[
https://issues.apache.org/jira/browse/HDFS-7076?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16644499#comment-16644499
]
Xiang Li commented on HDFS-7076:
--------------------------------
Hi all, if you are about to use the patch, please be aware that Mover does not
work but it is easy to fix.
*Stack trace*
{code:java}
java.lang.ArrayIndexOutOfBoundsException: 16
at
org.apache.hadoop.hdfs.server.mover.Mover.initStoragePolicies(Mover.java:159)
at org.apache.hadoop.hdfs.server.mover.Mover.init(Mover.java:141)
at org.apache.hadoop.hdfs.server.mover.Mover.run(Mover.java:165)
at org.apache.hadoop.hdfs.server.mover.Mover.run(Mover.java:568)
at org.apache.hadoop.hdfs.server.mover.Mover$Cli.run(Mover.java:696)
at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)
at org.apache.hadoop.hdfs.server.mover.Mover.main(Mover.java:727)
{code}
*Reason*
The size of BlockStoragePolicy array is hardcoded to 16, but the ID of a custom
storage policy is always >= 16
*Fix*
{code:java}
diff --git
a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/mover/Mover.java
b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/mover/Mover.java
index 5fcd29f..d721f6b 100644
---
a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/mover/Mover.java
+++
b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/mover/Mover.java
@@ -114,7 +114,7 @@ private StorageGroup getTarget(String uuid, StorageType
storageType) {
private final BlockStoragePolicy[] blockStoragePolicies;
- Mover(NameNodeConnector nnc, Configuration conf, AtomicInteger retryCount) {
+ Mover(NameNodeConnector nnc, Configuration conf, AtomicInteger retryCount)
throws IOException {
final long movedWinWidth = conf.getLong(
DFSConfigKeys.DFS_MOVER_MOVEDWINWIDTH_KEY,
DFSConfigKeys.DFS_MOVER_MOVEDWINWIDTH_DEFAULT);
@@ -133,8 +133,17 @@ private StorageGroup getTarget(String uuid, StorageType
storageType) {
maxConcurrentMovesPerNode, conf);
this.storages = new StorageMap();
this.targetPaths = nnc.getTargetPaths();
- this.blockStoragePolicies = new BlockStoragePolicy[1 <<
- BlockStoragePolicySuite.ID_BIT_LENGTH];
+
+ // Set the size of blockStoragePolicies array according to the current
HDFS setup
+ BlockStoragePolicy[] policies =
dispatcher.getDistributedFileSystem().getStoragePolicies();
+ int size = BlockStoragePolicySuite.RESERVED_POLICY_NUM;
+ for (BlockStoragePolicy policy : policies) {
+ int id = policy.getId();
+ if (size < id + 1) {
+ size = id + 1; // set size to the max id + 1
+ }
+ }
+ this.blockStoragePolicies = new BlockStoragePolicy[size];
}
{code}
Need some efforts to add the change above (and UT) into the latest patch (008)
> Allow users to define custom storage policies
> ---------------------------------------------
>
> Key: HDFS-7076
> URL: https://issues.apache.org/jira/browse/HDFS-7076
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Reporter: Jing Zhao
> Assignee: Jing Zhao
> Priority: Major
> Labels: BB2015-05-TBR
> Attachments: HDFS-7076.000.patch, HDFS-7076.001.patch,
> HDFS-7076.002.patch, HDFS-7076.003.patch, HDFS-7076.004.patch,
> HDFS-7076.005.patch, HDFS-7076.005.patch, HDFS-7076.007.patch,
> HDFS-7076.008.patch, editsStored
>
>
> Currently block storage policies are hard coded. This JIRA is to persist the
> policies in FSImage and Edit Log in order to support adding new policies or
> modifying existing policies.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
