Package org.apache.hadoop.hdfs.server.datanode

Class DirectoryScanner.ReportCompiler

java.lang.Object

org.apache.hadoop.hdfs.server.datanode.DirectoryScanner.ReportCompiler

All Implemented Interfaces:

java.util.concurrent.Callable<DirectoryScanner.ScanInfoVolumeReport>

Enclosing class:

DirectoryScanner

public class DirectoryScanner.ReportCompiler
extends java.lang.Object
implements java.util.concurrent.Callable<DirectoryScanner.ScanInfoVolumeReport>

The ReportCompiler class encapsulates the process of searching a datanode's disks for block information. It operates by performing a DFS of the volume to discover block information. When the ReportCompiler discovers block information, it create a new ScanInfo object for it and adds that object to its report list. The report list is returned by the call() method.

Constructor Summary

Constructors

Constructor	Description
ReportCompiler(FsVolumeSpi volume)	Create a report compiler for the given volume.

Method Summary

Modifier and Type	Method	Description
DirectoryScanner.ScanInfoVolumeReport	call()	Run this report compiler thread.
void	throttle()	Called by the thread before each potential disk scan so that a pause can be optionally inserted to limit the number of scans per second.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ReportCompiler

public ReportCompiler(FsVolumeSpi volume)

Create a report compiler for the given volume.

Parameters:

volume - the target volume

Method Detail

call

public DirectoryScanner.ScanInfoVolumeReport call()
                                           throws java.io.IOException

Run this report compiler thread.

Specified by:

call in interface java.util.concurrent.Callable<DirectoryScanner.ScanInfoVolumeReport>

Returns:

the block info report list

Throws:

java.io.IOException - if the block pool is not found

throttle

public void throttle()
              throws java.lang.InterruptedException

Called by the thread before each potential disk scan so that a pause can be optionally inserted to limit the number of scans per second. The limit is controlled by DFSConfigKeys.DFS_DATANODE_DIRECTORYSCAN_THROTTLE_LIMIT_MS_PER_SEC_KEY.

Throws:

java.lang.InterruptedException