[
https://issues.apache.org/jira/browse/HDDS-2203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16940499#comment-16940499
]
Istvan Fajth commented on HDDS-2203:
------------------------------------
I have some concerns about the design of this class, and I would like to have
some insights from others, especially from [~aengineer] as he reviewed the
change so far.
What do you think about removing this class, and move its parts back to the
calling code, to
#1 avoid the synchronization at all
#2 eliminate the need of the init call before the conversion methods as this I
think is an unfortunate design
#3 move this logic where it is used, as it is used from two places, but from
one place one of the methods are called, from the other, an other method is
called?
If you agree, I would create a new JIRA for that, and also I am happy to
provide a PR with the changes.
> Race condition in ByteStringHelper.init()
> -----------------------------------------
>
> Key: HDDS-2203
> URL: https://issues.apache.org/jira/browse/HDDS-2203
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: Ozone Client, SCM
> Reporter: Istvan Fajth
> Assignee: Istvan Fajth
> Priority: Critical
> Labels: pull-request-available, pull-requests-available
> Time Spent: 40m
> Remaining Estimate: 0h
>
> The current init method:
> {code}
> public static void init(boolean isUnsafeByteOperation) {
> final boolean set = INITIALIZED.compareAndSet(false, true);
> if (set) {
> ByteStringHelper.isUnsafeByteOperationsEnabled =
> isUnsafeByteOperation;
> } else {
> // already initialized, check values
> Preconditions.checkState(isUnsafeByteOperationsEnabled
> == isUnsafeByteOperation);
> }
> }
> {code}
> In a scenario when two thread accesses this method, and the execution order
> is the following, then the second thread runs into an exception from
> PreCondition.checkState() in the else branch.
> In an unitialized state:
> - T1 thread arrives to the method with true as the parameter, the class
> initialises the isUnsafeByteOperationsEnabled to false
> - T1 sets INITIALIZED true
> - T2 arrives to the method with true as the parameter
> - T2 reads the INITALIZED value and as it is not false goes to else branch
> - T2 tries to check if the internal boolean property is the same true as it
> wanted to set, and as T1 still to set the value, the checkState throws an
> IllegalArgumentException.
> This happens in certain Hive query cases, as it came from that testing, the
> exception we see there is the following:
> {code}
> Error: Error while processing statement: FAILED: Execution Error, return code
> 2 from org.apache.hadoop.hive.ql.exec.tez.TezTask. Vertex failed,
> vertexName=Map 2, vertexId=vertex_1569486223160_0334_1_02,
> diagnostics=[Vertex vertex_1569486223160_0334_1_02 [Map 2] killed/failed
> due to:ROOT_INPUT_INIT_FAILURE, Vertex Input: item initializer failed,
> vertex=vertex_1569486223160_0334_1_02 [Map 2], java.io.IOException: Couldn't
> create RpcClient protocol
> at
> org.apache.hadoop.ozone.client.OzoneClientFactory.getClientProtocol(OzoneClientFactory.java:263)
> at
> org.apache.hadoop.ozone.client.OzoneClientFactory.getClientProtocol(OzoneClientFactory.java:239)
> at
> org.apache.hadoop.ozone.client.OzoneClientFactory.getRpcClient(OzoneClientFactory.java:203)
> at
> org.apache.hadoop.ozone.client.OzoneClientFactory.getRpcClient(OzoneClientFactory.java:165)
> at
> org.apache.hadoop.fs.ozone.BasicOzoneClientAdapterImpl.<init>(BasicOzoneClientAdapterImpl.java:158)
> at
> org.apache.hadoop.fs.ozone.OzoneClientAdapterImpl.<init>(OzoneClientAdapterImpl.java:50)
> at
> org.apache.hadoop.fs.ozone.OzoneFileSystem.createAdapter(OzoneFileSystem.java:102)
> at
> org.apache.hadoop.fs.ozone.BasicOzoneFileSystem.initialize(BasicOzoneFileSystem.java:155)
> at org.apache.hadoop.fs.FileSystem.createFileSystem(FileSystem.java:3315)
> at org.apache.hadoop.fs.FileSystem.access$200(FileSystem.java:136)
> at org.apache.hadoop.fs.FileSystem$Cache.getInternal(FileSystem.java:3364)
> at org.apache.hadoop.fs.FileSystem$Cache.get(FileSystem.java:3332)
> at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:491)
> at org.apache.hadoop.fs.Path.getFileSystem(Path.java:361)
> at
> org.apache.hadoop.hive.ql.io.orc.OrcInputFormat.generateSplitsInfo(OrcInputFormat.java:1821)
> at
> org.apache.hadoop.hive.ql.io.orc.OrcInputFormat.getSplits(OrcInputFormat.java:2002)
> at
> org.apache.hadoop.hive.ql.io.HiveInputFormat.addSplitsForGroup(HiveInputFormat.java:524)
> at
> org.apache.hadoop.hive.ql.io.HiveInputFormat.getSplits(HiveInputFormat.java:781)
> at
> org.apache.hadoop.hive.ql.exec.tez.HiveSplitGenerator.initialize(HiveSplitGenerator.java:243)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable$1.run(RootInputInitializerManager.java:278)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable$1.run(RootInputInitializerManager.java:269)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1876)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable.call(RootInputInitializerManager.java:269)
> at
> org.apache.tez.dag.app.dag.RootInputInitializerManager$InputInitializerCallable.call(RootInputInitializerManager.java:253)
> at
> com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.runInterruptibly(TrustedListenableFutureTask.java:108)
> at
> com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:41)
> at
> com.google.common.util.concurrent.TrustedListenableFutureTask.run(TrustedListenableFutureTask.java:77)
> 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.IllegalStateException
> at
> org.apache.hadoop.ozone.shaded.com.google.common.base.Preconditions.checkState(Preconditions.java:129)
> at
> org.apache.hadoop.hdds.scm.ByteStringHelper.init(ByteStringHelper.java:47)
> at org.apache.hadoop.ozone.client.rpc.RpcClient.<init>(RpcClient.java:241)
> at
> org.apache.hadoop.ozone.client.OzoneClientFactory.getClientProtocol(OzoneClientFactory.java:256)
> ... 31 more
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
