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

HDDS-797. If DN is started before SCM, it does not register. Contributed by Hanisha Koneru.

Browse Source
This commit is contained in:
Arpit Agarwal 2018-11-05 09:40:00 -08:00
parent 15df2e7a75
commit c8ca1747c0
2 changed files with 95 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/states/endpoint/VersionEndpointTask.java
View File

@ -64,6 +64,8 @@ public VersionEndpointTask(EndpointStateMachine rpcEndPoint,
public EndpointStateMachine.EndPointStates call() throws Exception {
rpcEndPoint.lock();
try{
if (rpcEndPoint.getState().equals(
EndpointStateMachine.EndPointStates.GETVERSION)) {
SCMVersionResponseProto versionResponse =
rpcEndPoint.getEndPoint().getVersion(null);
VersionResponse response = VersionResponse.getFromProtobuf(
@ -85,6 +87,7 @@ public EndpointStateMachine.EndPointStates call() throws Exception {
"cannot be null");
// If version file does not exist create version file and also set scmId
for (Map.Entry<String, HddsVolume> entry : volumeMap.entrySet()) {
HddsVolume hddsVolume = entry.getValue();
boolean result = HddsVolumeUtil.checkVolume(hddsVolume, scmId,
@ -108,6 +111,10 @@ public EndpointStateMachine.EndPointStates call() throws Exception {
rpcEndPoint.getState().getNextState();
rpcEndPoint.setState(nextState);
rpcEndPoint.zeroMissedCount();
} else {
LOG.debug("Cannot execute GetVersion task as endpoint state machine " +
"is in {} state", rpcEndPoint.getState());
}
} catch (DiskOutOfSpaceException ex) {
rpcEndPoint.setState(EndpointStateMachine.EndPointStates.SHUTDOWN);
} catch(IOException ex) {

52
hadoop-ozone/integration-test/src/test/java/org/apache/hadoop/ozone/TestMiniOzoneCluster.java
View File

@ -23,9 +23,13 @@
import org.apache.hadoop.hdds.conf.OzoneConfiguration;
import org.apache.hadoop.hdds.protocol.DatanodeDetails;
import org.apache.hadoop.hdds.scm.pipeline.PipelineID;
import org.apache.hadoop.hdds.scm.server.StorageContainerManager;
import org.apache.hadoop.ozone.container.common.SCMTestUtils;
import org.apache.hadoop.ozone.container.common.helpers.ContainerUtils;
import org.apache.hadoop.ozone.container.common.statemachine.DatanodeStateMachine;
import org.apache.hadoop.ozone.container.common.statemachine
.DatanodeStateMachine;
import org.apache.hadoop.ozone.container.common.statemachine
.EndpointStateMachine;
import org.apache.hadoop.ozone.container.ozoneimpl.TestOzoneContainer;
import org.apache.hadoop.hdds.protocol.proto.HddsProtos;
import org.apache.hadoop.hdds.scm.TestUtils;
@ -214,4 +218,50 @@ private void createMalformedIDFile(File malformedFile)
out.write("malformed".getBytes());
out.close();
}
/**
* Test that a DN can register with SCM even if it was started before the SCM.
* @throws Exception
*/
@Test (timeout = 300_000)
public void testDNstartAfterSCM() throws Exception {
// Start a cluster with 1 DN
cluster = MiniOzoneCluster.newBuilder(conf)
.setNumDatanodes(1)
.build();
cluster.waitForClusterToBeReady();
// Stop the SCM
StorageContainerManager scm = cluster.getStorageContainerManager();
scm.stop();
// Restart DN
cluster.restartHddsDatanode(0, false);
// DN should be in GETVERSION state till the SCM is restarted.
// Check DN endpoint state for 20 seconds
DatanodeStateMachine dnStateMachine = cluster.getHddsDatanodes().get(0)
.getDatanodeStateMachine();
for (int i = 0; i < 20; i++) {
for (EndpointStateMachine endpoint :
dnStateMachine.getConnectionManager().getValues()) {
Assert.assertEquals(
EndpointStateMachine.EndPointStates.GETVERSION,
endpoint.getState());
}
Thread.sleep(1000);
}
// DN should successfully register with the SCM after SCM is restarted.
// Restart the SCM
cluster.restartStorageContainerManager();
// Wait for DN to register
cluster.waitForClusterToBeReady();
// DN should be in HEARTBEAT state after registering with the SCM
for (EndpointStateMachine endpoint :
dnStateMachine.getConnectionManager().getValues()) {
Assert.assertEquals(EndpointStateMachine.EndPointStates.HEARTBEAT,
endpoint.getState());
}
}
}