[HIVE-17803] With Pig multi-query, 2 HCatStorers writing to the same table will trample each other's outputs

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.2.0, 3.0.0
• Fix Version/s: 2.2.1, 2.4.0, 3.0.0
• Component/s: HCatalog
• External issue ID: YHIVE-699
• Assignee: Chris Drome
• Reporter: Mithun Radhakrishnan
• Created: 13/Oct/17 21:04
• Updated: 22/May/18 23:12
• Resolved: 19/Oct/17 17:58

Description

When Pig scripts use multi-query and HCatStorer with dynamic-partitioning, and use more than one HCatStorer instance to write to the same table, they might trample on each other's outputs. The failure looks as follows:

Caused by: org.apache.hive.hcatalog.common.HCatException : 2006 : Error adding partition to metastore. Cause : org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException): No lease on /projects/foo/bar/activity_date=2016022306/_placeholder (inode 2878224200): File does not exist. [Lease.  Holder: DFSClient_NONMAPREDUCE_-1281544466_4952, pendingcreates: 1]
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease(FSNamesystem.java:3429)
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFileInternal(FSNamesystem.java:3517)
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFile(FSNamesystem.java:3484)
	at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.complete(NameNodeRpcServer.java:791)
	at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.complete(ClientNamenodeProtocolServerSideTranslatorPB.java:537)
	at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)
	at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:608)
	at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:969)
	at org.apache.hadoop.ipc.Server.call(Server.java:2267)
	at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:648)
	at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:615)
	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:1679)
	at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2217)

	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.registerPartitions(FileOutputCommitterContainer.java:1022)
	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.commitJob(FileOutputCommitterContainer.java:269)
	... 20 more
Caused by: org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException): No lease on /projects/foo/bar/activity_date=2016022306/_placeholder (inode 2878224200): File does not exist. [Lease.  Holder: DFSClient_NONMAPREDUCE_-1281544466_4952, pendingcreates: 1]
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease(FSNamesystem.java:3429)
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFileInternal(FSNamesystem.java:3517)
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFile(FSNamesystem.java:3484)
	at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.complete(NameNodeRpcServer.java:791)
	at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.complete(ClientNamenodeProtocolServerSideTranslatorPB.java:537)
	at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)
	at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:608)
	at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:969)
	at org.apache.hadoop.ipc.Server.call(Server.java:2267)
	at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:648)
	at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:615)
	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:1679)
	at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2217)

	at org.apache.hadoop.ipc.Client.call(Client.java:1457)
	at org.apache.hadoop.ipc.Client.call(Client.java:1394)
	at org.apache.hadoop.ipc.ProtobufRpcEngine$Invoker.invoke(ProtobufRpcEngine.java:230)
	at com.sun.proxy.$Proxy11.complete(Unknown Source)
	at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolTranslatorPB.complete(ClientNamenodeProtocolTranslatorPB.java:462)
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:483)
	at org.apache.hadoop.io.retry.RetryInvocationHandler.invokeMethod(RetryInvocationHandler.java:191)
	at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:102)
	at com.sun.proxy.$Proxy12.complete(Unknown Source)
	at org.apache.hadoop.hdfs.DFSOutputStream.completeFile(DFSOutputStream.java:2300)
	at org.apache.hadoop.hdfs.DFSOutputStream.closeImpl(DFSOutputStream.java:2281)
	at org.apache.hadoop.hdfs.DFSOutputStream.close(DFSOutputStream.java:2245)
	at org.apache.hadoop.fs.FSDataOutputStream$PositionCache.close(FSDataOutputStream.java:72)
	at org.apache.hadoop.fs.FSDataOutputStream.close(FSDataOutputStream.java:106)
	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:624)
	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:606)
	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:606)
	at org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.registerPartitions(FileOutputCommitterContainer.java:988)

This is the result of the following code in the FileOutputCommitterContainer:

  private void moveTaskOutputs(FileSystem fs, Path file, Path srcDir,
                 Path destDir, final boolean dryRun, boolean immutable
      ) throws IOException {
  // ...
            if (dynamicPartitioningUsed) {

            // Optimization: if the first child is file, we have reached the leaf directory, move the parent directory itself
            // instead of moving each file under the directory. See HCATALOG-538
            // Note for future Append implementation : This optimization is another reason dynamic
            // partitioning is currently incompatible with append on mutable tables.

            final Path parentDir = finalOutputPath.getParent();
            // Create the directory
            Path placeholder = new Path(parentDir, "_placeholder"); // <--- HERE!
            if (fs.mkdirs(parentDir)) {
              // It is weired but we need a placeholder, 
              // otherwise rename cannot move file to the right place
              fs.create(placeholder).close();
            }
  // ...
  }

Two storers will use the same placeholder location during commit. We should have better isolation.

Issue Links

• blocks: HIVE-17802 Remove unnecessary calls to FileSystem.setOwner() from FileOutputCommitterContainer