本文整理汇总了Java中org.apache.hadoop.hdfs.DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT属性的典型用法代码示例。如果您正苦于以下问题：Java DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT属性的具体用法？Java DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT怎么用？Java DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT使用的例子？那么, 这里精选的属性代码示例或许可以为您提供帮助。您也可以进一步了解该属性所在类org.apache.hadoop.hdfs.DFSConfigKeys的用法示例。
在下文中一共展示了DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT属性的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: testChooseReplicaToDelete
/**
* The test verifies that replica for deletion is chosen on a node,
* with the oldest heartbeat, when this heartbeat is larger than the
* tolerable heartbeat interval.
* It creates a file with several blocks and replication 4.
* The last DN is configured to send heartbeats rarely.
*
* Test waits until the tolerable heartbeat interval expires, and reduces
* replication of the file. All replica deletions should be scheduled for the
* last node. No replicas will actually be deleted, since last DN doesn't
* send heartbeats.
*/
@Test
public void testChooseReplicaToDelete() throws Exception {
MiniDFSCluster cluster = null;
FileSystem fs = null;
try {
Configuration conf = new HdfsConfiguration();
conf.setLong(DFSConfigKeys.DFS_BLOCK_SIZE_KEY, SMALL_BLOCK_SIZE);
cluster = new MiniDFSCluster.Builder(conf).numDataNodes(3).build();
fs = cluster.getFileSystem();
final FSNamesystem namesystem = cluster.getNamesystem();
conf.setLong(DFSConfigKeys.DFS_HEARTBEAT_INTERVAL_KEY, 300);
cluster.startDataNodes(conf, 1, true, null, null, null);
DataNode lastDN = cluster.getDataNodes().get(3);
DatanodeRegistration dnReg = DataNodeTestUtils.getDNRegistrationForBP(
lastDN, namesystem.getBlockPoolId());
String lastDNid = dnReg.getDatanodeUuid();
final Path fileName = new Path("/foo2");
DFSTestUtil.createFile(fs, fileName, SMALL_FILE_LENGTH, (short)4, 0L);
DFSTestUtil.waitReplication(fs, fileName, (short)4);
// Wait for tolerable number of heartbeats plus one
DatanodeDescriptor nodeInfo = null;
long lastHeartbeat = 0;
long waitTime = DFSConfigKeys.DFS_HEARTBEAT_INTERVAL_DEFAULT * 1000 *
(DFSConfigKeys.DFS_NAMENODE_TOLERATE_HEARTBEAT_MULTIPLIER_DEFAULT + 1);
do {
nodeInfo = namesystem.getBlockManager().getDatanodeManager()
.getDatanode(dnReg);
lastHeartbeat = nodeInfo.getLastUpdateMonotonic();
} while (monotonicNow() - lastHeartbeat < waitTime);
fs.setReplication(fileName, (short)3);
BlockLocation locs[] = fs.getFileBlockLocations(
fs.getFileStatus(fileName), 0, Long.MAX_VALUE);
// All replicas for deletion should be scheduled on lastDN.
// And should not actually be deleted, because lastDN does not heartbeat.
namesystem.readLock();
Collection<Block> dnBlocks =
namesystem.getBlockManager().excessReplicateMap.get(lastDNid);
assertEquals("Replicas on node " + lastDNid + " should have been deleted",
SMALL_FILE_LENGTH / SMALL_BLOCK_SIZE, dnBlocks.size());
namesystem.readUnlock();
for(BlockLocation location : locs)
assertEquals("Block should still have 4 replicas",
4, location.getNames().length);
} finally {
if(fs != null) fs.close();
if(cluster != null) cluster.shutdown();
}
}