public void testFormat() throws Exception {
localFs = FileSystem.getLocal(defaultConf);
localFs.delete(workDir, true);
Job job = new Job(new Configuration(defaultConf));
Path file = new Path(workDir, "test.txt");
int seed = new Random().nextInt();
Random random = new Random(seed);
// for a variety of lengths
for (int length = 0; length < MAX_LENGTH; length += random.nextInt(MAX_LENGTH / 10) + 1) {
// create a file with length entries
Writer writer = new OutputStreamWriter(localFs.create(file));
try {
MyClass mc = new MyClass();
for (int i = 0; i < length; i++) {
mc.s = Integer.toString(i);
mc.v = i;
byte[] raw = MessagePack.pack(mc);
byte[] b64e = base64_.encodeBase64(raw);
byte[] b64d = base64_.decode(b64e);
MyClass mc2 = MessagePack.unpack(b64d, mc.getClass());
assertEquals(mc.s, mc2.s);
assertEquals(mc.v, mc2.v);
writer.write(base64_.encodeToString(raw));
}
} finally {
writer.close();
}
checkFormat(job);
}
}
public void inject(Path crawlDb, Path urlDir) throws IOException {
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
long start = System.currentTimeMillis();
if (LOG.isInfoEnabled()) {
LOG.info("Injector: starting at " + sdf.format(start));
LOG.info("Injector: crawlDb: " + crawlDb);
LOG.info("Injector: urlDir: " + urlDir);
}
Path tempDir =
new Path(
getConf().get("mapred.temp.dir", ".")
+ "/inject-temp-"
+ Integer.toString(new Random().nextInt(Integer.MAX_VALUE)));
// map text input file to a <url,CrawlDatum> file
if (LOG.isInfoEnabled()) {
LOG.info("Injector: Converting injected urls to crawl db entries.");
}
JobConf sortJob = new NutchJob(getConf());
sortJob.setJobName("inject " + urlDir);
FileInputFormat.addInputPath(sortJob, urlDir);
sortJob.setMapperClass(InjectMapper.class);
FileOutputFormat.setOutputPath(sortJob, tempDir);
sortJob.setOutputFormat(SequenceFileOutputFormat.class);
sortJob.setOutputKeyClass(Text.class);
sortJob.setOutputValueClass(CrawlDatum.class);
sortJob.setLong("injector.current.time", System.currentTimeMillis());
RunningJob mapJob = JobClient.runJob(sortJob);
long urlsInjected = mapJob.getCounters().findCounter("injector", "urls_injected").getValue();
long urlsFiltered = mapJob.getCounters().findCounter("injector", "urls_filtered").getValue();
LOG.info("Injector: total number of urls rejected by filters: " + urlsFiltered);
LOG.info(
"Injector: total number of urls injected after normalization and filtering: "
+ urlsInjected);
// merge with existing crawl db
if (LOG.isInfoEnabled()) {
LOG.info("Injector: Merging injected urls into crawl db.");
}
JobConf mergeJob = CrawlDb.createJob(getConf(), crawlDb);
FileInputFormat.addInputPath(mergeJob, tempDir);
mergeJob.setReducerClass(InjectReducer.class);
JobClient.runJob(mergeJob);
CrawlDb.install(mergeJob, crawlDb);
// clean up
FileSystem fs = FileSystem.get(getConf());
fs.delete(tempDir, true);
long end = System.currentTimeMillis();
LOG.info(
"Injector: finished at "
+ sdf.format(end)
+ ", elapsed: "
+ TimingUtil.elapsedTime(start, end));
}
private void dirCleanup() {
Configuration conf = new Configuration();
try {
FileSystem fs = FileSystem.get(conf);
Path dirPath = new Path(testPath);
if (fs.exists(dirPath)) {
fs.delete(dirPath, true);
}
} catch (IOException ex) {
LOG.warn("IO Error in test cleanup", ex);
}
}
public void copyInitialState(Path origAppDir) throws IOException {
// locate previous snapshot
String newAppDir = this.dag.assertAppPath();
FSRecoveryHandler recoveryHandler = new FSRecoveryHandler(origAppDir.toString(), conf);
// read snapshot against new dependencies
Object snapshot = recoveryHandler.restore();
if (snapshot == null) {
throw new IllegalArgumentException("No previous application state found in " + origAppDir);
}
InputStream logIs = recoveryHandler.getLog();
// modify snapshot state to switch app id
((StreamingContainerManager.CheckpointState) snapshot).setApplicationId(this.dag, conf);
Path checkpointPath = new Path(newAppDir, LogicalPlan.SUBDIR_CHECKPOINTS);
FileSystem fs = FileSystem.newInstance(origAppDir.toUri(), conf);
// remove the path that was created by the storage agent during deserialization and replacement
fs.delete(checkpointPath, true);
// write snapshot to new location
recoveryHandler = new FSRecoveryHandler(newAppDir, conf);
recoveryHandler.save(snapshot);
OutputStream logOs = recoveryHandler.rotateLog();
IOUtils.copy(logIs, logOs);
logOs.flush();
logOs.close();
logIs.close();
// copy sub directories that are not present in target
FileStatus[] lFiles = fs.listStatus(origAppDir);
for (FileStatus f : lFiles) {
if (f.isDirectory()) {
String targetPath = f.getPath().toString().replace(origAppDir.toString(), newAppDir);
if (!fs.exists(new Path(targetPath))) {
LOG.debug("Copying {} to {}", f.getPath(), targetPath);
FileUtil.copy(fs, f.getPath(), fs, new Path(targetPath), false, conf);
// FSUtil.copy(fs, f, fs, new Path(targetPath), false, false, conf);
} else {
LOG.debug("Ignoring {} as it already exists under {}", f.getPath(), targetPath);
// FSUtil.setPermission(fs, new Path(targetPath), new FsPermission((short)0777));
}
}
}
}
public void testFormat() throws Exception {
JobConf job = new JobConf(conf);
FileSystem fs = FileSystem.getLocal(conf);
Path dir = new Path(System.getProperty("test.build.data", ".") + "/mapred");
Path file = new Path(dir, "test.seq");
Reporter reporter = Reporter.NULL;
int seed = new Random().nextInt();
// LOG.info("seed = "+seed);
Random random = new Random(seed);
fs.delete(dir, true);
FileInputFormat.setInputPaths(job, dir);
// for a variety of lengths
for (int length = 0; length < MAX_LENGTH; length += random.nextInt(MAX_LENGTH / 10) + 1) {
// LOG.info("creating; entries = " + length);
// create a file with length entries
SequenceFile.Writer writer =
SequenceFile.createWriter(fs, conf, file, IntWritable.class, BytesWritable.class);
try {
for (int i = 0; i < length; i++) {
IntWritable key = new IntWritable(i);
byte[] data = new byte[random.nextInt(10)];
random.nextBytes(data);
BytesWritable value = new BytesWritable(data);
writer.append(key, value);
}
} finally {
writer.close();
}
// try splitting the file in a variety of sizes
InputFormat<IntWritable, BytesWritable> format =
new SequenceFileInputFormat<IntWritable, BytesWritable>();
IntWritable key = new IntWritable();
BytesWritable value = new BytesWritable();
for (int i = 0; i < 3; i++) {
int numSplits = random.nextInt(MAX_LENGTH / (SequenceFile.SYNC_INTERVAL / 20)) + 1;
// LOG.info("splitting: requesting = " + numSplits);
InputSplit[] splits = format.getSplits(job, numSplits);
// LOG.info("splitting: got = " + splits.length);
// check each split
BitSet bits = new BitSet(length);
for (int j = 0; j < splits.length; j++) {
RecordReader<IntWritable, BytesWritable> reader =
format.getRecordReader(splits[j], job, reporter);
try {
int count = 0;
while (reader.next(key, value)) {
// if (bits.get(key.get())) {
// LOG.info("splits["+j+"]="+splits[j]+" : " +
// key.get());
// LOG.info("@"+reader.getPos());
// }
assertFalse("Key in multiple partitions.", bits.get(key.get()));
bits.set(key.get());
count++;
}
// LOG.info("splits["+j+"]="+splits[j]+" count=" +
// count);
} finally {
reader.close();
}
}
assertEquals("Some keys in no partition.", length, bits.cardinality());
}
}
}
public void doTestTextBatchAppend() throws Exception {
LOG.debug("Starting...");
final long rollCount = 10;
final long batchSize = 2;
final String fileName = "PageView";
String newPath = testPath + "/singleTextBucket";
int totalEvents = 0;
int i = 1, j = 1;
// clear the test directory
Configuration conf = new Configuration();
FileSystem fs = FileSystem.get(conf);
Path dirPath = new Path(newPath);
fs.delete(dirPath, true);
fs.mkdirs(dirPath);
Context context = new Context();
context.put("hdfs.path", newPath);
context.put("hdfs.rollCount", String.valueOf(rollCount));
context.put("hdfs.batchSize", String.valueOf(batchSize));
context.put("hdfs.filePrefix", "pageview");
Channel channel = new MemoryChannel();
Configurables.configure(channel, context);
sink.setChannel(channel);
sink.start();
Calendar eventDate = Calendar.getInstance();
Date currentDate = new Date();
Map<String, String> header = new HashMap<String, String>();
header.put("topic", "PageView");
List<String> bodies = Lists.newArrayList();
// 将测试的事件推入到通道中
for (i = 1; i <= (rollCount * 10) / batchSize; i++) {
Transaction txn = channel.getTransaction();
txn.begin();
for (j = 1; j <= batchSize; j++) {
header.put("timestamp", String.valueOf(currentDate.getTime()));
Event event = new SimpleEvent();
eventDate.clear();
eventDate.set(2014, i, i, i, 0);
String body = "Test." + i + "." + j;
event.setHeaders(header);
event.setBody(body.getBytes());
bodies.add(body);
channel.put(event);
totalEvents++;
}
txn.commit();
txn.close();
// execute sink to process the events
sink.process();
}
sink.stop();
FileStatus[] dirStat = fs.listStatus(dirPath);
Path fList[] = FileUtil.stat2Paths(dirStat);
long expectedFiles = totalEvents / rollCount;
if (totalEvents % rollCount > 0) {
expectedFiles++;
}
Assert.assertEquals(
"num files wrong, found: " + Lists.newArrayList(fList), expectedFiles, fList.length);
// 检查所有写入文件的内容
verifyOutputTextFiles(fs, conf, dirPath.toUri().getPath(), fileName, bodies);
}
static void checkRecords(
Configuration defaults, int noMaps, int noReduces, Path sortInput, Path sortOutput)
throws IOException {
JobConf jobConf = new JobConf(defaults, RecordChecker.class);
jobConf.setJobName("sortvalidate-record-checker");
jobConf.setInputFormat(SequenceFileInputFormat.class);
jobConf.setOutputFormat(SequenceFileOutputFormat.class);
jobConf.setOutputKeyClass(BytesWritable.class);
jobConf.setOutputValueClass(IntWritable.class);
jobConf.setMapperClass(Map.class);
jobConf.setReducerClass(Reduce.class);
JobClient client = new JobClient(jobConf);
ClusterStatus cluster = client.getClusterStatus();
if (noMaps == -1) {
noMaps = cluster.getTaskTrackers() * jobConf.getInt("test.sortvalidate.maps_per_host", 10);
}
if (noReduces == -1) {
noReduces = (int) (cluster.getMaxReduceTasks() * 0.9);
String sortReduces = jobConf.get("test.sortvalidate.reduces_per_host");
if (sortReduces != null) {
noReduces = cluster.getTaskTrackers() * Integer.parseInt(sortReduces);
}
}
jobConf.setNumMapTasks(noMaps);
jobConf.setNumReduceTasks(noReduces);
FileInputFormat.setInputPaths(jobConf, sortInput);
FileInputFormat.addInputPath(jobConf, sortOutput);
Path outputPath = new Path("/tmp/sortvalidate/recordchecker");
FileSystem fs = FileSystem.get(defaults);
if (fs.exists(outputPath)) {
fs.delete(outputPath, true);
}
FileOutputFormat.setOutputPath(jobConf, outputPath);
// Uncomment to run locally in a single process
// job_conf.set("mapred.job.tracker", "local");
Path[] inputPaths = FileInputFormat.getInputPaths(jobConf);
System.out.println(
"\nSortValidator.RecordChecker: Running on "
+ cluster.getTaskTrackers()
+ " nodes to validate sort from "
+ inputPaths[0]
+ ", "
+ inputPaths[1]
+ " into "
+ FileOutputFormat.getOutputPath(jobConf)
+ " with "
+ noReduces
+ " reduces.");
Date startTime = new Date();
System.out.println("Job started: " + startTime);
JobClient.runJob(jobConf);
Date end_time = new Date();
System.out.println("Job ended: " + end_time);
System.out.println(
"The job took " + (end_time.getTime() - startTime.getTime()) / 1000 + " seconds.");
}
static void checkRecords(Configuration defaults, Path sortInput, Path sortOutput)
throws IOException {
FileSystem inputfs = sortInput.getFileSystem(defaults);
FileSystem outputfs = sortOutput.getFileSystem(defaults);
FileSystem defaultfs = FileSystem.get(defaults);
JobConf jobConf = new JobConf(defaults, RecordStatsChecker.class);
jobConf.setJobName("sortvalidate-recordstats-checker");
int noSortReduceTasks = outputfs.listStatus(sortOutput, sortPathsFilter).length;
jobConf.setInt("sortvalidate.sort.reduce.tasks", noSortReduceTasks);
int noSortInputpaths = inputfs.listStatus(sortInput).length;
jobConf.setInputFormat(NonSplitableSequenceFileInputFormat.class);
jobConf.setOutputFormat(SequenceFileOutputFormat.class);
jobConf.setOutputKeyClass(IntWritable.class);
jobConf.setOutputValueClass(RecordStatsChecker.RecordStatsWritable.class);
jobConf.setMapperClass(Map.class);
jobConf.setCombinerClass(Reduce.class);
jobConf.setReducerClass(Reduce.class);
jobConf.setNumMapTasks(noSortReduceTasks);
jobConf.setNumReduceTasks(1);
FileInputFormat.setInputPaths(jobConf, sortInput);
FileInputFormat.addInputPath(jobConf, sortOutput);
Path outputPath = new Path("/tmp/sortvalidate/recordstatschecker");
if (defaultfs.exists(outputPath)) {
defaultfs.delete(outputPath, true);
}
FileOutputFormat.setOutputPath(jobConf, outputPath);
// Uncomment to run locally in a single process
// job_conf.set("mapred.job.tracker", "local");
Path[] inputPaths = FileInputFormat.getInputPaths(jobConf);
System.out.println(
"\nSortValidator.RecordStatsChecker: Validate sort "
+ "from "
+ inputPaths[0]
+ " ("
+ noSortInputpaths
+ " files), "
+ inputPaths[1]
+ " ("
+ noSortReduceTasks
+ " files) into "
+ FileOutputFormat.getOutputPath(jobConf)
+ " with 1 reducer.");
Date startTime = new Date();
System.out.println("Job started: " + startTime);
JobClient.runJob(jobConf);
Date end_time = new Date();
System.out.println("Job ended: " + end_time);
System.out.println(
"The job took " + (end_time.getTime() - startTime.getTime()) / 1000 + " seconds.");
// Check to ensure that the statistics of the
// framework's sort-input and sort-output match
SequenceFile.Reader stats =
new SequenceFile.Reader(defaultfs, new Path(outputPath, "part-00000"), defaults);
IntWritable k1 = new IntWritable();
IntWritable k2 = new IntWritable();
RecordStatsWritable v1 = new RecordStatsWritable();
RecordStatsWritable v2 = new RecordStatsWritable();
if (!stats.next(k1, v1)) {
throw new IOException("Failed to read record #1 from reduce's output");
}
if (!stats.next(k2, v2)) {
throw new IOException("Failed to read record #2 from reduce's output");
}
if ((v1.getBytes() != v2.getBytes())
|| (v1.getRecords() != v2.getRecords())
|| v1.getChecksum() != v2.getChecksum()) {
throw new IOException(
"("
+ v1.getBytes()
+ ", "
+ v1.getRecords()
+ ", "
+ v1.getChecksum()
+ ") v/s ("
+ v2.getBytes()
+ ", "
+ v2.getRecords()
+ ", "
+ v2.getChecksum()
+ ")");
}
}
