private static void ignoreAndWait(final Exception e, boolean printException) {
H2O.ignore(e, "Hit HDFS reset problem, retrying...", printException);
try {
Thread.sleep(500);
} catch (InterruptedException ie) {
}
}
/**
* The run method lives for the life of the JobTracker, and removes Jobs that are not still
* running, but which finished a long time ago.
*/
public void run() {
while (shouldRun) {
try {
Thread.sleep(RETIRE_JOB_CHECK_INTERVAL);
} catch (InterruptedException ie) {
}
synchronized (jobs) {
synchronized (jobInitQueue) {
synchronized (jobsByArrival) {
for (Iterator it = jobs.keySet().iterator(); it.hasNext(); ) {
String jobid = (String) it.next();
JobInProgress job = (JobInProgress) jobs.get(jobid);
if (job.getStatus().getRunState() != JobStatus.RUNNING
&& job.getStatus().getRunState() != JobStatus.PREP
&& (job.getFinishTime() + RETIRE_JOB_INTERVAL < System.currentTimeMillis())) {
it.remove();
jobInitQueue.remove(job);
jobsByArrival.remove(job);
}
}
}
}
}
}
}
public static InputStream openStream(Key k, ProgressMonitor pmon) throws IOException {
H2OHdfsInputStream res = null;
Path p = new Path(k.toString());
try {
res = new H2OHdfsInputStream(p, 0, pmon);
} catch (IOException e) {
try {
Thread.sleep(1000);
} catch (Exception ex) {
}
Log.warn("Error while opening HDFS key " + k.toString() + ", will wait and retry.");
res = new H2OHdfsInputStream(p, 0, pmon);
}
return res;
}
/**
* The run method lives for the life of the JobTracker, and removes TaskTrackers that have not
* checked in for some time.
*/
public void run() {
while (shouldRun) {
//
// Thread runs periodically to check whether trackers should be expired.
// The sleep interval must be no more than half the maximum expiry time
// for a task tracker.
//
try {
Thread.sleep(TASKTRACKER_EXPIRY_INTERVAL / 3);
} catch (InterruptedException ie) {
}
//
// Loop through all expired items in the queue
//
synchronized (taskTrackers) {
synchronized (trackerExpiryQueue) {
long now = System.currentTimeMillis();
TaskTrackerStatus leastRecent = null;
while ((trackerExpiryQueue.size() > 0)
&& ((leastRecent = (TaskTrackerStatus) trackerExpiryQueue.first()) != null)
&& (now - leastRecent.getLastSeen() > TASKTRACKER_EXPIRY_INTERVAL)) {
// Remove profile from head of queue
trackerExpiryQueue.remove(leastRecent);
String trackerName = leastRecent.getTrackerName();
// Figure out if last-seen time should be updated, or if tracker is dead
TaskTrackerStatus newProfile =
(TaskTrackerStatus) taskTrackers.get(leastRecent.getTrackerName());
// Items might leave the taskTracker set through other means; the
// status stored in 'taskTrackers' might be null, which means the
// tracker has already been destroyed.
if (newProfile != null) {
if (now - newProfile.getLastSeen() > TASKTRACKER_EXPIRY_INTERVAL) {
// Remove completely
updateTaskTrackerStatus(trackerName, null);
lostTaskTracker(leastRecent.getTrackerName());
} else {
// Update time by inserting latest profile
trackerExpiryQueue.add(newProfile);
}
}
}
}
}
}
}
public static void startTracker(Configuration conf) throws IOException {
if (tracker != null) throw new IOException("JobTracker already running.");
while (true) {
try {
tracker = new JobTracker(conf);
break;
} catch (IOException e) {
LOG.log(Level.WARNING, "Starting tracker", e);
}
try {
Thread.sleep(1000);
} catch (InterruptedException e) {
}
}
tracker.offerService();
}
private static String lock(String lock) {
String realPath = "";
String parent = "/lock";
String lockName = parent + "/" + lock;
logger.debug("Getting lock " + lockName);
try {
if (zkInstance.exists(parent, false) == null)
zkInstance.create(parent, new byte[0], ZooDefs.Ids.OPEN_ACL_UNSAFE, CreateMode.fromFlag(0));
} catch (Exception E) {
logger.error("Error creating lock node: " + E.toString());
return null;
}
List<String> children = new LinkedList<String>();
try {
// List <ACL> ACLList = zkInstance.getACL(lockName, zkInstance.exists(lock, false));
realPath =
zkInstance.create(
lockName, new byte[0], ZooDefs.Ids.OPEN_ACL_UNSAFE, CreateMode.EPHEMERAL_SEQUENTIAL);
// children = zkInstance.getChildren(realPath, false);
checkLock:
while (true) {
children = zkInstance.getChildren(parent, false);
for (String curChild : children) {
String child = parent + "/" + curChild;
// System.out.println(child + " " + realPath + " " +
// Integer.toString(child.compareTo(realPath)));
if (child.compareTo(realPath) < 0
&& child.length() == realPath.length()
&& curChild.startsWith(lock)) {
// System.out.println(child + " cmp to " + realPath);
Thread.sleep(300);
continue checkLock;
}
}
logger.info("Got lock " + lockName);
return realPath;
}
} catch (Exception E) {
logger.error("Exception while trying to get lock " + lockName + " :" + E.toString());
E.printStackTrace();
return null;
}
}
/**
* Map method. Copies one file from source file system to destination.
*
* @param key src len
* @param value FilePair (FileStatus src, Path dst)
* @param out Log of failed copies
* @param reporter
*/
public void map(
LongWritable key,
FilePair value,
OutputCollector<WritableComparable<?>, Text> out,
Reporter reporter)
throws IOException {
final FileStatus srcstat = value.input;
final Path relativedst = new Path(value.output);
try {
copy(srcstat, relativedst, out, reporter);
} catch (IOException e) {
++failcount;
reporter.incrCounter(Counter.FAIL, 1);
updateStatus(reporter);
final String sfailure = "FAIL " + relativedst + " : " + StringUtils.stringifyException(e);
out.collect(null, new Text(sfailure));
LOG.info(sfailure);
try {
for (int i = 0; i < 3; ++i) {
try {
final Path tmp = new Path(job.get(TMP_DIR_LABEL), relativedst);
if (destFileSys.delete(tmp, true)) break;
} catch (Throwable ex) {
// ignore, we are just cleaning up
LOG.debug("Ignoring cleanup exception", ex);
}
// update status, so we don't get timed out
updateStatus(reporter);
Thread.sleep(3 * 1000);
}
} catch (InterruptedException inte) {
throw (IOException) new IOException().initCause(inte);
}
} finally {
updateStatus(reporter);
}
}
public void incrementAndThrottle(int bytes) {
if (bytesPerSec < 1) { // no throttle at all
return;
}
long currentTime = System.currentTimeMillis();
long timeDiff = currentTime - lastTime;
if (timeDiff == 0) {
timeDiff = 1;
}
double bytesPerSec = (bytes / (double) timeDiff) * 1000;
if (bytesPerSec > this.bytesPerSec) {
// Throttle
double exceededByFactorOf = bytesPerSec / this.bytesPerSec;
try {
long mustSleep = (long) ((exceededByFactorOf - 1) * timeDiff);
Thread.sleep(mustSleep);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
lastTime = System.currentTimeMillis();
}
/**
* Method to move files from HDFS to local filesystem
*
* <p>localPath: Path on the machines filesystem fs:FileSystem object from HDFS pathList:List of
* paths for files that might need to be backed up size:max size in bytes to be backed up
*
* <p>ReturnsDate of the last files backed up if reached size limit, else, zero
*/
public long backupFiles(
String localPath, String preservePath, FileSystem fs, ArrayList<Path> pathList, long size) {
Path fsPath;
long tmpSize = 0;
long tmpDate = 0;
// Start iterating over all paths
for (Path hdfsPath : pathList) {
try {
long nFileSize = fs.getContentSummary(hdfsPath).getLength();
tmpSize = tmpSize + nFileSize;
if ((tmpSize <= size) || (size == 0)) {
FileStatus stat = fs.getFileStatus(hdfsPath);
System.err.println(
"File "
+ hdfsPath.toUri().getPath()
+ " "
+ nFileSize
+ " bytes, "
+ "perms: "
+ stat.getOwner()
+ "/"
+ stat.getGroup()
+ ", "
+ stat.getPermission().toString());
tmpDate = stat.getModificationTime() / 1000;
String sFsPath = localPath + hdfsPath.toUri().getPath();
fsPath = new Path(sFsPath);
File f = new File(sFsPath);
// COMMENTED OUT: until a few backup cycles run
// and the mtime gets in fact set on all copied
// files.
//
// ignore it if the file exists and has the same mtime
// if (f.exists() && f.isFile() && f.lastModified() == stat.getModificationTime())
// {
// System.out.println("no need to backup " + f.toString() + ", mtime matches hdfs");
// continue;
// }
if (false == m_bDryRun) {
// check if we need to back up the local file
// (not directory), if it already exists.
if (f.exists() && f.isFile()) {
// ignore files with substrings in the
// no-preserve file
if (true == doPreserveFile(sFsPath)) {
// move it to the backup path
String sNewPath = preservePath + hdfsPath.toUri().getPath();
File newFile = new File(sNewPath);
// create directory structure for new file?
if (false == newFile.getParentFile().exists()) {
if (false == newFile.getParentFile().mkdirs()) {
System.err.println("Failed to mkdirs " + newFile.getParentFile().toString());
System.exit(1);
}
}
// rename existing file to new location
if (false == f.renameTo(newFile)) {
System.err.println(
"Failed to renameTo " + f.toString() + " to " + newFile.toString());
System.exit(1);
}
System.out.println("preserved " + f.toString() + " into " + newFile.toString());
} else {
System.out.println("skipped preservation of " + f.toString());
}
}
// copy from hdfs to local filesystem
fs.copyToLocalFile(hdfsPath, fsPath);
// set the mtime to match hdfs file
f.setLastModified(stat.getModificationTime());
// compare checksums on both files
compareChecksums(fs, hdfsPath, sFsPath);
}
// don't print the progress after every file -- go
// by at least 1% increments
long nPercentDone = (long) (100 * tmpSize / m_nTotalBytes);
if (nPercentDone > m_nLastPercentBytesDone) {
System.out.println(
"progress: copied "
+ prettyPrintBytes(tmpSize)
+ ", "
+ nPercentDone
+ "% done"
+ ", tstamp="
+ tmpDate);
m_nLastPercentBytesDone = nPercentDone;
}
if (m_nSleepSeconds > 0) {
try {
Thread.sleep(1000 * m_nSleepSeconds);
} catch (Exception e2) {
// ignore
}
}
} else {
return tmpDate;
}
} catch (IOException e) {
System.err.println("FATAL ERROR: Something wrong with the file");
System.err.println(e);
System.out.println(tmpDate);
System.exit(1);
return 0;
}
}
return 0;
}