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 2019/04/03 19:41:36 UTC
[hadoop] branch branch-3.1 updated: HDFS-10477. Stop decommission a
rack of DataNodes caused NameNode fail over to standby. Contributed by
yunjiong zhao and Wei-Chiu Chuang.
This is an automated email from the ASF dual-hosted git repository.
weichiu pushed a commit to branch branch-3.1
in repository https://gitbox.apache.org/repos/asf/hadoop.git
The following commit(s) were added to refs/heads/branch-3.1 by this push:
new 084fb9d HDFS-10477. Stop decommission a rack of DataNodes caused NameNode fail over to standby. Contributed by yunjiong zhao and Wei-Chiu Chuang.
084fb9d is described below
commit 084fb9de22a811bbcf17b1da0a4da40660efae32
Author: Wei-Chiu Chuang <we...@apache.org>
AuthorDate: Wed Apr 3 11:00:12 2019 -0700
HDFS-10477. Stop decommission a rack of DataNodes caused NameNode fail over to standby. Contributed by yunjiong zhao and Wei-Chiu Chuang.
(cherry picked from commit be488b6070a124234c77f16193ee925d32ca9a20)
(cherry picked from commit c8703dda0727e17d759d7ad27f0caee88103a530)
---
.../hdfs/server/blockmanagement/BlockManager.java | 44 ++++++++++++++++------
1 file changed, 32 insertions(+), 12 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 b08ba20..9765bb6 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
@@ -4186,21 +4186,41 @@ public class BlockManager implements BlockStatsMXBean {
if (!isPopulatingReplQueues()) {
return;
}
- final Iterator<BlockInfo> it = srcNode.getBlockIterator();
+
int numExtraRedundancy = 0;
- while(it.hasNext()) {
- final BlockInfo block = it.next();
- if (block.isDeleted()) {
- //Orphan block, will be handled eventually, skip
+ for (DatanodeStorageInfo datanodeStorageInfo : srcNode.getStorageInfos()) {
+ // the namesystem lock is released between iterations. Make sure the
+ // storage is not removed before continuing.
+ if (srcNode.getStorageInfo(datanodeStorageInfo.getStorageID()) == null) {
continue;
}
- int expectedReplication = this.getExpectedRedundancyNum(block);
- NumberReplicas num = countNodes(block);
- if (shouldProcessExtraRedundancy(num, expectedReplication)) {
- // extra redundancy block
- processExtraRedundancyBlock(block, (short) expectedReplication, null,
- null);
- numExtraRedundancy++;
+ final Iterator<BlockInfo> it = datanodeStorageInfo.getBlockIterator();
+ while(it.hasNext()) {
+ final BlockInfo block = it.next();
+ if (block.isDeleted()) {
+ //Orphan block, will be handled eventually, skip
+ continue;
+ }
+ int expectedReplication = this.getExpectedRedundancyNum(block);
+ NumberReplicas num = countNodes(block);
+ if (shouldProcessExtraRedundancy(num, expectedReplication)) {
+ // extra redundancy block
+ processExtraRedundancyBlock(block, (short) expectedReplication, null,
+ null);
+ numExtraRedundancy++;
+ }
+ }
+ // When called by tests like TestDefaultBlockPlacementPolicy.
+ // testPlacementWithLocalRackNodesDecommissioned, it is not protected by
+ // lock, only when called by DatanodeManager.refreshNodes have writeLock
+ if (namesystem.hasWriteLock()) {
+ namesystem.writeUnlock();
+ try {
+ Thread.sleep(1);
+ } catch (InterruptedException e) {
+ Thread.currentThread().interrupt();
+ }
+ namesystem.writeLock();
}
}
LOG.info("Invalidated {} extra redundancy blocks on {} after "
---------------------------------------------------------------------
To unsubscribe, e-mail: common-commits-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-commits-help@hadoop.apache.org