[
https://issues.apache.org/jira/browse/HDFS-16732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17581969#comment-17581969
]
ASF GitHub Bot commented on HDFS-16732:
---------------------------------------
xkrogen commented on code in PR #4756:
URL: https://github.com/apache/hadoop/pull/4756#discussion_r950420437
##########
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSNamesystem.java:
##########
@@ -3470,6 +3471,16 @@ HdfsFileStatus getFileInfo(final String src, boolean
resolveLink,
logAuditEvent(false, operationName, src);
throw e;
}
+ if (needLocation && haEnabled && haContext != null &&
+ haContext.getState().getServiceState() == OBSERVER &&
+ stat instanceof HdfsLocatedFileStatus) {
+ LocatedBlocks lbs = ((HdfsLocatedFileStatus) stat).getLocatedBlocks();
+ for (LocatedBlock b : lbs.getLocatedBlocks()) {
+ if (b.getLocations() == null || b.getLocations().length == 0) {
+ throw new ObserverRetryOnActiveException("Zero blocklocations for "
+ src);
+ }
+ }
+ }
Review Comment:
Can we pull this into a common method like:
```java
private void checkBlockLocationsIfObserver(Iterator<LocatedBlocks>
blocksIter) throws ObserverRetryOnActive {
if (haEnabled && haContext != null &&
haContext.getState().getServiceState() == OBSERVER) {
...
}
}
```
or two methods like
```java
private boolean isObserver() { return haEnabled && haContext != null &&
haContext.getState().getServiceState() == OBSERVER; }
private void checkBlockLocationsForObserver(LocatedBlocks blocks) throws
ObserverRetryOnActive { ... }
```
Point being that we have 3 places with almost identical logic here, we
should try to consolidate.
> [SBN READ] Avoid get location from observer when the block report is delayed.
> -----------------------------------------------------------------------------
>
> Key: HDFS-16732
> URL: https://issues.apache.org/jira/browse/HDFS-16732
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs
> Affects Versions: 3.2.1
> Reporter: zhengchenyu
> Assignee: zhengchenyu
> Priority: Critical
> Labels: pull-request-available
>
> Hive on tez application fail occasionally after observer is enable, log show
> below.
> {code:java}
> 2022-08-18 15:22:06,914 [ERROR] [Dispatcher thread {Central}]
> |impl.VertexImpl|: Vertex Input: namenodeinfo_stg initializer failed,
> vertex=vertex_1660618571916_4839_1_00 [Map 1]
> org.apache.tez.dag.app.dag.impl.AMUserCodeException:
> java.lang.ArrayIndexOutOfBoundsException: 0
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallback.onFailure(RootInputInitializerManager.java:329)
> at
> com.google.common.util.concurrent.Futures$CallbackListener.run(Futures.java:1056)
> at
> com.google.common.util.concurrent.DirectExecutor.execute(DirectExecutor.java:30)
> at
> com.google.common.util.concurrent.AbstractFuture.executeListener(AbstractFuture.java:1138)
> at
> com.google.common.util.concurrent.AbstractFuture.complete(AbstractFuture.java:958)
> at
> com.google.common.util.concurrent.AbstractFuture.setException(AbstractFuture.java:748)
> at
> com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.afterRanInterruptibly(TrustedListenableFutureTask.java:133)
> at
> com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:80)
> at
> com.google.common.util.concurrent.TrustedListenableFutureTask.run(TrustedListenableFutureTask.java:78)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> Caused by: java.lang.ArrayIndexOutOfBoundsException: 0
> at
> org.apache.hadoop.mapred.FileInputFormat.identifyHosts(FileInputFormat.java:748)
> at
> org.apache.hadoop.mapred.FileInputFormat.getSplitHostsAndCachedHosts(FileInputFormat.java:714)
> at
> org.apache.hadoop.mapred.FileInputFormat.getSplits(FileInputFormat.java:378)
> at
> org.apache.hadoop.hive.ql.io.HiveInputFormat.addSplitsForGroup(HiveInputFormat.java:306)
> at
> org.apache.hadoop.hive.ql.io.HiveInputFormat.getSplits(HiveInputFormat.java:408)
> at
> org.apache.hadoop.hive.ql.exec.tez.HiveSplitGenerator.initialize(HiveSplitGenerator.java:159)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable$1.run(RootInputInitializerManager.java:279)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable$1.run(RootInputInitializerManager.java:270)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1730)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable.call(RootInputInitializerManager.java:270)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable.call(RootInputInitializerManager.java:254)
> at
> com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.runInterruptibly(TrustedListenableFutureTask.java:125)
> at
> com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:57)
> ... 4 more {code}
> As describe in MAPREDUCE-7082, when the block is missing, then will throw
> this exception, but my cluster had no missing block.
> In this example, I found getListing return location information. When block
> report of observer is delayed, will return the block without location.
> HDFS-13924 is introduce to solve this problem, but only consider
> getBlockLocations.
> In observer node, all method which may return location should check whether
> locations is empty or not.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
