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

HDFS-16479. EC: NameNode should not send a reconstruction work when the source datanodes are insufficient (#4138)

Browse Source 
(cherry picked from commit 2efab92959)
This commit is contained in:
Takanobu Asanuma 2022-04-14 11:23:38 +09:00
parent 52c6d77274
commit 52abc9f132
2 changed files with 106 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManager.java
View File

@ -2118,6 +2118,16 @@ BlockReconstructionWork scheduleReconstruction(BlockInfo block,
return null; return null;
} }
// skip if source datanodes for reconstructing ec block are not enough
if (block.isStriped()) {
BlockInfoStriped stripedBlock = (BlockInfoStriped) block;
if (stripedBlock.getRealDataBlockNum() > srcNodes.length) {
LOG.debug("Block {} cannot be reconstructed due to shortage of source datanodes ", block);
NameNode.getNameNodeMetrics().incNumTimesReReplicationNotScheduled();
return null;
}
}
// liveReplicaNodes can include READ_ONLY_SHARED replicas which are // liveReplicaNodes can include READ_ONLY_SHARED replicas which are
// not included in the numReplicas.liveReplicas() count // not included in the numReplicas.liveReplicas() count
assert liveReplicaNodes.size() >= numReplicas.liveReplicas(); assert liveReplicaNodes.size() >= numReplicas.liveReplicas();

96
hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestBlockManager.java
View File

@ -852,6 +852,102 @@ public void testChooseSrcDNWithDupECInDecommissioningNode() throws Exception {
0, numReplicas.redundantInternalBlocks()); 0, numReplicas.redundantInternalBlocks());
} }
@Test
public void testSkipReconstructionWithManyBusyNodes() {
long blockId = -9223372036854775776L; // real ec block id
// RS-3-2 EC policy
ErasureCodingPolicy ecPolicy =
SystemErasureCodingPolicies.getPolicies().get(1);
// create an EC block group: 3 data blocks + 2 parity blocks
Block aBlockGroup = new Block(blockId, ecPolicy.getCellSize() * ecPolicy.getNumDataUnits(), 0);
BlockInfoStriped aBlockInfoStriped = new BlockInfoStriped(aBlockGroup, ecPolicy);
// create 4 storageInfo, which means 1 block is missing
DatanodeStorageInfo ds1 = DFSTestUtil.createDatanodeStorageInfo(
"storage1", "1.1.1.1", "rack1", "host1");
DatanodeStorageInfo ds2 = DFSTestUtil.createDatanodeStorageInfo(
"storage2", "2.2.2.2", "rack2", "host2");
DatanodeStorageInfo ds3 = DFSTestUtil.createDatanodeStorageInfo(
"storage3", "3.3.3.3", "rack3", "host3");
DatanodeStorageInfo ds4 = DFSTestUtil.createDatanodeStorageInfo(
"storage4", "4.4.4.4", "rack4", "host4");
// link block with storage
aBlockInfoStriped.addStorage(ds1, aBlockGroup);
aBlockInfoStriped.addStorage(ds2, new Block(blockId + 1, 0, 0));
aBlockInfoStriped.addStorage(ds3, new Block(blockId + 2, 0, 0));
aBlockInfoStriped.addStorage(ds4, new Block(blockId + 3, 0, 0));
addEcBlockToBM(blockId, ecPolicy);
aBlockInfoStriped.setBlockCollectionId(mockINodeId);
// reconstruction should be scheduled
BlockReconstructionWork work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
assertNotNull(work);
// simulate the 2 nodes reach maxReplicationStreams
for(int i = 0; i < bm.maxReplicationStreams; i++){
ds3.getDatanodeDescriptor().incrementPendingReplicationWithoutTargets();
ds4.getDatanodeDescriptor().incrementPendingReplicationWithoutTargets();
}
// reconstruction should be skipped since the number of non-busy nodes are not enough
work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
assertNull(work);
}
@Test
public void testSkipReconstructionWithManyBusyNodes2() {
long blockId = -9223372036854775776L; // real ec block id
// RS-3-2 EC policy
ErasureCodingPolicy ecPolicy =
SystemErasureCodingPolicies.getPolicies().get(1);
// create an EC block group: 2 data blocks + 2 parity blocks
Block aBlockGroup = new Block(blockId,
ecPolicy.getCellSize() * (ecPolicy.getNumDataUnits() - 1), 0);
BlockInfoStriped aBlockInfoStriped = new BlockInfoStriped(aBlockGroup, ecPolicy);
// create 3 storageInfo, which means 1 block is missing
DatanodeStorageInfo ds1 = DFSTestUtil.createDatanodeStorageInfo(
"storage1", "1.1.1.1", "rack1", "host1");
DatanodeStorageInfo ds2 = DFSTestUtil.createDatanodeStorageInfo(
"storage2", "2.2.2.2", "rack2", "host2");
DatanodeStorageInfo ds3 = DFSTestUtil.createDatanodeStorageInfo(
"storage3", "3.3.3.3", "rack3", "host3");
// link block with storage
aBlockInfoStriped.addStorage(ds1, aBlockGroup);
aBlockInfoStriped.addStorage(ds2, new Block(blockId + 1, 0, 0));
aBlockInfoStriped.addStorage(ds3, new Block(blockId + 2, 0, 0));
addEcBlockToBM(blockId, ecPolicy);
aBlockInfoStriped.setBlockCollectionId(mockINodeId);
// reconstruction should be scheduled
BlockReconstructionWork work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
assertNotNull(work);
// simulate the 1 node reaches maxReplicationStreams
for(int i = 0; i < bm.maxReplicationStreams; i++){
ds2.getDatanodeDescriptor().incrementPendingReplicationWithoutTargets();
}
// reconstruction should still be scheduled since there are 2 source nodes to create 2 blocks
work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
assertNotNull(work);
// simulate the 1 more node reaches maxReplicationStreams
for(int i = 0; i < bm.maxReplicationStreams; i++){
ds3.getDatanodeDescriptor().incrementPendingReplicationWithoutTargets();
}
// reconstruction should be skipped since the number of non-busy nodes are not enough
work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
assertNull(work);
}
@Test @Test
public void testFavorDecomUntilHardLimit() throws Exception { public void testFavorDecomUntilHardLimit() throws Exception {
bm.maxReplicationStreams = 0; bm.maxReplicationStreams = 0;