This is an automated email from the ASF dual-hosted git repository.
hexiaoqiao 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 700147b4ac1 HDFS-16964. Improve processing of excess redundancy after
failover. (#5510). Contributed by Shuyan Zhang.
700147b4ac1 is described below
commit 700147b4ac18ceca5137e9d7fc8f53a5619768d4
Author: zhangshuyan <81411509+zhangshuy...@users.noreply.github.com>
AuthorDate: Tue Mar 28 16:14:59 2023 +0800
HDFS-16964. Improve processing of excess redundancy after failover.
(#5510). Contributed by Shuyan Zhang.
Signed-off-by: He Xiaoqiao <hexiaoq...@apache.org>
---
.../hdfs/server/blockmanagement/BlockManager.java | 59 ++++++++++++++--------
1 file changed, 38 insertions(+), 21 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 e5a6cf73b69..ec8bbf82498 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
@@ -3987,17 +3987,11 @@ public class BlockManager implements BlockStatsMXBean {
}
if (shouldProcessExtraRedundancy(num, expectedRedundancy)) {
- if (num.replicasOnStaleNodes() > 0) {
- // If any of the replicas of this block are on nodes that are
- // considered "stale", then these replicas may in fact have
- // already been deleted. So, we cannot safely act on the
- // over-replication until a later point in time, when
- // the "stale" nodes have block reported.
+ // extra redundancy block
+ if (!processExtraRedundancyBlockWithoutPostpone(block,
expectedRedundancy,
+ null, null)) {
return MisReplicationResult.POSTPONE;
}
-
- // extra redundancy block
- processExtraRedundancyBlock(block, expectedRedundancy, null, null);
return MisReplicationResult.OVER_REPLICATED;
}
@@ -4020,12 +4014,26 @@ public class BlockManager implements BlockStatsMXBean {
}
}
+ /**
+ * Process blocks with redundant replicas. If there are replicas in
+ * stale storages, mark them in the postponedMisreplicatedBlocks.
+ */
+ private void processExtraRedundancyBlock(final BlockInfo block,
+ final short replication, final DatanodeDescriptor addedNode,
+ DatanodeDescriptor delNodeHint) {
+ if (!processExtraRedundancyBlockWithoutPostpone(block, replication,
+ addedNode, delNodeHint)) {
+ postponeBlock(block);
+ }
+ }
+
/**
* Find how many of the containing nodes are "extra", if any.
* If there are any extras, call chooseExcessRedundancies() to
* mark them in the excessRedundancyMap.
+ * @return true if all redundancy replicas are removed.
*/
- private void processExtraRedundancyBlock(final BlockInfo block,
+ private boolean processExtraRedundancyBlockWithoutPostpone(final BlockInfo
block,
final short replication, final DatanodeDescriptor addedNode,
DatanodeDescriptor delNodeHint) {
assert namesystem.hasWriteLock();
@@ -4035,17 +4043,17 @@ public class BlockManager implements BlockStatsMXBean {
Collection<DatanodeStorageInfo> nonExcess = new ArrayList<>();
Collection<DatanodeDescriptor> corruptNodes = corruptReplicas
.getNodes(block);
+ boolean hasStaleStorage = false;
+ Set<DatanodeStorageInfo> staleStorages = new HashSet<>();
for (DatanodeStorageInfo storage : blocksMap.getStorages(block)) {
if (storage.getState() != State.NORMAL) {
continue;
}
final DatanodeDescriptor cur = storage.getDatanodeDescriptor();
if (storage.areBlockContentsStale()) {
- LOG.trace("BLOCK* processExtraRedundancyBlock: Postponing {}"
- + " since storage {} does not yet have up-to-date information.",
- block, storage);
- postponeBlock(block);
- return;
+ hasStaleStorage = true;
+ staleStorages.add(storage);
+ continue;
}
if (!isExcess(cur, block)) {
if (cur.isInService()) {
@@ -4058,6 +4066,13 @@ public class BlockManager implements BlockStatsMXBean {
}
chooseExcessRedundancies(nonExcess, block, replication, addedNode,
delNodeHint);
+ if (hasStaleStorage) {
+ LOG.trace("BLOCK* processExtraRedundancyBlockWithoutPostpone: Postponing
{}"
+ + " since storages {} does not yet have up-to-date information.",
+ block, staleStorages);
+ return false;
+ }
+ return true;
}
private void chooseExcessRedundancies(
@@ -4071,12 +4086,14 @@ public class BlockManager implements BlockStatsMXBean {
if (storedBlock.isStriped()) {
chooseExcessRedundancyStriped(bc, nonExcess, storedBlock, delNodeHint);
} else {
- final BlockStoragePolicy storagePolicy = storagePolicySuite.getPolicy(
- bc.getStoragePolicyID());
- final List<StorageType> excessTypes = storagePolicy.chooseExcess(
- replication, DatanodeStorageInfo.toStorageTypes(nonExcess));
- chooseExcessRedundancyContiguous(nonExcess, storedBlock, replication,
- addedNode, delNodeHint, excessTypes);
+ if (nonExcess.size() > replication) {
+ final BlockStoragePolicy storagePolicy = storagePolicySuite.getPolicy(
+ bc.getStoragePolicyID());
+ final List<StorageType> excessTypes = storagePolicy.chooseExcess(
+ replication, DatanodeStorageInfo.toStorageTypes(nonExcess));
+ chooseExcessRedundancyContiguous(nonExcess, storedBlock, replication,
+ addedNode, delNodeHint, excessTypes);
+ }
}
}
---------------------------------------------------------------------
To unsubscribe, e-mail: common-commits-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-commits-h...@hadoop.apache.org
