HDFS-16764. [SBN Read] ObserverNamenode should throw ObserverRetryOnActiveException instead of FileNotFoundException during processing of addBlock rpc (#4872)

Signed-off-by: Erik Krogen <xkrogen@apache.org>
Co-authored-by: zengqiang.xu <zengqiang.xu@shopee.com>

(cherry picked from commit 8d221255f2)
This commit is contained in:
ZanderXu 2022-12-21 07:50:58 +08:00 committed by Erik Krogen
parent b6fd322307
commit a49378cb93
2 changed files with 54 additions and 4 deletions

View File

@ -2968,12 +2968,12 @@ public class FSNamesystem implements Namesystem, FSNamesystemMBean,
LocatedBlock[] onRetryBlock = new LocatedBlock[1];
FSDirWriteFileOp.ValidateAddBlockResult r;
checkOperation(OperationCategory.READ);
checkOperation(OperationCategory.WRITE);
final FSPermissionChecker pc = getPermissionChecker();
FSPermissionChecker.setOperationType(operationName);
readLock();
try {
checkOperation(OperationCategory.READ);
checkOperation(OperationCategory.WRITE);
r = FSDirWriteFileOp.validateAddBlock(this, pc, src, fileId, clientName,
previous, onRetryBlock);
} finally {
@ -3019,12 +3019,15 @@ public class FSNamesystem implements Namesystem, FSNamesystemMBean,
final byte storagePolicyID;
final List<DatanodeStorageInfo> chosen;
final BlockType blockType;
checkOperation(OperationCategory.READ);
checkOperation(OperationCategory.WRITE);
final FSPermissionChecker pc = getPermissionChecker();
FSPermissionChecker.setOperationType(null);
readLock();
try {
checkOperation(OperationCategory.READ);
// Changing this operation category to WRITE instead of making getAdditionalDatanode as a
// read method is aim to let Active NameNode to handle this RPC, because Active NameNode
// contains a more complete DN selection context than Observer NameNode.
checkOperation(OperationCategory.WRITE);
//check safe mode
checkNameNodeSafeMode("Cannot add datanode; src=" + src + ", blk=" + blk);
final INodesInPath iip = dir.resolvePath(pc, src, fileId);

View File

@ -28,22 +28,32 @@ import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_IP_PROXY_USERS;
import static org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_RPC_BIND_HOST_KEY;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertNotEquals;
import static org.junit.Assert.assertNull;
import static org.junit.Assert.assertNotNull;
import java.io.IOException;
import java.nio.charset.StandardCharsets;
import java.util.EnumSet;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hdfs.AddBlockFlag;
import org.apache.hadoop.hdfs.DFSTestUtil;
import org.apache.hadoop.hdfs.DistributedFileSystem;
import org.apache.hadoop.hdfs.HdfsConfiguration;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.hdfs.protocol.HdfsFileStatus;
import org.apache.hadoop.hdfs.protocol.LocatedBlocks;
import org.apache.hadoop.hdfs.qjournal.MiniQJMHACluster;
import org.apache.hadoop.ipc.CallerContext;
import org.apache.hadoop.ipc.ObserverRetryOnActiveException;
import org.apache.hadoop.security.UserGroupInformation;
import org.apache.hadoop.test.GenericTestUtils;
import org.apache.hadoop.test.LambdaTestUtils;
import org.junit.Test;
import org.junit.jupiter.api.Timeout;
public class TestNameNodeRpcServer {
@ -91,6 +101,43 @@ public class TestNameNodeRpcServer {
// trials. 1/3^20=3e-10, so that should be good enough.
static final int ITERATIONS_TO_USE = 20;
@Test
@Timeout(30000)
public void testObserverHandleAddBlock() throws Exception {
String baseDir = GenericTestUtils.getRandomizedTempPath();
Configuration conf = new HdfsConfiguration();
MiniQJMHACluster.Builder builder = new MiniQJMHACluster.Builder(conf).setNumNameNodes(3);
builder.getDfsBuilder().numDataNodes(3);
try (MiniQJMHACluster qjmhaCluster = builder.baseDir(baseDir).build()) {
MiniDFSCluster dfsCluster = qjmhaCluster.getDfsCluster();
dfsCluster.waitActive();
dfsCluster.transitionToActive(0);
dfsCluster.transitionToObserver(2);
NameNode activeNN = dfsCluster.getNameNode(0);
NameNode observerNN = dfsCluster.getNameNode(2);
// Stop the editLogTailer of Observer NameNode
observerNN.getNamesystem().getEditLogTailer().stop();
DistributedFileSystem dfs = dfsCluster.getFileSystem(0);
Path testPath = new Path("/testObserverHandleAddBlock/file.txt");
try (FSDataOutputStream ignore = dfs.create(testPath)) {
HdfsFileStatus fileStatus = activeNN.getRpcServer().getFileInfo(testPath.toUri().getPath());
assertNotNull(fileStatus);
assertNull(observerNN.getRpcServer().getFileInfo(testPath.toUri().getPath()));
LambdaTestUtils.intercept(ObserverRetryOnActiveException.class, () -> {
observerNN.getRpcServer().addBlock(testPath.toUri().getPath(),
dfs.getClient().getClientName(), null, null,
fileStatus.getFileId(), null, EnumSet.noneOf(AddBlockFlag.class));
});
} finally {
dfs.delete(testPath, true);
}
}
}
/**
* A test to make sure that if an authorized user adds "clientIp:" to their
* caller context, it will be used to make locality decisions on the NN.