HDFS-8827. Erasure Coding: Fix NPE when NameNode processes over-replicated striped blocks. Contributed by Walter Su and Takuya Fukudome.

This commit is contained in:
Jing Zhao 2015-08-07 11:25:20 -07:00
parent 8799363db1
commit fbf7e81ca0
3 changed files with 168 additions and 9 deletions

View File

@ -391,3 +391,6 @@
HDFS-8857. Erasure Coding: Fix ArrayIndexOutOfBoundsException in
TestWriteStripedFileWithFailure. (Li Bo)
HDFS-8827. Erasure Coding: Fix NPE when NameNode processes over-replicated
striped blocks. (Walter Su and Takuya Fukudome via jing9)

View File

@ -3135,14 +3135,13 @@ private void chooseExcessReplicates(
assert namesystem.hasWriteLock();
// first form a rack to datanodes map and
BlockCollection bc = getBlockCollection(storedBlock);
final BlockStoragePolicy storagePolicy = storagePolicySuite.getPolicy(
bc.getStoragePolicyID());
final List<StorageType> excessTypes = storagePolicy.chooseExcess(
replication, DatanodeStorageInfo.toStorageTypes(nonExcess));
if (storedBlock.isStriped()) {
chooseExcessReplicasStriped(bc, nonExcess, storedBlock, delNodeHint,
excessTypes);
chooseExcessReplicasStriped(bc, nonExcess, storedBlock, delNodeHint);
} else {
final BlockStoragePolicy storagePolicy = storagePolicySuite.getPolicy(
bc.getStoragePolicyID());
final List<StorageType> excessTypes = storagePolicy.chooseExcess(
replication, DatanodeStorageInfo.toStorageTypes(nonExcess));
chooseExcessReplicasContiguous(bc, nonExcess, storedBlock,
replication, addedNode, delNodeHint, excessTypes);
}
@ -3216,8 +3215,7 @@ private void chooseExcessReplicasContiguous(BlockCollection bc,
private void chooseExcessReplicasStriped(BlockCollection bc,
final Collection<DatanodeStorageInfo> nonExcess,
BlockInfo storedBlock,
DatanodeDescriptor delNodeHint,
List<StorageType> excessTypes) {
DatanodeDescriptor delNodeHint) {
assert storedBlock instanceof BlockInfoStriped;
BlockInfoStriped sblk = (BlockInfoStriped) storedBlock;
short groupSize = sblk.getTotalBlockNum();
@ -3237,6 +3235,14 @@ private void chooseExcessReplicasStriped(BlockCollection bc,
found.set(index);
storage2index.put(storage, index);
}
// the number of target left replicas equals to the of number of the found
// indices.
int numOfTarget = found.cardinality();
final BlockStoragePolicy storagePolicy = storagePolicySuite.getPolicy(
bc.getStoragePolicyID());
final List<StorageType> excessTypes = storagePolicy.chooseExcess(
(short)numOfTarget, DatanodeStorageInfo.toStorageTypes(nonExcess));
// use delHint only if delHint is duplicated
final DatanodeStorageInfo delStorageHint =

View File

@ -24,9 +24,14 @@
import org.apache.hadoop.hdfs.DistributedFileSystem;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.hdfs.protocol.Block;
import org.apache.hadoop.hdfs.protocol.DatanodeInfo;
import org.apache.hadoop.hdfs.protocol.HdfsConstants;
import org.apache.hadoop.hdfs.protocol.LocatedBlocks;
import org.apache.hadoop.hdfs.protocol.LocatedStripedBlock;
import org.apache.hadoop.hdfs.server.blockmanagement.BlockInfoStriped;
import org.apache.hadoop.hdfs.server.blockmanagement.BlockManager;
import org.apache.hadoop.hdfs.server.blockmanagement.BlockManagerTestUtil;
import org.apache.hadoop.hdfs.server.datanode.DataNode;
import org.apache.hadoop.hdfs.server.datanode.SimulatedFSDataset;
import org.junit.After;
import org.junit.Before;
@ -35,6 +40,7 @@
import java.io.IOException;
import java.util.Arrays;
import java.util.HashSet;
import java.util.List;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertTrue;
@ -49,7 +55,7 @@ public class TestAddOverReplicatedStripedBlocks {
private final short PARITY_BLK_NUM = HdfsConstants.NUM_PARITY_BLOCKS;
private final short GROUP_SIZE = DATA_BLK_NUM + PARITY_BLK_NUM;
private final int CELLSIZE = HdfsConstants.BLOCK_STRIPED_CELL_SIZE;
private final int NUM_STRIPE_PER_BLOCK = 1;
private final int NUM_STRIPE_PER_BLOCK = 4;
private final int BLOCK_SIZE = NUM_STRIPE_PER_BLOCK * CELLSIZE;
private final int numDNs = GROUP_SIZE + 3;
@ -57,6 +63,8 @@ public class TestAddOverReplicatedStripedBlocks {
public void setup() throws IOException {
Configuration conf = new Configuration();
conf.setLong(DFSConfigKeys.DFS_BLOCK_SIZE_KEY, BLOCK_SIZE);
// disable block recovery
conf.setInt(DFSConfigKeys.DFS_NAMENODE_REPLICATION_MAX_STREAMS_KEY, 0);
SimulatedFSDataset.setFactory(conf);
cluster = new MiniDFSCluster.Builder(conf).numDataNodes(numDNs).build();
cluster.waitActive();
@ -113,4 +121,146 @@ public void testProcessOverReplicatedStripedBlock() throws Exception {
filePath.toString(), 0, fileLen);
DFSTestUtil.verifyLocatedStripedBlocks(lbs, GROUP_SIZE);
}
@Test
public void testProcessOverReplicatedSBSmallerThanFullBlocks()
throws Exception {
// Create a EC file which doesn't fill full internal blocks.
int fileLen = CELLSIZE * (DATA_BLK_NUM - 1);
byte[] content = new byte[fileLen];
DFSTestUtil.writeFile(fs, filePath, new String(content));
LocatedBlocks lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
LocatedStripedBlock bg = (LocatedStripedBlock) (lbs.get(0));
long gs = bg.getBlock().getGenerationStamp();
String bpid = bg.getBlock().getBlockPoolId();
long groupId = bg.getBlock().getBlockId();
Block blk = new Block(groupId, BLOCK_SIZE, gs);
cluster.triggerBlockReports();
List<DatanodeInfo> infos = Arrays.asList(bg.getLocations());
// let a internal block be over replicated with 2 redundant blocks.
// Therefor number of internal blocks is over GROUP_SIZE. (5 data blocks +
// 3 parity blocks + 2 redundant blocks > GROUP_SIZE)
blk.setBlockId(groupId + 2);
List<DataNode> dataNodeList = cluster.getDataNodes();
for (int i = 0; i < numDNs; i++) {
if (!infos.contains(dataNodeList.get(i).getDatanodeId())) {
cluster.injectBlocks(i, Arrays.asList(blk), bpid);
System.out.println("XXX: inject block into datanode " + i);
}
}
// update blocksMap
cluster.triggerBlockReports();
// add to invalidates
cluster.triggerHeartbeats();
// datanode delete block
cluster.triggerHeartbeats();
// update blocksMap
cluster.triggerBlockReports();
// verify that all internal blocks exists
lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
DFSTestUtil.verifyLocatedStripedBlocks(lbs, GROUP_SIZE - 1);
}
@Test
public void testProcessOverReplicatedAndCorruptStripedBlock()
throws Exception {
long fileLen = DATA_BLK_NUM * BLOCK_SIZE;
DFSTestUtil.createStripedFile(cluster, filePath, null, 1,
NUM_STRIPE_PER_BLOCK, false);
LocatedBlocks lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
LocatedStripedBlock bg = (LocatedStripedBlock) (lbs.get(0));
long gs = bg.getBlock().getGenerationStamp();
String bpid = bg.getBlock().getBlockPoolId();
long groupId = bg.getBlock().getBlockId();
Block blk = new Block(groupId, BLOCK_SIZE, gs);
BlockInfoStriped blockInfo = new BlockInfoStriped(blk,
ErasureCodingSchemaManager.getSystemDefaultSchema(), CELLSIZE);
for (int i = 0; i < GROUP_SIZE; i++) {
blk.setBlockId(groupId + i);
cluster.injectBlocks(i, Arrays.asList(blk), bpid);
}
cluster.triggerBlockReports();
// let a internal block be corrupt
BlockManager bm = cluster.getNamesystem().getBlockManager();
List<DatanodeInfo> infos = Arrays.asList(bg.getLocations());
List<String> storages = Arrays.asList(bg.getStorageIDs());
cluster.getNamesystem().writeLock();
try {
bm.findAndMarkBlockAsCorrupt(lbs.getLastLocatedBlock().getBlock(),
infos.get(0), storages.get(0), "TEST");
} finally {
cluster.getNamesystem().writeUnlock();
}
assertEquals(1, bm.countNodes(blockInfo).corruptReplicas());
// let a internal block be over replicated with 2 redundant block.
blk.setBlockId(groupId + 2);
cluster.injectBlocks(numDNs - 3, Arrays.asList(blk), bpid);
cluster.injectBlocks(numDNs - 2, Arrays.asList(blk), bpid);
// update blocksMap
cluster.triggerBlockReports();
// add to invalidates
cluster.triggerHeartbeats();
// datanode delete block
cluster.triggerHeartbeats();
// update blocksMap
cluster.triggerBlockReports();
// verify that all internal blocks exists
lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
DFSTestUtil.verifyLocatedStripedBlocks(lbs, GROUP_SIZE);
}
@Test
public void testProcessOverReplicatedAndMissingStripedBlock()
throws Exception {
long fileLen = CELLSIZE * DATA_BLK_NUM;
DFSTestUtil.createStripedFile(cluster, filePath, null, 1,
NUM_STRIPE_PER_BLOCK, false);
LocatedBlocks lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
LocatedStripedBlock bg = (LocatedStripedBlock) (lbs.get(0));
long gs = bg.getBlock().getGenerationStamp();
String bpid = bg.getBlock().getBlockPoolId();
long groupId = bg.getBlock().getBlockId();
Block blk = new Block(groupId, BLOCK_SIZE, gs);
// only inject GROUP_SIZE - 1 blocks, so there is one block missing
for (int i = 0; i < GROUP_SIZE - 1; i++) {
blk.setBlockId(groupId + i);
cluster.injectBlocks(i, Arrays.asList(blk), bpid);
}
cluster.triggerBlockReports();
// let a internal block be over replicated with 2 redundant blocks.
// Therefor number of internal blocks is over GROUP_SIZE. (5 data blocks +
// 3 parity blocks + 2 redundant blocks > GROUP_SIZE)
blk.setBlockId(groupId + 2);
cluster.injectBlocks(numDNs - 3, Arrays.asList(blk), bpid);
cluster.injectBlocks(numDNs - 2, Arrays.asList(blk), bpid);
// update blocksMap
cluster.triggerBlockReports();
// add to invalidates
cluster.triggerHeartbeats();
// datanode delete block
cluster.triggerHeartbeats();
// update blocksMap
cluster.triggerBlockReports();
// Since one block is missing, when over-replicated blocks got deleted,
// we are left GROUP_SIZE - 1 blocks.
lbs = cluster.getNameNodeRpc().getBlockLocations(
filePath.toString(), 0, fileLen);
DFSTestUtil.verifyLocatedStripedBlocks(lbs, GROUP_SIZE - 1);
}
}