本文整理汇总了Java中org.apache.hadoop.hdfs.DistributedFileSystem.listLocatedStatus方法的典型用法代码示例。如果您正苦于以下问题:Java DistributedFileSystem.listLocatedStatus方法的具体用法?Java DistributedFileSystem.listLocatedStatus怎么用?Java DistributedFileSystem.listLocatedStatus使用的例子?那么, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类org.apache.hadoop.hdfs.DistributedFileSystem
的用法示例。
在下文中一共展示了DistributedFileSystem.listLocatedStatus方法的1个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Java代码示例。
示例1: testDecommissionStatusAfterDNRestart
import org.apache.hadoop.hdfs.DistributedFileSystem; //导入方法依赖的package包/类
/**
* Verify a DN remains in DECOMMISSION_INPROGRESS state if it is marked
* as dead before decommission has completed. That will allow DN to resume
* the replication process after it rejoins the cluster.
*/
@Test(timeout=120000)
public void testDecommissionStatusAfterDNRestart() throws Exception {
DistributedFileSystem fileSys =
(DistributedFileSystem)cluster.getFileSystem();
// Create a file with one block. That block has one replica.
Path f = new Path("decommission.dat");
DFSTestUtil.createFile(fileSys, f, fileSize, fileSize, fileSize,
(short)1, seed);
// Find the DN that owns the only replica.
RemoteIterator<LocatedFileStatus> fileList = fileSys.listLocatedStatus(f);
BlockLocation[] blockLocations = fileList.next().getBlockLocations();
String dnName = blockLocations[0].getNames()[0];
// Decommission the DN.
FSNamesystem fsn = cluster.getNamesystem();
final DatanodeManager dm = fsn.getBlockManager().getDatanodeManager();
decommissionNode(fsn, localFileSys, dnName);
dm.refreshNodes(conf);
// Stop the DN when decommission is in progress.
// Given DFS_DATANODE_BALANCE_BANDWIDTHPERSEC_KEY is to 1 and the size of
// the block, it will take much longer time that test timeout value for
// the decommission to complete. So when stopDataNode is called,
// decommission should be in progress.
DataNodeProperties dataNodeProperties = cluster.stopDataNode(dnName);
final List<DatanodeDescriptor> dead = new ArrayList<DatanodeDescriptor>();
while (true) {
dm.fetchDatanodes(null, dead, false);
if (dead.size() == 1) {
break;
}
Thread.sleep(1000);
}
// Force removal of the dead node's blocks.
BlockManagerTestUtil.checkHeartbeat(fsn.getBlockManager());
// Force DatanodeManager to check decommission state.
BlockManagerTestUtil.recheckDecommissionState(dm);
// Verify that the DN remains in DECOMMISSION_INPROGRESS state.
assertTrue("the node should be DECOMMISSION_IN_PROGRESSS",
dead.get(0).isDecommissionInProgress());
// Check DatanodeManager#getDecommissionNodes, make sure it returns
// the node as decommissioning, even if it's dead
List<DatanodeDescriptor> decomlist = dm.getDecommissioningNodes();
assertTrue("The node should be be decommissioning", decomlist.size() == 1);
// Delete the under-replicated file, which should let the
// DECOMMISSION_IN_PROGRESS node become DECOMMISSIONED
cleanupFile(fileSys, f);
BlockManagerTestUtil.recheckDecommissionState(dm);
assertTrue("the node should be decommissioned",
dead.get(0).isDecommissioned());
// Add the node back
cluster.restartDataNode(dataNodeProperties, true);
cluster.waitActive();
// Call refreshNodes on FSNamesystem with empty exclude file.
// This will remove the datanodes from decommissioning list and
// make them available again.
writeConfigFile(localFileSys, excludeFile, null);
dm.refreshNodes(conf);
}