@Override
public int run(String[] args) throws Exception {
String locatorHost = args[0];
int locatorPort = Integer.parseInt(args[1]);
String hdfsHomeDir = args[2];
System.out.println(
"KnownKeysMRv2 invoked with args (locatorHost = "
+ locatorHost
+ " locatorPort = "
+ locatorPort
+ " hdfsHomeDir = "
+ hdfsHomeDir);
Configuration conf = getConf();
conf.set(GFInputFormat.INPUT_REGION, "partitionedRegion");
conf.set(GFInputFormat.HOME_DIR, hdfsHomeDir);
conf.setBoolean(GFInputFormat.CHECKPOINT, false);
conf.set(GFOutputFormat.REGION, "validationRegion");
conf.set(GFOutputFormat.LOCATOR_HOST, locatorHost);
conf.setInt(GFOutputFormat.LOCATOR_PORT, locatorPort);
Job job = Job.getInstance(conf, "knownKeysMRv2");
job.setInputFormatClass(GFInputFormat.class);
job.setOutputFormatClass(GFOutputFormat.class);
job.setMapperClass(KnownKeysMRv2Mapper.class);
job.setMapOutputKeyClass(GFKey.class);
job.setMapOutputValueClass(PEIWritable.class);
job.setReducerClass(KnownKeysMRv2Reducer.class);
// job.setOutputKeyClass(String.class);
// job.setOutputValueClass(ValueHolder.class);
return job.waitForCompletion(false) ? 0 : 1;
}
/**
* Test how IPC cache map works.
*
* @throws Exception If failed.
*/
@SuppressWarnings("unchecked")
public void testIpcCache() throws Exception {
Field cacheField = GridGgfsHadoopIpcIo.class.getDeclaredField("ipcCache");
cacheField.setAccessible(true);
Field activeCntField = GridGgfsHadoopIpcIo.class.getDeclaredField("activeCnt");
activeCntField.setAccessible(true);
Map<String, GridGgfsHadoopIpcIo> cache =
(Map<String, GridGgfsHadoopIpcIo>) cacheField.get(null);
String name = "ggfs:" + getTestGridName(0) + "@";
Configuration cfg = new Configuration();
cfg.addResource(U.resolveGridGainUrl(HADOOP_FS_CFG));
cfg.setBoolean("fs.ggfs.impl.disable.cache", true);
cfg.setBoolean(String.format(GridGgfsHadoopUtils.PARAM_GGFS_ENDPOINT_NO_EMBED, name), true);
// Ensure that existing IO is reused.
FileSystem fs1 = FileSystem.get(new URI("ggfs://" + name + "/"), cfg);
assertEquals(1, cache.size());
GridGgfsHadoopIpcIo io = null;
System.out.println("CACHE: " + cache);
for (String key : cache.keySet()) {
if (key.contains("10500")) {
io = cache.get(key);
break;
}
}
assert io != null;
assertEquals(1, ((AtomicInteger) activeCntField.get(io)).get());
// Ensure that when IO is used by multiple file systems and one of them is closed, IO is not
// stopped.
FileSystem fs2 = FileSystem.get(new URI("ggfs://" + name + "/abc"), cfg);
assertEquals(1, cache.size());
assertEquals(2, ((AtomicInteger) activeCntField.get(io)).get());
fs2.close();
assertEquals(1, cache.size());
assertEquals(1, ((AtomicInteger) activeCntField.get(io)).get());
Field stopField = GridGgfsHadoopIpcIo.class.getDeclaredField("stopping");
stopField.setAccessible(true);
assert !(Boolean) stopField.get(io);
// Ensure that IO is stopped when nobody else is need it.
fs1.close();
assert cache.isEmpty();
assert (Boolean) stopField.get(io);
}