.getSlowNodesUuidSet();
boolean isSlownode = slownodes.contains(nodeReg.getDatanodeUuid());
return new HeartbeatResponse(cmds, haState, rollingUpgradeInfo,
blockReportLeaseId);
blockReportLeaseId, isSlownode);
} finally {
readUnlock("handleHeartbeat");
}

15
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/protocol/HeartbeatResponse.java
View File

@ -36,14 +36,23 @@ public class HeartbeatResponse {
private final RollingUpgradeStatus rollingUpdateStatus;
private final long fullBlockReportLeaseId;
private final boolean isSlownode;
public HeartbeatResponse(DatanodeCommand[] cmds,
NNHAStatusHeartbeat haStatus, RollingUpgradeStatus rollingUpdateStatus,
long fullBlockReportLeaseId) {
this(cmds, haStatus, rollingUpdateStatus, fullBlockReportLeaseId, false);
}
public HeartbeatResponse(DatanodeCommand[] cmds,
NNHAStatusHeartbeat haStatus, RollingUpgradeStatus rollingUpdateStatus,
long fullBlockReportLeaseId, boolean isSlownode) {
commands = cmds;
this.haStatus = haStatus;
this.rollingUpdateStatus = rollingUpdateStatus;
this.fullBlockReportLeaseId = fullBlockReportLeaseId;
this.isSlownode = isSlownode;
}
public DatanodeCommand[] getCommands() {
@ -61,4 +70,8 @@ public RollingUpgradeStatus getRollingUpdateStatus() {
public long getFullBlockReportLeaseId() {
return fullBlockReportLeaseId;
}
public boolean getIsSlownode() {
return isSlownode;
}
}

1
hadoop-hdfs-project/hadoop-hdfs/src/main/proto/DatanodeProtocol.proto
View File

@ -223,6 +223,7 @@ message HeartbeatResponseProto {
optional RollingUpgradeStatusProto rollingUpgradeStatus = 3;
optional RollingUpgradeStatusProto rollingUpgradeStatusV2 = 4;
optional uint64 fullBlockReportLeaseId = 5 [ default = 0 ];
optional bool isSlownode = 6 [ default = false ];
}
/**

59
hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/datanode/TestBPOfferService.java
View File

@ -30,6 +30,7 @@
import static org.apache.hadoop.test.MetricsAsserts.getLongCounter;
import static org.apache.hadoop.test.MetricsAsserts.getMetrics;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertNull;
import static org.junit.Assert.assertNotNull;
import static org.junit.Assert.assertSame;
@ -127,7 +128,8 @@ public class TestBPOfferService {
private final int[] heartbeatCounts = new int[3];
private DataNode mockDn;
private FsDatasetSpi<?> mockFSDataset;
private boolean isSlownode;
@Before
public void setupMocks() throws Exception {
mockNN1 = setupNNMock(0);
@ -216,6 +218,23 @@ public HeartbeatResponse answer(InvocationOnMock invocation)
}
}
private class HeartbeatIsSlownodeAnswer implements Answer<HeartbeatResponse> {
private final int nnIdx;
HeartbeatIsSlownodeAnswer(int nnIdx) {
this.nnIdx = nnIdx;
}
@Override
public HeartbeatResponse answer(InvocationOnMock invocation)
throws Throwable {
HeartbeatResponse heartbeatResponse = new HeartbeatResponse(
datanodeCommands[nnIdx], mockHaStatuses[nnIdx], null,
0, isSlownode);
return heartbeatResponse;
}
}
private class HeartbeatRegisterAnswer implements Answer<HeartbeatResponse> {
private final int nnIdx;
@ -1182,6 +1201,44 @@ public Object answer(InvocationOnMock invocation)
}
}
@Test(timeout = 15000)
public void testSetIsSlownode() throws Exception {
assertEquals(mockDn.isSlownode(), false);
Mockito.when(mockNN1.sendHeartbeat(
Mockito.any(DatanodeRegistration.class),
Mockito.any(StorageReport[].class),
Mockito.anyLong(),
Mockito.anyLong(),
Mockito.anyInt(),
Mockito.anyInt(),
Mockito.anyInt(),
Mockito.any(VolumeFailureSummary.class),
Mockito.anyBoolean(),
Mockito.any(SlowPeerReports.class),
Mockito.any(SlowDiskReports.class)))
.thenAnswer(new HeartbeatIsSlownodeAnswer(0));
BPOfferService bpos = setupBPOSForNNs(mockNN1);
bpos.start();
try {
waitForInitialization(bpos);
bpos.triggerHeartbeatForTests();
assertFalse(bpos.isSlownode());
isSlownode = true;
bpos.triggerHeartbeatForTests();
assertTrue(bpos.isSlownode());
isSlownode = false;
bpos.triggerHeartbeatForTests();
assertFalse(bpos.isSlownode());
} finally {
bpos.stop();
}
}
@Test(timeout = 15000)
public void testCommandProcessingThread() throws Exception {
Configuration conf = new HdfsConfiguration();