You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-commits@hadoop.apache.org by we...@apache.org on 2022/11/30 01:43:34 UTC
[hadoop] branch trunk updated: HDFS-16839 It should consider EC reconstruction work when we determine if a node is busy (#5128)
This is an automated email from the ASF dual-hosted git repository.
weichiu pushed a commit to branch trunk
in repository https://gitbox.apache.org/repos/asf/hadoop.git
The following commit(s) were added to refs/heads/trunk by this push:
new 72749a4ff8a HDFS-16839 It should consider EC reconstruction work when we determine if a node is busy (#5128)
72749a4ff8a is described below
commit 72749a4ff8a240867a822bb9ae1382d3235083ea
Author: Kidd5368 <57...@users.noreply.github.com>
AuthorDate: Wed Nov 30 09:43:15 2022 +0800
HDFS-16839 It should consider EC reconstruction work when we determine if a node is busy (#5128)
Co-authored-by: Takanobu Asanuma <ta...@apache.org>
Reviewed-by: Tao Li <to...@apache.org>
---
.../hdfs/server/blockmanagement/BlockManager.java | 6 ++-
.../server/blockmanagement/TestBlockManager.java | 52 ++++++++++++++++++++++
2 files changed, 56 insertions(+), 2 deletions(-)
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManager.java
index ded48930832..9836baa0040 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManager.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockManager.java
@@ -2599,7 +2599,8 @@ public class BlockManager implements BlockStatsMXBean {
if (priority != LowRedundancyBlocks.QUEUE_HIGHEST_PRIORITY
&& (!node.isDecommissionInProgress() && !node.isEnteringMaintenance())
- && node.getNumberOfBlocksToBeReplicated() >= maxReplicationStreams) {
+ && node.getNumberOfBlocksToBeReplicated() +
+ node.getNumberOfBlocksToBeErasureCoded() >= maxReplicationStreams) {
if (isStriped && (state == StoredReplicaState.LIVE
|| state == StoredReplicaState.DECOMMISSIONING)) {
liveBusyBlockIndices.add(blockIndex);
@@ -2609,7 +2610,8 @@ public class BlockManager implements BlockStatsMXBean {
continue; // already reached replication limit
}
- if (node.getNumberOfBlocksToBeReplicated() >= replicationStreamsHardLimit) {
+ if (node.getNumberOfBlocksToBeReplicated() +
+ node.getNumberOfBlocksToBeErasureCoded() >= replicationStreamsHardLimit) {
if (isStriped && (state == StoredReplicaState.LIVE
|| state == StoredReplicaState.DECOMMISSIONING)) {
liveBusyBlockIndices.add(blockIndex);
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestBlockManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestBlockManager.java
index 8ebcbfe2e34..04f474f67ef 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestBlockManager.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestBlockManager.java
@@ -957,6 +957,58 @@ public class TestBlockManager {
assertNull(work);
}
+ @Test
+ public void testSkipReconstructionWithManyBusyNodes3() {
+ NameNode.initMetrics(new Configuration(), HdfsServerConstants.NamenodeRole.NAMENODE);
+ 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);
+
+ ExtendedBlock dummyBlock = new ExtendedBlock("bpid", 1, 1, 1);
+ DatanodeDescriptor dummyDD = ds1.getDatanodeDescriptor();
+ DatanodeDescriptor[] dummyDDArray = new DatanodeDescriptor[]{dummyDD};
+ DatanodeStorageInfo[] dummyDSArray = new DatanodeStorageInfo[]{ds1};
+ // Simulate the 2 nodes reach maxReplicationStreams.
+ for(int i = 0; i < bm.maxReplicationStreams; i++){ //Add some dummy EC reconstruction task.
+ ds3.getDatanodeDescriptor().addBlockToBeErasureCoded(dummyBlock, dummyDDArray,
+ dummyDSArray, new byte[0], new byte[0], ecPolicy);
+ ds4.getDatanodeDescriptor().addBlockToBeErasureCoded(dummyBlock, dummyDDArray,
+ dummyDSArray, new byte[0], new byte[0], ecPolicy);
+ }
+
+ // Reconstruction should be skipped since the number of non-busy nodes are not enough.
+ work = bm.scheduleReconstruction(aBlockInfoStriped, 3);
+ assertNull(work);
+ }
+
@Test
public void testFavorDecomUntilHardLimit() throws Exception {
bm.maxReplicationStreams = 0;
---------------------------------------------------------------------
To unsubscribe, e-mail: common-commits-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-commits-help@hadoop.apache.org