int start(
Configuration conf, String traceIn, Path ioPath, long genbytes, UserResolver userResolver)
throws IOException, InterruptedException {
InputStream trace = null;
try {
Path scratchDir = new Path(ioPath, conf.get(GRIDMIX_OUT_DIR, "gridmix"));
final FileSystem scratchFs = scratchDir.getFileSystem(conf);
scratchFs.mkdirs(scratchDir, new FsPermission((short) 0777));
scratchFs.setPermission(scratchDir, new FsPermission((short) 0777));
// add shutdown hook for SIGINT, etc.
Runtime.getRuntime().addShutdownHook(sdh);
CountDownLatch startFlag = new CountDownLatch(1);
try {
// Create, start job submission threads
startThreads(conf, traceIn, ioPath, scratchDir, startFlag, userResolver);
// Write input data if specified
if (genbytes > 0) {
writeInputData(genbytes, ioPath);
}
// scan input dir contents
submitter.refreshFilePool();
factory.start();
statistics.start();
} catch (Throwable e) {
LOG.error("Startup failed", e);
if (factory != null) factory.abort(); // abort pipeline
} finally {
// signal for factory to start; sets start time
startFlag.countDown();
}
if (factory != null) {
// wait for input exhaustion
factory.join(Long.MAX_VALUE);
final Throwable badTraceException = factory.error();
if (null != badTraceException) {
LOG.error("Error in trace", badTraceException);
throw new IOException("Error in trace", badTraceException);
}
// wait for pending tasks to be submitted
submitter.shutdown();
submitter.join(Long.MAX_VALUE);
// wait for running tasks to complete
monitor.shutdown();
monitor.join(Long.MAX_VALUE);
statistics.shutdown();
statistics.join(Long.MAX_VALUE);
}
} finally {
IOUtils.cleanup(LOG, trace);
}
return 0;
}
/**
* Write random bytes at the path provided.
*
* @see org.apache.hadoop.mapred.gridmix.GenerateData
*/
protected void writeInputData(long genbytes, Path ioPath)
throws IOException, InterruptedException {
final Configuration conf = getConf();
final GridmixJob genData = new GenerateData(conf, ioPath, genbytes);
submitter.add(genData);
LOG.info("Generating " + StringUtils.humanReadableInt(genbytes) + " of test data...");
// TODO add listeners, use for job dependencies
TimeUnit.SECONDS.sleep(10);
try {
genData.getJob().waitForCompletion(false);
} catch (ClassNotFoundException e) {
throw new IOException("Internal error", e);
}
if (!genData.getJob().isSuccessful()) {
throw new IOException("Data generation failed!");
}
FsShell shell = new FsShell(conf);
try {
LOG.info("Changing the permissions for inputPath " + ioPath.toString());
shell.run(new String[] {"-chmod", "-R", "777", ioPath.toString()});
} catch (Exception e) {
LOG.error("Couldnt change the file permissions ", e);
throw new IOException(e);
}
LOG.info("Done.");
}
/**
* Create each component in the pipeline and start it.
*
* @param conf Configuration data, no keys specific to this context
* @param traceIn Either a Path to the trace data or "-" for stdin
* @param ioPath <ioPath>/input/ is the dir from which input data is read and
* <ioPath>/distributedCache/ is the gridmix distributed cache directory.
* @param scratchDir Path into which job output is written
* @param startFlag Semaphore for starting job trace pipeline
*/
private void startThreads(
Configuration conf,
String traceIn,
Path ioPath,
Path scratchDir,
CountDownLatch startFlag,
UserResolver userResolver)
throws IOException {
try {
Path inputDir = getGridmixInputDataPath(ioPath);
GridmixJobSubmissionPolicy policy = getJobSubmissionPolicy(conf);
LOG.info(" Submission policy is " + policy.name());
statistics = new Statistics(conf, policy.getPollingInterval(), startFlag);
monitor = createJobMonitor(statistics, conf);
int noOfSubmitterThreads =
(policy == GridmixJobSubmissionPolicy.SERIAL)
? 1
: Runtime.getRuntime().availableProcessors() + 1;
int numThreads = conf.getInt(GRIDMIX_SUB_THR, noOfSubmitterThreads);
int queueDep = conf.getInt(GRIDMIX_QUE_DEP, 5);
submitter =
createJobSubmitter(
monitor,
numThreads,
queueDep,
new FilePool(conf, inputDir),
userResolver,
statistics);
distCacheEmulator = new DistributedCacheEmulator(conf, ioPath);
factory = createJobFactory(submitter, traceIn, scratchDir, conf, startFlag, userResolver);
factory.jobCreator.setDistCacheEmulator(distCacheEmulator);
if (policy == GridmixJobSubmissionPolicy.SERIAL) {
statistics.addJobStatsListeners(factory);
} else {
statistics.addClusterStatsObservers(factory);
}
// add the gridmix run summarizer to the statistics
statistics.addJobStatsListeners(summarizer.getExecutionSummarizer());
statistics.addClusterStatsObservers(summarizer.getClusterSummarizer());
monitor.start();
submitter.start();
} catch (Exception e) {
LOG.error(" Exception at start ", e);
throw new IOException(e);
}
}
/**
* Called when a new submission message is received. Unpacks the job data and sends the
* information as a SubmissionData to a JobSubmitter for bridging into the Heritrix submission
* system.
*/
public synchronized void onMessage(Message msg) {
if (msg instanceof MapMessage) {
UrlSubmissionMsg subMsg = new UrlSubmissionMsg((MapMessage) msg);
try {
if (!subMsg.validate()) {
throw new JMSException("Invalid UrlSubmissionMsg");
}
SubmissionData data = new SubmissionData();
data.setMetaName(subMsg.getJobID());
data.setSeeds(subMsg.getUrl());
data.setJobDescription(subMsg.getDescription());
CrawlJob job = jobSubmitter.submit(data);
JobFinishedListener listener = new JobFinishedListener(data, msg, this);
inProcess = true;
handler.addJob(job);
CrawlController controller = job.getController();
// This sucks, but an alternative would require heavy hacking.
while (controller == null) {
Thread.sleep(50);
controller = job.getController();
}
controller.addCrawlStatusListener(listener);
sendJobStarted(subMsg.getJobID(), subMsg.getUrl());
// Only take one message at a time.
while (inProcess) {
try {
this.wait();
if (!inProcess) {
break;
}
} catch (InterruptedException e) {
}
}
} catch (JMSException e) {
e.printStackTrace();
} catch (Exception e) {
// Submission went awry
e.printStackTrace();
}
}
} // - onMessage
/**
* Create each component in the pipeline and start it.
*
* @param conf Configuration data, no keys specific to this context
* @param traceIn Either a Path to the trace data or "-" for stdin
* @param ioPath Path from which input data is read
* @param scratchDir Path into which job output is written
* @param startFlag Semaphore for starting job trace pipeline
*/
private void startThreads(
Configuration conf,
String traceIn,
Path ioPath,
Path scratchDir,
CountDownLatch startFlag,
UserResolver userResolver)
throws IOException {
try {
GridmixJobSubmissionPolicy policy =
GridmixJobSubmissionPolicy.getPolicy(conf, GridmixJobSubmissionPolicy.STRESS);
LOG.info(" Submission policy is " + policy.name());
statistics = new Statistics(conf, policy.getPollingInterval(), startFlag);
monitor = createJobMonitor(statistics);
int noOfSubmitterThreads =
(policy == GridmixJobSubmissionPolicy.SERIAL)
? 1
: Runtime.getRuntime().availableProcessors() + 1;
submitter =
createJobSubmitter(
monitor,
conf.getInt(GRIDMIX_SUB_THR, noOfSubmitterThreads),
conf.getInt(GRIDMIX_QUE_DEP, 5),
new FilePool(conf, ioPath),
userResolver,
statistics);
factory = createJobFactory(submitter, traceIn, scratchDir, conf, startFlag, userResolver);
if (policy == GridmixJobSubmissionPolicy.SERIAL) {
statistics.addJobStatsListeners(factory);
} else {
statistics.addClusterStatsObservers(factory);
}
monitor.start();
submitter.start();
} catch (Exception e) {
LOG.error(" Exception at start ", e);
throw new IOException(e);
}
}
// Launch Input/DistCache Data Generation job and wait for completion
void launchGridmixJob(GridmixJob job) throws IOException, InterruptedException {
submitter.add(job);
// TODO add listeners, use for job dependencies
try {
while (!job.isSubmitted()) {
try {
Thread.sleep(100); // sleep
} catch (InterruptedException ie) {
}
}
// wait for completion
job.getJob().waitForCompletion(false);
} catch (ClassNotFoundException e) {
throw new IOException("Internal error", e);
}
if (!job.getJob().isSuccessful()) {
throw new IOException(job.getJob().getJobName() + " job failed!");
}
}
/**
* @param conf gridmix configuration
* @param traceIn trace file path(if it is '-', then trace comes from the stream stdin)
* @param ioPath Working directory for gridmix. GenerateData job will generate data in the
* directory <ioPath>/input/ and distributed cache data is generated in the directory
* <ioPath>/distributedCache/, if -generate option is specified.
* @param genbytes size of input data to be generated under the directory <ioPath>/input/
* @param userResolver gridmix user resolver
* @param generate true if -generate option was specified
* @return exit code
* @throws IOException
* @throws InterruptedException
*/
int start(
Configuration conf,
String traceIn,
Path ioPath,
long genbytes,
UserResolver userResolver,
boolean generate)
throws IOException, InterruptedException {
DataStatistics stats = null;
InputStream trace = null;
ioPath = ioPath.makeQualified(ioPath.getFileSystem(conf));
try {
Path scratchDir = new Path(ioPath, conf.get(GRIDMIX_OUT_DIR, "gridmix"));
// add shutdown hook for SIGINT, etc.
Runtime.getRuntime().addShutdownHook(sdh);
CountDownLatch startFlag = new CountDownLatch(1);
try {
// Create, start job submission threads
startThreads(conf, traceIn, ioPath, scratchDir, startFlag, userResolver);
Path inputDir = getGridmixInputDataPath(ioPath);
// Write input data if specified
if (genbytes > 0) {
writeInputData(genbytes, inputDir);
}
// publish the data statistics
stats = GenerateData.publishDataStatistics(inputDir, genbytes, conf);
// scan input dir contents
submitter.refreshFilePool();
// set up the needed things for emulation of various loads
int exitCode = setupEmulation(conf, traceIn, scratchDir, ioPath, generate);
if (exitCode != 0) {
return exitCode;
}
// start the summarizer
summarizer.start(conf);
factory.start();
statistics.start();
} catch (Throwable e) {
LOG.error("Startup failed", e);
if (factory != null) factory.abort(); // abort pipeline
} finally {
// signal for factory to start; sets start time
startFlag.countDown();
}
if (factory != null) {
// wait for input exhaustion
factory.join(Long.MAX_VALUE);
final Throwable badTraceException = factory.error();
if (null != badTraceException) {
LOG.error("Error in trace", badTraceException);
throw new IOException("Error in trace", badTraceException);
}
// wait for pending tasks to be submitted
submitter.shutdown();
submitter.join(Long.MAX_VALUE);
// wait for running tasks to complete
monitor.shutdown();
monitor.join(Long.MAX_VALUE);
statistics.shutdown();
statistics.join(Long.MAX_VALUE);
}
} finally {
if (factory != null) {
summarizer.finalize(factory, traceIn, genbytes, userResolver, stats, conf);
}
IOUtils.cleanup(LOG, trace);
}
return 0;
}
