private static void addFolder2(
FileSystem fs, Path p, ArrayList<String> keys, ArrayList<String> failed) {
try {
if (fs == null) return;
Futures futures = new Futures();
for (FileStatus file : fs.listStatus(p)) {
Path pfs = file.getPath();
if (file.isDir()) {
addFolder2(fs, pfs, keys, failed);
} else {
long size = file.getLen();
Key res;
if (pfs.getName().endsWith(Extensions.JSON)) {
throw H2O.unimpl();
} else if (pfs.getName().endsWith(Extensions.HEX)) { // Hex file?
throw H2O.unimpl();
} else {
Key k = null;
keys.add((k = HdfsFileVec.make(file, futures)).toString());
Log.info("PersistHdfs: DKV.put(" + k + ")");
}
}
}
} catch (Exception e) {
Log.err(e);
failed.add(p.toString());
}
}
/**
* Check whether the contents of src and dst are the same.
*
* <p>Return false if dstpath does not exist
*
* <p>If the files have different sizes, return false.
*
* <p>If the files have the same sizes, the file checksums will be compared.
*
* <p>When file checksum is not supported in any of file systems, two files are considered as the
* same if they have the same size.
*/
private static boolean sameFile(
FileSystem srcfs, FileStatus srcstatus, FileSystem dstfs, Path dstpath) throws IOException {
FileStatus dststatus;
try {
dststatus = dstfs.getFileStatus(dstpath);
} catch (FileNotFoundException fnfe) {
return false;
}
// same length?
if (srcstatus.getLen() != dststatus.getLen()) {
return false;
}
// compare checksums
try {
final FileChecksum srccs = srcfs.getFileChecksum(srcstatus.getPath());
final FileChecksum dstcs = dstfs.getFileChecksum(dststatus.getPath());
// return true if checksum is not supported
// (i.e. some of the checksums is null)
return srccs == null || dstcs == null || srccs.equals(dstcs);
} catch (FileNotFoundException fnfe) {
return false;
}
}
/*
* Fetch a file that is in a Hadoop file system. Return a local File.
* Interruptible.
*/
private File hdfsFetch(Path fromPath, Reporter reporter)
throws IOException, InterruptedException {
UUID uniqueId = UUID.randomUUID();
File toFile = new File(tempDir, uniqueId.toString() + "/" + fromPath.getName());
File toDir = new File(toFile.getParent());
if (toDir.exists()) {
FileUtils.deleteDirectory(toDir);
}
toDir.mkdirs();
Path toPath = new Path(toFile.getCanonicalPath());
FileSystem fS = fromPath.getFileSystem(hadoopConf);
FileSystem tofS = FileSystem.getLocal(hadoopConf);
Throttler throttler = new Throttler((double) bytesPerSecThrottle);
try {
for (FileStatus fStatus : fS.globStatus(fromPath)) {
log.info("Copying " + fStatus.getPath() + " to " + toPath);
long bytesSoFar = 0;
FSDataInputStream iS = fS.open(fStatus.getPath());
FSDataOutputStream oS = tofS.create(toPath);
byte[] buffer = new byte[downloadBufferSize];
int nRead;
while ((nRead = iS.read(buffer, 0, buffer.length)) != -1) {
// Needed to being able to be interrupted at any moment.
if (Thread.interrupted()) {
iS.close();
oS.close();
cleanDirNoExceptions(toDir);
throw new InterruptedException();
}
bytesSoFar += nRead;
oS.write(buffer, 0, nRead);
throttler.incrementAndThrottle(nRead);
if (bytesSoFar >= bytesToReportProgress) {
reporter.progress(bytesSoFar);
bytesSoFar = 0l;
}
}
if (reporter != null) {
reporter.progress(bytesSoFar);
}
oS.close();
iS.close();
}
return toDir;
} catch (ClosedByInterruptException e) {
// This can be thrown by the method read.
cleanDirNoExceptions(toDir);
throw new InterruptedIOException();
}
}
private static void addFolder(FileSystem fs, Path p, JsonArray succeeded, JsonArray failed) {
try {
if (fs == null) return;
for (FileStatus file : fs.listStatus(p)) {
Path pfs = file.getPath();
if (file.isDir()) {
addFolder(fs, pfs, succeeded, failed);
} else {
Key k = Key.make(pfs.toString());
long size = file.getLen();
Value val = null;
if (pfs.getName().endsWith(Extensions.JSON)) {
JsonParser parser = new JsonParser();
JsonObject json = parser.parse(new InputStreamReader(fs.open(pfs))).getAsJsonObject();
JsonElement v = json.get(Constants.VERSION);
if (v == null) throw new RuntimeException("Missing version");
JsonElement type = json.get(Constants.TYPE);
if (type == null) throw new RuntimeException("Missing type");
Class c = Class.forName(type.getAsString());
OldModel model = (OldModel) c.newInstance();
model.fromJson(json);
} else if (pfs.getName().endsWith(Extensions.HEX)) { // Hex file?
FSDataInputStream s = fs.open(pfs);
int sz = (int) Math.min(1L << 20, size); // Read up to the 1st meg
byte[] mem = MemoryManager.malloc1(sz);
s.readFully(mem);
// Convert to a ValueArray (hope it fits in 1Meg!)
ValueArray ary = new ValueArray(k, 0).read(new AutoBuffer(mem));
val = new Value(k, ary, Value.HDFS);
} else if (size >= 2 * ValueArray.CHUNK_SZ) {
val =
new Value(
k,
new ValueArray(k, size),
Value.HDFS); // ValueArray byte wrapper over a large file
} else {
val = new Value(k, (int) size, Value.HDFS); // Plain Value
val.setdsk();
}
DKV.put(k, val);
Log.info("PersistHdfs: DKV.put(" + k + ")");
JsonObject o = new JsonObject();
o.addProperty(Constants.KEY, k.toString());
o.addProperty(Constants.FILE, pfs.toString());
o.addProperty(Constants.VALUE_SIZE, file.getLen());
succeeded.add(o);
}
}
} catch (Exception e) {
Log.err(e);
JsonObject o = new JsonObject();
o.addProperty(Constants.FILE, p.toString());
o.addProperty(Constants.ERROR, e.getMessage());
failed.add(o);
}
}
/**
* Generate the list of files and make them into FileSplits. This needs to be copied to insert a
* filter on acceptable data
*/
@Override
public List<InputSplit> getSplits(JobContext job) throws IOException {
long minSize = Math.max(getFormatMinSplitSize(), getMinSplitSize(job));
long maxSize = getMaxSplitSize(job);
long desiredMappers =
job.getConfiguration().getLong("org.systemsbiology.jxtandem.DesiredXMLInputMappers", 0);
// generate splits
List<InputSplit> splits = new ArrayList<InputSplit>();
List<FileStatus> fileStatuses = listStatus(job);
boolean forceNumberMappers = fileStatuses.size() == 1;
for (FileStatus file : fileStatuses) {
Path path = file.getPath();
if (!isPathAcceptable(path)) // filter acceptable data
continue;
FileSystem fs = path.getFileSystem(job.getConfiguration());
long length = file.getLen();
BlockLocation[] blkLocations = fs.getFileBlockLocations(file, 0, length);
if ((length != 0) && isSplitable(job, path)) {
long blockSize = file.getBlockSize();
// use desired mappers to force more splits
if (forceNumberMappers && desiredMappers > 0)
maxSize = Math.min(maxSize, (length / desiredMappers));
long splitSize = computeSplitSize(blockSize, minSize, maxSize);
long bytesRemaining = length;
while (withinSlop(splitSize, bytesRemaining)) {
int blkIndex = getBlockIndex(blkLocations, length - bytesRemaining);
splits.add(
new FileSplit(
path, length - bytesRemaining, splitSize, blkLocations[blkIndex].getHosts()));
bytesRemaining -= splitSize;
}
if (bytesRemaining != 0) {
splits.add(
new FileSplit(
path,
length - bytesRemaining,
bytesRemaining,
blkLocations[blkLocations.length - 1].getHosts()));
}
} else if (length != 0) {
splits.add(new FileSplit(path, 0, length, blkLocations[0].getHosts()));
} else {
// Create empty hosts array for zero length files
splits.add(new FileSplit(path, 0, length, new String[0]));
}
}
System.out.println("Total # of splits: " + splits.size());
// LOG.debug("Total # of splits: " + splits.size());
return splits;
}
/**
* Produce splits such that each is no greater than the quotient of the total size and the
* number of splits requested.
*
* @param job The handle to the JobConf object
* @param numSplits Number of splits requested
*/
public InputSplit[] getSplits(JobConf job, int numSplits) throws IOException {
int cnfiles = job.getInt(SRC_COUNT_LABEL, -1);
long cbsize = job.getLong(TOTAL_SIZE_LABEL, -1);
String srcfilelist = job.get(SRC_LIST_LABEL, "");
if (cnfiles < 0 || cbsize < 0 || "".equals(srcfilelist)) {
throw new RuntimeException(
"Invalid metadata: #files("
+ cnfiles
+ ") total_size("
+ cbsize
+ ") listuri("
+ srcfilelist
+ ")");
}
Path src = new Path(srcfilelist);
FileSystem fs = src.getFileSystem(job);
FileStatus srcst = fs.getFileStatus(src);
ArrayList<FileSplit> splits = new ArrayList<FileSplit>(numSplits);
LongWritable key = new LongWritable();
FilePair value = new FilePair();
final long targetsize = cbsize / numSplits;
long pos = 0L;
long last = 0L;
long acc = 0L;
long cbrem = srcst.getLen();
SequenceFile.Reader sl = null;
try {
sl = new SequenceFile.Reader(fs, src, job);
for (; sl.next(key, value); last = sl.getPosition()) {
// if adding this split would put this split past the target size,
// cut the last split and put this next file in the next split.
if (acc + key.get() > targetsize && acc != 0) {
long splitsize = last - pos;
splits.add(new FileSplit(src, pos, splitsize, (String[]) null));
cbrem -= splitsize;
pos = last;
acc = 0L;
}
acc += key.get();
}
} finally {
checkAndClose(sl);
}
if (cbrem != 0) {
splits.add(new FileSplit(src, pos, cbrem, (String[]) null));
}
return splits.toArray(new FileSplit[splits.size()]);
}
private static Path checkDest(String srcName, FileSystem dstFS, Path dst, boolean overwrite)
throws IOException {
if (dstFS.exists(dst)) {
FileStatus sdst = dstFS.getFileStatus(dst);
if (sdst.isDir()) {
if (null == srcName) {
throw new IOException("Target " + dst + " is a directory");
}
return checkDest(null, dstFS, new Path(dst, srcName), overwrite);
} else if (!overwrite) {
throw new IOException("Target " + dst + " already exists");
}
}
return dst;
}
// Checks if the cache has already been localized and is fresh
private static boolean ifExistsAndFresh(
Configuration conf,
FileSystem fs,
URI cache,
long confFileStamp,
CacheStatus lcacheStatus,
FileStatus fileStatus)
throws IOException {
// check for existence of the cache
long dfsFileStamp;
if (fileStatus != null) {
dfsFileStamp = fileStatus.getModificationTime();
} else {
dfsFileStamp = getTimestamp(conf, cache);
}
// ensure that the file on hdfs hasn't been modified since the job started
if (dfsFileStamp != confFileStamp) {
LOG.fatal("File: " + cache + " has changed on HDFS since job started");
throw new IOException("File: " + cache + " has changed on HDFS since job started");
}
if (dfsFileStamp != lcacheStatus.mtime) {
// needs refreshing
return false;
}
return true;
}
/** Added for back compatibility. */
public static Path getLocalCache(
URI cache,
Configuration conf,
Path subdir,
FileStatus fileStatus,
boolean isArchive,
long confFileStamp,
Path currentWorkDir,
boolean honorSymLinkConf,
MRAsyncDiskService asyncDiskService,
LocalDirAllocator lDirAllocator)
throws IOException {
return getLocalCache(
cache,
conf,
subdir,
fileStatus,
isArchive,
confFileStamp,
fileStatus.getLen(),
currentWorkDir,
honorSymLinkConf,
asyncDiskService,
lDirAllocator);
}
/** Implement the delete(Path, boolean) in checksum file system. */
public boolean delete(Path f, boolean recursive) throws IOException {
FileStatus fstatus = null;
try {
fstatus = fs.getFileStatus(f);
} catch (FileNotFoundException e) {
return false;
}
if (fstatus.isDir()) {
// this works since the crcs are in the same
// directories and the files. so we just delete
// everything in the underlying filesystem
return fs.delete(f, recursive);
} else {
Path checkFile = getChecksumFile(f);
if (fs.exists(checkFile)) {
fs.delete(checkFile, true);
}
return fs.delete(f, true);
}
}
private FSDataOutputStream create(Path f, Reporter reporter, FileStatus srcstat)
throws IOException {
if (destFileSys.exists(f)) {
destFileSys.delete(f, false);
}
if (!preserve_status) {
return destFileSys.create(f, true, sizeBuf, reporter);
}
FsPermission permission =
preseved.contains(FileAttribute.PERMISSION) ? srcstat.getPermission() : null;
short replication =
preseved.contains(FileAttribute.REPLICATION)
? srcstat.getReplication()
: destFileSys.getDefaultReplication();
long blockSize =
preseved.contains(FileAttribute.BLOCK_SIZE)
? srcstat.getBlockSize()
: destFileSys.getDefaultBlockSize();
return destFileSys.create(f, permission, true, sizeBuf, replication, blockSize, reporter);
}
/**
* list subfiles
*
* @param hdfsPath !null path probably exists
* @return !null list of enclused file names - emptu if file of not exists
*/
@Override
public String[] ls(final String hdfsPath) {
if (isRunningAsUser()) {
return super.ls(hdfsPath);
}
final FileSystem fs = getDFS();
try {
FileStatus[] statuses = fs.listStatus(new Path(hdfsPath));
if (statuses == null) return new String[0];
List<String> holder = new ArrayList<String>();
for (int i = 0; i < statuses.length; i++) {
FileStatus statuse = statuses[i];
String s = statuse.getPath().getName();
holder.add(s);
}
String[] ret = new String[holder.size()];
holder.toArray(ret);
return ret;
} catch (IOException e) {
throw new RuntimeException(e);
}
}
public static boolean copy(
FileSystem srcFS,
Path[] srcs,
FileSystem dstFS,
Path dst,
boolean deleteSource,
boolean overwrite,
Configuration conf)
throws IOException {
boolean gotException = false;
boolean returnVal = true;
StringBuffer exceptions = new StringBuffer();
if (srcs.length == 1) return copy(srcFS, srcs[0], dstFS, dst, deleteSource, overwrite, conf);
// Check if dest is directory
if (!dstFS.exists(dst)) {
throw new IOException("`" + dst + "': specified destination directory " + "doest not exist");
} else {
FileStatus sdst = dstFS.getFileStatus(dst);
if (!sdst.isDir())
throw new IOException(
"copying multiple files, but last argument `" + dst + "' is not a directory");
}
for (Path src : srcs) {
try {
if (!copy(srcFS, src, dstFS, dst, deleteSource, overwrite, conf)) returnVal = false;
} catch (IOException e) {
gotException = true;
exceptions.append(e.getMessage());
exceptions.append("\n");
}
}
if (gotException) {
throw new IOException(exceptions.toString());
}
return returnVal;
}
/**
* The src file is under FS, and the dst is on the local disk. Copy it from FS control to the
* local dst name. If src and dst are directories, the copyCrc parameter determines whether to
* copy CRC files.
*/
public void copyToLocalFile(Path src, Path dst, boolean copyCrc) throws IOException {
if (!fs.isDirectory(src)) { // source is a file
fs.copyToLocalFile(src, dst);
FileSystem localFs = getLocal(getConf()).getRawFileSystem();
if (localFs.isDirectory(dst)) {
dst = new Path(dst, src.getName());
}
dst = getChecksumFile(dst);
if (localFs.exists(dst)) { // remove old local checksum file
localFs.delete(dst, true);
}
Path checksumFile = getChecksumFile(src);
if (copyCrc && fs.exists(checksumFile)) { // copy checksum file
fs.copyToLocalFile(checksumFile, dst);
}
} else {
FileStatus[] srcs = listStatus(src);
for (FileStatus srcFile : srcs) {
copyToLocalFile(srcFile.getPath(), new Path(dst, srcFile.getPath().getName()), copyCrc);
}
}
}
public static void recursePath(Configuration conf, Path path, Job job) {
try {
FileSystem fs = path.getFileSystem(conf);
FileStatus[] fstats = fs.listStatus(path);
if (fstats != null) {
for (FileStatus f : fstats) {
Path p = f.getPath();
;
if (fs.isFile(p)) {
// connection times out otherwise
System.err.println("file:" + p.toString());
FileInputFormat.addInputPath(job, p);
} else {
System.err.println("dir:" + p.toString());
recursePath(conf, p, job);
}
}
}
} catch (IOException e) {
// shouldn't be here
throw new RuntimeException(e);
}
}
/**
* Get the locally cached file or archive; it could either be previously cached (and valid) or
* copy it from the {@link FileSystem} now.
*
* @param cache the cache to be localized, this should be specified as new
* URI(hdfs://hostname:port/absolute_path_to_file#LINKNAME). If no schema or hostname:port is
* provided the file is assumed to be in the filesystem being used in the Configuration
* @param conf The Confguration file which contains the filesystem
* @param baseDir The base cache Dir where you wnat to localize the files/archives
* @param fileStatus The file status on the dfs.
* @param isArchive if the cache is an archive or a file. In case it is an archive with a .zip or
* .jar or .tar or .tgz or .tar.gz extension it will be unzipped/unjarred/untarred
* automatically and the directory where the archive is unzipped/unjarred/untarred is returned
* as the Path. In case of a file, the path to the file is returned
* @param confFileStamp this is the hdfs file modification timestamp to verify that the file to be
* cached hasn't changed since the job started
* @param currentWorkDir this is the directory where you would want to create symlinks for the
* locally cached files/archives
* @return the path to directory where the archives are unjarred in case of archives, the path to
* the file where the file is copied locally
* @throws IOException
*/
public static Path getLocalCache(
URI cache,
Configuration conf,
Path baseDir,
FileStatus fileStatus,
boolean isArchive,
long confFileStamp,
Path currentWorkDir,
MRAsyncDiskService asyncDiskService)
throws IOException {
return getLocalCache(
cache,
conf,
baseDir,
fileStatus,
isArchive,
confFileStamp,
fileStatus.getLen(),
currentWorkDir,
true,
asyncDiskService,
new LocalDirAllocator("mapred.local.dir"));
}
/**
* Initialize DFSCopyFileMapper specific job-configuration.
*
* @param conf : The dfs/mapred configuration.
* @param jobConf : The handle to the jobConf object to be initialized.
* @param args Arguments
*/
private static void setup(Configuration conf, JobConf jobConf, final Arguments args)
throws IOException {
jobConf.set(DST_DIR_LABEL, args.dst.toUri().toString());
// set boolean values
final boolean update = args.flags.contains(Options.UPDATE);
final boolean overwrite = !update && args.flags.contains(Options.OVERWRITE);
jobConf.setBoolean(Options.UPDATE.propertyname, update);
jobConf.setBoolean(Options.OVERWRITE.propertyname, overwrite);
jobConf.setBoolean(
Options.IGNORE_READ_FAILURES.propertyname,
args.flags.contains(Options.IGNORE_READ_FAILURES));
jobConf.setBoolean(
Options.PRESERVE_STATUS.propertyname, args.flags.contains(Options.PRESERVE_STATUS));
final String randomId = getRandomId();
JobClient jClient = new JobClient(jobConf);
Path jobDirectory = new Path(jClient.getSystemDir(), NAME + "_" + randomId);
jobConf.set(JOB_DIR_LABEL, jobDirectory.toString());
FileSystem dstfs = args.dst.getFileSystem(conf);
boolean dstExists = dstfs.exists(args.dst);
boolean dstIsDir = false;
if (dstExists) {
dstIsDir = dstfs.getFileStatus(args.dst).isDir();
}
// default logPath
Path logPath = args.log;
if (logPath == null) {
String filename = "_distcp_logs_" + randomId;
if (!dstExists || !dstIsDir) {
Path parent = args.dst.getParent();
if (!dstfs.exists(parent)) {
dstfs.mkdirs(parent);
}
logPath = new Path(parent, filename);
} else {
logPath = new Path(args.dst, filename);
}
}
FileOutputFormat.setOutputPath(jobConf, logPath);
// create src list, dst list
FileSystem jobfs = jobDirectory.getFileSystem(jobConf);
Path srcfilelist = new Path(jobDirectory, "_distcp_src_files");
jobConf.set(SRC_LIST_LABEL, srcfilelist.toString());
SequenceFile.Writer src_writer =
SequenceFile.createWriter(
jobfs,
jobConf,
srcfilelist,
LongWritable.class,
FilePair.class,
SequenceFile.CompressionType.NONE);
Path dstfilelist = new Path(jobDirectory, "_distcp_dst_files");
SequenceFile.Writer dst_writer =
SequenceFile.createWriter(
jobfs, jobConf, dstfilelist, Text.class, Text.class, SequenceFile.CompressionType.NONE);
Path dstdirlist = new Path(jobDirectory, "_distcp_dst_dirs");
jobConf.set(DST_DIR_LIST_LABEL, dstdirlist.toString());
SequenceFile.Writer dir_writer =
SequenceFile.createWriter(
jobfs,
jobConf,
dstdirlist,
Text.class,
FilePair.class,
SequenceFile.CompressionType.NONE);
// handle the case where the destination directory doesn't exist
// and we've only a single src directory OR we're updating/overwriting
// the contents of the destination directory.
final boolean special = (args.srcs.size() == 1 && !dstExists) || update || overwrite;
int srcCount = 0, cnsyncf = 0, dirsyn = 0;
long fileCount = 0L, byteCount = 0L, cbsyncs = 0L;
try {
for (Iterator<Path> srcItr = args.srcs.iterator(); srcItr.hasNext(); ) {
final Path src = srcItr.next();
FileSystem srcfs = src.getFileSystem(conf);
FileStatus srcfilestat = srcfs.getFileStatus(src);
Path root = special && srcfilestat.isDir() ? src : src.getParent();
if (srcfilestat.isDir()) {
++srcCount;
}
Stack<FileStatus> pathstack = new Stack<FileStatus>();
for (pathstack.push(srcfilestat); !pathstack.empty(); ) {
FileStatus cur = pathstack.pop();
FileStatus[] children = srcfs.listStatus(cur.getPath());
for (int i = 0; i < children.length; i++) {
boolean skipfile = false;
final FileStatus child = children[i];
final String dst = makeRelative(root, child.getPath());
++srcCount;
if (child.isDir()) {
pathstack.push(child);
} else {
// skip file if the src and the dst files are the same.
skipfile = update && sameFile(srcfs, child, dstfs, new Path(args.dst, dst));
// skip file if it exceed file limit or size limit
skipfile |=
fileCount == args.filelimit || byteCount + child.getLen() > args.sizelimit;
if (!skipfile) {
++fileCount;
byteCount += child.getLen();
if (LOG.isTraceEnabled()) {
LOG.trace("adding file " + child.getPath());
}
++cnsyncf;
cbsyncs += child.getLen();
if (cnsyncf > SYNC_FILE_MAX || cbsyncs > BYTES_PER_MAP) {
src_writer.sync();
dst_writer.sync();
cnsyncf = 0;
cbsyncs = 0L;
}
}
}
if (!skipfile) {
src_writer.append(
new LongWritable(child.isDir() ? 0 : child.getLen()), new FilePair(child, dst));
}
dst_writer.append(new Text(dst), new Text(child.getPath().toString()));
}
if (cur.isDir()) {
String dst = makeRelative(root, cur.getPath());
dir_writer.append(new Text(dst), new FilePair(cur, dst));
if (++dirsyn > SYNC_FILE_MAX) {
dirsyn = 0;
dir_writer.sync();
}
}
}
}
} finally {
checkAndClose(src_writer);
checkAndClose(dst_writer);
checkAndClose(dir_writer);
}
FileStatus dststatus = null;
try {
dststatus = dstfs.getFileStatus(args.dst);
} catch (FileNotFoundException fnfe) {
LOG.info(args.dst + " does not exist.");
}
// create dest path dir if copying > 1 file
if (dststatus == null) {
if (srcCount > 1 && !dstfs.mkdirs(args.dst)) {
throw new IOException("Failed to create" + args.dst);
}
}
final Path sorted = new Path(jobDirectory, "_distcp_sorted");
checkDuplication(jobfs, dstfilelist, sorted, conf);
if (dststatus != null && args.flags.contains(Options.DELETE)) {
deleteNonexisting(dstfs, dststatus, sorted, jobfs, jobDirectory, jobConf, conf);
}
Path tmpDir =
new Path(
(dstExists && !dstIsDir) || (!dstExists && srcCount == 1)
? args.dst.getParent()
: args.dst,
"_distcp_tmp_" + randomId);
jobConf.set(TMP_DIR_LABEL, tmpDir.toUri().toString());
LOG.info("srcCount=" + srcCount);
jobConf.setInt(SRC_COUNT_LABEL, srcCount);
jobConf.setLong(TOTAL_SIZE_LABEL, byteCount);
setMapCount(byteCount, jobConf);
}
private static void updatePermissions(
FileStatus src, FileStatus dst, EnumSet<FileAttribute> preseved, FileSystem destFileSys)
throws IOException {
String owner = null;
String group = null;
if (preseved.contains(FileAttribute.USER) && !src.getOwner().equals(dst.getOwner())) {
owner = src.getOwner();
}
if (preseved.contains(FileAttribute.GROUP) && !src.getGroup().equals(dst.getGroup())) {
group = src.getGroup();
}
if (owner != null || group != null) {
destFileSys.setOwner(dst.getPath(), owner, group);
}
if (preseved.contains(FileAttribute.PERMISSION)
&& !src.getPermission().equals(dst.getPermission())) {
destFileSys.setPermission(dst.getPath(), src.getPermission());
}
}
/**
* Copy a file to a destination.
*
* @param srcstat src path and metadata
* @param dstpath dst path
* @param reporter
*/
private void copy(
FileStatus srcstat,
Path relativedst,
OutputCollector<WritableComparable<?>, Text> outc,
Reporter reporter)
throws IOException {
Path absdst = new Path(destPath, relativedst);
int totfiles = job.getInt(SRC_COUNT_LABEL, -1);
assert totfiles >= 0 : "Invalid file count " + totfiles;
// if a directory, ensure created even if empty
if (srcstat.isDir()) {
if (destFileSys.exists(absdst)) {
if (!destFileSys.getFileStatus(absdst).isDir()) {
throw new IOException("Failed to mkdirs: " + absdst + " is a file.");
}
} else if (!destFileSys.mkdirs(absdst)) {
throw new IOException("Failed to mkdirs " + absdst);
}
// TODO: when modification times can be set, directories should be
// emitted to reducers so they might be preserved. Also, mkdirs does
// not currently return an error when the directory already exists;
// if this changes, all directory work might as well be done in reduce
return;
}
if (destFileSys.exists(absdst) && !overwrite && !needsUpdate(srcstat, destFileSys, absdst)) {
outc.collect(null, new Text("SKIP: " + srcstat.getPath()));
++skipcount;
reporter.incrCounter(Counter.SKIP, 1);
updateStatus(reporter);
return;
}
Path tmpfile = new Path(job.get(TMP_DIR_LABEL), relativedst);
long cbcopied = 0L;
FSDataInputStream in = null;
FSDataOutputStream out = null;
try {
// open src file
try {
in = srcstat.getPath().getFileSystem(job).open(srcstat.getPath());
} catch (IOException e) {
LOG.error("Failed to open src file " + srcstat.getPath() + ", ignore and return");
in = null;
return;
}
reporter.incrCounter(Counter.BYTESEXPECTED, srcstat.getLen());
// open tmp file
out = create(tmpfile, reporter, srcstat);
// copy file
for (int cbread; (cbread = in.read(buffer)) >= 0; ) {
out.write(buffer, 0, cbread);
cbcopied += cbread;
reporter.setStatus(
String.format("%.2f ", cbcopied * 100.0 / srcstat.getLen())
+ absdst
+ " [ "
+ StringUtils.humanReadableInt(cbcopied)
+ " / "
+ StringUtils.humanReadableInt(srcstat.getLen())
+ " ]");
}
} finally {
checkAndClose(in);
checkAndClose(out);
}
if (cbcopied != srcstat.getLen()) {
if (srcstat.getLen() == 0 && cbcopied > 0) {
LOG.info("most likely see a WAL file corruption: " + srcstat.getPath());
} else {
throw new IOException(
"File size not matched: copied "
+ bytesString(cbcopied)
+ " to tmpfile (="
+ tmpfile
+ ") but expected "
+ bytesString(srcstat.getLen())
+ " from "
+ srcstat.getPath());
}
} else {
if (totfiles == 1) {
// Copying a single file; use dst path provided by user as destination
// rather than destination directory, if a file
Path dstparent = absdst.getParent();
if (!(destFileSys.exists(dstparent) && destFileSys.getFileStatus(dstparent).isDir())) {
absdst = dstparent;
}
}
if (destFileSys.exists(absdst) && destFileSys.getFileStatus(absdst).isDir()) {
throw new IOException(absdst + " is a directory");
}
if (!destFileSys.mkdirs(absdst.getParent())) {
throw new IOException("Failed to craete parent dir: " + absdst.getParent());
}
rename(tmpfile, absdst);
FileStatus dststat = destFileSys.getFileStatus(absdst);
if (dststat.getLen() != srcstat.getLen()) {
destFileSys.delete(absdst, false);
throw new IOException(
"File size not matched: copied "
+ bytesString(dststat.getLen())
+ " to dst (="
+ absdst
+ ") but expected "
+ bytesString(srcstat.getLen())
+ " from "
+ srcstat.getPath());
}
updatePermissions(srcstat, dststat);
}
// report at least once for each file
++copycount;
reporter.incrCounter(Counter.BYTESCOPIED, cbcopied);
reporter.incrCounter(Counter.COPY, 1);
updateStatus(reporter);
}
/**
* Return true if dst should be replaced by src and the update flag is set. Right now, this
* merely checks that the src and dst len are not equal. This should be improved on once
* modification times, CRCs, etc. can be meaningful in this context.
*
* @throws IOException
*/
private boolean needsUpdate(FileStatus srcstatus, FileSystem dstfs, Path dstpath)
throws IOException {
return update && !sameFile(srcstatus.getPath().getFileSystem(job), srcstatus, dstfs, dstpath);
}
/**
* 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;
}
public void write(DataOutput out) throws IOException {
input.write(out);
Text.writeString(out, output);
}
/** * Compare the checksums of the hdfs file as well as the local copied file. * * @author [email protected] * @date Fri Jan 27 06:06:00 2012 */ boolean compareChecksums(FileSystem fs, Path p, String sFsPath) { try { // get hdfs file info FileStatus stat = fs.getFileStatus(p); // get HDFS checksum FileChecksum ck = fs.getFileChecksum(p); String sCk, sCkShort; if (ck == null) { sCk = sCkShort = "<null>"; } else { sCk = ck.toString(); sCkShort = sCk.replaceAll("^.*:", ""); } // System.out.println(p.toUri().getPath() + " len=" + stat.getLen() // + " " + stat.getOwner() + "/" + stat.getGroup() // + " checksum=" + sCk); // find the local file File fLocal = new File(sFsPath); if (!fLocal.exists()) { System.out.println("CHECKSUM-ERROR: file does not exist: " + sFsPath); return false; } if (!fLocal.isFile()) { System.out.println("CHECKSUM-ERROR: path is not a file: " + sFsPath); return false; } if (stat.getLen() != fLocal.length()) { System.out.println( "CHECKSUM-ERROR: length mismatch: " + sFsPath + " hdfslen=" + stat.getLen() + " fslen=" + fLocal.length()); return false; } // get local fs checksum FileChecksum ckLocal = getLocalFileChecksum(sFsPath); if (ckLocal == null) { System.out.println("ERROR Failed to get checksum for local file " + sFsPath); return false; } // compare checksums as a string, after stripping the // algorithm name from the beginning String sCkLocal = ckLocal.toString(); String sCkLocalShort = sCkLocal.replaceAll("^.*:", ""); if (false == sCkShort.equals(sCkLocalShort)) { System.out.println( "CHECKSUM-ERROR: checksum mismatch: " + sFsPath + "\nhdfs = " + sCk + "\nlocal= " + sCkLocal); return false; } return true; } catch (IOException e) { System.out.println("CHECKSUM-ERROR: " + sFsPath + " exception " + e.toString()); } return false; }
public int run(String[] args) throws IOException, ClassNotFoundException, InterruptedException {
if (args.length != 2) {
System.out.println("Usage: FeatureMatching ID <inputName.jpeg/inputName.jpg>");
System.exit(1);
}
SimpleDateFormat sdf = new SimpleDateFormat("", Locale.US);
sdf.applyPattern("yyyy-MM-dd_HH-mm-ss");
String time = sdf.format(new Date());
Job job = Job.getInstance();
ID = "/" + args[0];
String filename = args[1];
filename = filename.toLowerCase();
System.out.println("current filename:" + filename);
// Detect illegal username (if the username dir doesn't exist)
File userPath = new File(LOCAL_USER_DIR + ID);
if (!userPath.exists()) {
System.out.println("Unauthorized username!!!\nExiting......");
System.exit(1);
}
// Preprocess the input image.jpg from local dir: /local.../user/ID/input/image.jpg
// Save the features to local dir: hdfs://.../user/ID/input/image.jpg
extractQueryFeatures2HDFS(filename, job);
// Add the feature file to the hdfs cache
String featureFileName = filename.substring(0, filename.lastIndexOf(".")) + ".json";
// job.addCacheFile(new Path(HDFS_HOME + USER + ID + INPUT + "/" +
// featureFileName).toUri());
job.getConfiguration()
.set("featureFilePath", HDFS_HOME + USER + ID + INPUT + "/" + featureFileName);
// Check the file type. Only support jpeg/jpg type images
String type = filename.substring(args[1].lastIndexOf("."));
if (!(type.equals(".jpg") || type.equals(".jpeg"))) {
System.out.println("Image type not supported!!!\nExiting");
System.exit(1);
}
// Input: hdfs://.../features/
// The feature dir is a location of all features extracted from the database
String inputPathStr = HDFS_HOME + FEATURES;
// Output: hdfs://.../user/ID/output/
String outputPathStr = HDFS_HOME + USER + ID + OUTPUT + "/" + time;
job.setInputFormatClass(KeyValueTextInputFormat.class);
// job.setOutputFormatClass(TextOutputFormat.class);
// Get the lists of all feature files: /.../features/data/part-*
FileSystem fs = FileSystem.get(job.getConfiguration());
FileStatus[] statuses = fs.listStatus(new Path(inputPathStr));
StringBuffer sb = new StringBuffer();
for (FileStatus fileStatus : statuses) {
sb.append(fileStatus.getPath() + ",");
}
sb.deleteCharAt(sb.lastIndexOf(","));
job.setJarByClass(FeatureMatching.class);
job.setMapperClass(FeatureMatchMapper.class);
job.setReducerClass(FeatureMatchReducer.class);
// only need one reducer to collect the result
job.setNumReduceTasks(1);
job.setMapOutputKeyClass(IntWritable.class);
job.setMapOutputValueClass(Text.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(DoubleWritable.class);
// Input a directory, so need the recursive input
FileInputFormat.setInputDirRecursive(job, true);
// Set the PathFilter, to omit _SUCCESS files
// (This is not working correctly, as the PathFilter class is an interface rather than a class.
// But the 2nd arg asks me to extend the PathFilter)
// FileInputFormat.setInputPathFilter(job, MyPathFilter.class);
//
// FileInputFormat.setInputPaths(job, new Path(inputPathStr));
FileInputFormat.setInputPaths(job, sb.toString());
FileOutputFormat.setOutputPath(job, new Path(outputPathStr));
boolean success = job.waitForCompletion(true);
return success ? 0 : 1;
}
@Override
public final void run() {
try {
// before preparing the job localize
// all the archives
TaskAttemptID taskid = t.getTaskID();
LocalDirAllocator lDirAlloc = new LocalDirAllocator("mapred.local.dir");
File jobCacheDir = null;
if (conf.getJar() != null) {
jobCacheDir = new File(new Path(conf.getJar()).getParent().toString());
}
File workDir =
new File(
lDirAlloc
.getLocalPathToRead(
TaskTracker.getJobCacheSubdir()
+ Path.SEPARATOR
+ t.getJobID()
+ Path.SEPARATOR
+ t.getTaskID()
+ Path.SEPARATOR
+ MRConstants.WORKDIR,
conf)
.toString());
URI[] archives = DistributedCache.getCacheArchives(conf);
URI[] files = DistributedCache.getCacheFiles(conf);
FileStatus fileStatus;
FileSystem fileSystem;
Path localPath;
String baseDir;
if ((archives != null) || (files != null)) {
if (archives != null) {
String[] archivesTimestamps = DistributedCache.getArchiveTimestamps(conf);
Path[] p = new Path[archives.length];
for (int i = 0; i < archives.length; i++) {
fileSystem = FileSystem.get(archives[i], conf);
fileStatus = fileSystem.getFileStatus(new Path(archives[i].getPath()));
String cacheId = DistributedCache.makeRelative(archives[i], conf);
String cachePath = TaskTracker.getCacheSubdir() + Path.SEPARATOR + cacheId;
if (lDirAlloc.ifExists(cachePath, conf)) {
localPath = lDirAlloc.getLocalPathToRead(cachePath, conf);
} else {
localPath = lDirAlloc.getLocalPathForWrite(cachePath, fileStatus.getLen(), conf);
}
baseDir = localPath.toString().replace(cacheId, "");
p[i] =
DistributedCache.getLocalCache(
archives[i],
conf,
new Path(baseDir),
fileStatus,
true,
Long.parseLong(archivesTimestamps[i]),
new Path(workDir.getAbsolutePath()),
false);
}
DistributedCache.setLocalArchives(conf, stringifyPathArray(p));
}
if ((files != null)) {
String[] fileTimestamps = DistributedCache.getFileTimestamps(conf);
Path[] p = new Path[files.length];
for (int i = 0; i < files.length; i++) {
fileSystem = FileSystem.get(files[i], conf);
fileStatus = fileSystem.getFileStatus(new Path(files[i].getPath()));
String cacheId = DistributedCache.makeRelative(files[i], conf);
String cachePath = TaskTracker.getCacheSubdir() + Path.SEPARATOR + cacheId;
if (lDirAlloc.ifExists(cachePath, conf)) {
localPath = lDirAlloc.getLocalPathToRead(cachePath, conf);
} else {
localPath = lDirAlloc.getLocalPathForWrite(cachePath, fileStatus.getLen(), conf);
}
baseDir = localPath.toString().replace(cacheId, "");
p[i] =
DistributedCache.getLocalCache(
files[i],
conf,
new Path(baseDir),
fileStatus,
false,
Long.parseLong(fileTimestamps[i]),
new Path(workDir.getAbsolutePath()),
false);
}
DistributedCache.setLocalFiles(conf, stringifyPathArray(p));
}
Path localTaskFile = new Path(t.getJobFile());
FileSystem localFs = FileSystem.getLocal(conf);
localFs.delete(localTaskFile, true);
OutputStream out = localFs.create(localTaskFile);
try {
conf.writeXml(out);
} finally {
out.close();
}
}
if (!prepare()) {
return;
}
String sep = System.getProperty("path.separator");
StringBuffer classPath = new StringBuffer();
// start with same classpath as parent process
classPath.append(System.getProperty("java.class.path"));
classPath.append(sep);
if (!workDir.mkdirs()) {
if (!workDir.isDirectory()) {
LOG.fatal("Mkdirs failed to create " + workDir.toString());
}
}
String jar = conf.getJar();
if (jar != null) {
// if jar exists, it into workDir
File[] libs = new File(jobCacheDir, "lib").listFiles();
if (libs != null) {
for (int i = 0; i < libs.length; i++) {
classPath.append(sep); // add libs from jar to classpath
classPath.append(libs[i]);
}
}
classPath.append(sep);
classPath.append(new File(jobCacheDir, "classes"));
classPath.append(sep);
classPath.append(jobCacheDir);
}
// include the user specified classpath
// archive paths
Path[] archiveClasspaths = DistributedCache.getArchiveClassPaths(conf);
if (archiveClasspaths != null && archives != null) {
Path[] localArchives = DistributedCache.getLocalCacheArchives(conf);
if (localArchives != null) {
for (int i = 0; i < archives.length; i++) {
for (int j = 0; j < archiveClasspaths.length; j++) {
if (archives[i].getPath().equals(archiveClasspaths[j].toString())) {
classPath.append(sep);
classPath.append(localArchives[i].toString());
}
}
}
}
}
// file paths
Path[] fileClasspaths = DistributedCache.getFileClassPaths(conf);
if (fileClasspaths != null && files != null) {
Path[] localFiles = DistributedCache.getLocalCacheFiles(conf);
if (localFiles != null) {
for (int i = 0; i < files.length; i++) {
for (int j = 0; j < fileClasspaths.length; j++) {
if (files[i].getPath().equals(fileClasspaths[j].toString())) {
classPath.append(sep);
classPath.append(localFiles[i].toString());
}
}
}
}
}
classPath.append(sep);
classPath.append(workDir);
// Build exec child jmv args.
Vector<String> vargs = new Vector<String>(8);
File jvm = // use same jvm as parent
new File(new File(System.getProperty("java.home"), "bin"), "java");
vargs.add(jvm.toString());
// Add child (task) java-vm options.
//
// The following symbols if present in mapred.child.java.opts value are
// replaced:
// + @taskid@ is interpolated with value of TaskID.
// Other occurrences of @ will not be altered.
//
// Example with multiple arguments and substitutions, showing
// jvm GC logging, and start of a passwordless JVM JMX agent so can
// connect with jconsole and the likes to watch child memory, threads
// and get thread dumps.
//
// <property>
// <name>mapred.child.java.opts</name>
// <value>-verbose:gc -Xloggc:/tmp/@[email protected] \
// -Dcom.sun.management.jmxremote.authenticate=false \
// -Dcom.sun.management.jmxremote.ssl=false \
// </value>
// </property>
//
String javaOpts = conf.get("mapred.child.java.opts", "-Xmx200m");
javaOpts = javaOpts.replace("@taskid@", taskid.toString());
String[] javaOptsSplit = javaOpts.split(" ");
// Add java.library.path; necessary for loading native libraries.
//
// 1. To support native-hadoop library i.e. libhadoop.so, we add the
// parent processes' java.library.path to the child.
// 2. We also add the 'cwd' of the task to it's java.library.path to help
// users distribute native libraries via the DistributedCache.
// 3. The user can also specify extra paths to be added to the
// java.library.path via mapred.child.java.opts.
//
String libraryPath = System.getProperty("java.library.path");
if (libraryPath == null) {
libraryPath = workDir.getAbsolutePath();
} else {
libraryPath += sep + workDir;
}
boolean hasUserLDPath = false;
for (int i = 0; i < javaOptsSplit.length; i++) {
if (javaOptsSplit[i].startsWith("-Djava.library.path=")) {
javaOptsSplit[i] += sep + libraryPath;
hasUserLDPath = true;
break;
}
}
if (!hasUserLDPath) {
vargs.add("-Djava.library.path=" + libraryPath);
}
for (int i = 0; i < javaOptsSplit.length; i++) {
vargs.add(javaOptsSplit[i]);
}
// add java.io.tmpdir given by mapred.child.tmp
String tmp = conf.get("mapred.child.tmp", "./tmp");
Path tmpDir = new Path(tmp);
// if temp directory path is not absolute
// prepend it with workDir.
if (!tmpDir.isAbsolute()) {
tmpDir = new Path(workDir.toString(), tmp);
}
FileSystem localFs = FileSystem.getLocal(conf);
if (!localFs.mkdirs(tmpDir) && !localFs.getFileStatus(tmpDir).isDir()) {
throw new IOException("Mkdirs failed to create " + tmpDir.toString());
}
vargs.add("-Djava.io.tmpdir=" + tmpDir.toString());
// Add classpath.
vargs.add("-classpath");
vargs.add(classPath.toString());
// Setup the log4j prop
long logSize = TaskLog.getTaskLogLength(conf);
vargs.add(
"-Dhadoop.log.dir=" + new File(System.getProperty("hadoop.log.dir")).getAbsolutePath());
vargs.add("-Dhadoop.root.logger=INFO,TLA");
vargs.add("-Dhadoop.tasklog.taskid=" + taskid);
vargs.add("-Dhadoop.tasklog.totalLogFileSize=" + logSize);
if (conf.getProfileEnabled()) {
if (conf.getProfileTaskRange(t.isMapTask()).isIncluded(t.getPartition())) {
File prof = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.PROFILE);
vargs.add(String.format(conf.getProfileParams(), prof.toString()));
}
}
// Add main class and its arguments
vargs.add(Child.class.getName()); // main of Child
// pass umbilical address
InetSocketAddress address = tracker.getTaskTrackerReportAddress();
vargs.add(address.getAddress().getHostAddress());
vargs.add(Integer.toString(address.getPort()));
vargs.add(taskid.toString()); // pass task identifier
String pidFile = null;
if (tracker.isTaskMemoryManagerEnabled()) {
pidFile =
lDirAlloc
.getLocalPathForWrite(
(TaskTracker.getPidFilesSubdir() + Path.SEPARATOR + taskid), this.conf)
.toString();
}
// set memory limit using ulimit if feasible and necessary ...
String[] ulimitCmd = Shell.getUlimitMemoryCommand(conf);
List<String> setup = null;
if (ulimitCmd != null) {
setup = new ArrayList<String>();
for (String arg : ulimitCmd) {
setup.add(arg);
}
}
// Set up the redirection of the task's stdout and stderr streams
File stdout = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDOUT);
File stderr = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDERR);
stdout.getParentFile().mkdirs();
tracker.getTaskTrackerInstrumentation().reportTaskLaunch(taskid, stdout, stderr);
Map<String, String> env = new HashMap<String, String>();
StringBuffer ldLibraryPath = new StringBuffer();
ldLibraryPath.append(workDir.toString());
String oldLdLibraryPath = null;
oldLdLibraryPath = System.getenv("LD_LIBRARY_PATH");
if (oldLdLibraryPath != null) {
ldLibraryPath.append(sep);
ldLibraryPath.append(oldLdLibraryPath);
}
env.put("LD_LIBRARY_PATH", ldLibraryPath.toString());
jvmManager.launchJvm(
this,
jvmManager.constructJvmEnv(
setup, vargs, stdout, stderr, logSize, workDir, env, pidFile, conf));
synchronized (lock) {
while (!done) {
lock.wait();
}
}
tracker.getTaskTrackerInstrumentation().reportTaskEnd(t.getTaskID());
if (exitCodeSet) {
if (!killed && exitCode != 0) {
if (exitCode == 65) {
tracker.getTaskTrackerInstrumentation().taskFailedPing(t.getTaskID());
}
throw new IOException("Task process exit with nonzero status of " + exitCode + ".");
}
}
} catch (FSError e) {
LOG.fatal("FSError", e);
try {
tracker.fsError(t.getTaskID(), e.getMessage());
} catch (IOException ie) {
LOG.fatal(t.getTaskID() + " reporting FSError", ie);
}
} catch (Throwable throwable) {
LOG.warn(t.getTaskID() + " Child Error", throwable);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
throwable.printStackTrace(new PrintStream(baos));
try {
tracker.reportDiagnosticInfo(t.getTaskID(), baos.toString());
} catch (IOException e) {
LOG.warn(t.getTaskID() + " Reporting Diagnostics", e);
}
} finally {
try {
URI[] archives = DistributedCache.getCacheArchives(conf);
URI[] files = DistributedCache.getCacheFiles(conf);
if (archives != null) {
for (int i = 0; i < archives.length; i++) {
DistributedCache.releaseCache(archives[i], conf);
}
}
if (files != null) {
for (int i = 0; i < files.length; i++) {
DistributedCache.releaseCache(files[i], conf);
}
}
} catch (IOException ie) {
LOG.warn("Error releasing caches : Cache files might not have been cleaned up");
}
tracker.reportTaskFinished(t.getTaskID(), false);
if (t.isMapTask()) {
tracker.addFreeMapSlot();
} else {
tracker.addFreeReduceSlot();
}
}
}
/** Delete the dst files/dirs which do not exist in src */
private static void deleteNonexisting(
FileSystem dstfs,
FileStatus dstroot,
Path dstsorted,
FileSystem jobfs,
Path jobdir,
JobConf jobconf,
Configuration conf)
throws IOException {
if (!dstroot.isDir()) {
throw new IOException(
"dst must be a directory when option "
+ Options.DELETE.cmd
+ " is set, but dst (= "
+ dstroot.getPath()
+ ") is not a directory.");
}
// write dst lsr results
final Path dstlsr = new Path(jobdir, "_distcp_dst_lsr");
final SequenceFile.Writer writer =
SequenceFile.createWriter(
jobfs,
jobconf,
dstlsr,
Text.class,
FileStatus.class,
SequenceFile.CompressionType.NONE);
try {
// do lsr to get all file statuses in dstroot
final Stack<FileStatus> lsrstack = new Stack<FileStatus>();
for (lsrstack.push(dstroot); !lsrstack.isEmpty(); ) {
final FileStatus status = lsrstack.pop();
if (status.isDir()) {
for (FileStatus child : dstfs.listStatus(status.getPath())) {
String relative = makeRelative(dstroot.getPath(), child.getPath());
writer.append(new Text(relative), child);
lsrstack.push(child);
}
}
}
} finally {
checkAndClose(writer);
}
// sort lsr results
final Path sortedlsr = new Path(jobdir, "_distcp_dst_lsr_sorted");
SequenceFile.Sorter sorter =
new SequenceFile.Sorter(
jobfs, new Text.Comparator(), Text.class, FileStatus.class, jobconf);
sorter.sort(dstlsr, sortedlsr);
// compare lsr list and dst list
SequenceFile.Reader lsrin = null;
SequenceFile.Reader dstin = null;
try {
lsrin = new SequenceFile.Reader(jobfs, sortedlsr, jobconf);
dstin = new SequenceFile.Reader(jobfs, dstsorted, jobconf);
// compare sorted lsr list and sorted dst list
final Text lsrpath = new Text();
final FileStatus lsrstatus = new FileStatus();
final Text dstpath = new Text();
final Text dstfrom = new Text();
final FsShell shell = new FsShell(conf);
final String[] shellargs = {"-rmr", null};
boolean hasnext = dstin.next(dstpath, dstfrom);
for (; lsrin.next(lsrpath, lsrstatus); ) {
int dst_cmp_lsr = dstpath.compareTo(lsrpath);
for (; hasnext && dst_cmp_lsr < 0; ) {
hasnext = dstin.next(dstpath, dstfrom);
dst_cmp_lsr = dstpath.compareTo(lsrpath);
}
if (dst_cmp_lsr == 0) {
// lsrpath exists in dst, skip it
hasnext = dstin.next(dstpath, dstfrom);
} else {
// lsrpath does not exist, delete it
String s = new Path(dstroot.getPath(), lsrpath.toString()).toString();
if (shellargs[1] == null || !isAncestorPath(shellargs[1], s)) {
shellargs[1] = s;
int r = 0;
try {
r = shell.run(shellargs);
} catch (Exception e) {
throw new IOException("Exception from shell.", e);
}
if (r != 0) {
throw new IOException(
"\"" + shellargs[0] + " " + shellargs[1] + "\" returns non-zero value " + r);
}
}
}
}
} finally {
checkAndClose(lsrin);
checkAndClose(dstin);
}
}
public void readFields(DataInput in) throws IOException {
input.readFields(in);
output = Text.readString(in);
}
@Test
public void testListStatus() throws Exception {
final String hPrefix = "test/hadoop";
final String[] dirs = {
hPrefix + "/a",
hPrefix + "/b",
hPrefix + "/c",
hPrefix + "/1",
hPrefix + "/#@#@",
hPrefix + "/&*#$#$@234"
};
ArrayList<Path> testDirs = new ArrayList<Path>();
for (String d : dirs) {
testDirs.add(qualifiedPath(d, fc2));
}
Assert.assertFalse(exists(fc1, testDirs.get(0)));
for (Path path : testDirs) {
fc1.mkdir(path, FsPermission.getDefault(), true);
}
// test listStatus that returns an array of FileStatus
FileStatus[] paths = fc1.util().listStatus(qualifiedPath("test", fc1));
Assert.assertEquals(1, paths.length);
Assert.assertEquals(qualifiedPath(hPrefix, fc1), paths[0].getPath());
paths = fc1.util().listStatus(qualifiedPath(hPrefix, fc1));
Assert.assertEquals(6, paths.length);
for (int i = 0; i < dirs.length; i++) {
boolean found = false;
for (int j = 0; j < paths.length; j++) {
if (qualifiedPath(dirs[i], fc1).equals(paths[j].getPath())) {
found = true;
}
}
Assert.assertTrue(dirs[i] + " not found", found);
}
paths = fc1.util().listStatus(qualifiedPath(dirs[0], fc1));
Assert.assertEquals(0, paths.length);
// test listStatus that returns an iterator of FileStatus
RemoteIterator<FileStatus> pathsItor = fc1.listStatus(qualifiedPath("test", fc1));
Assert.assertEquals(qualifiedPath(hPrefix, fc1), pathsItor.next().getPath());
Assert.assertFalse(pathsItor.hasNext());
pathsItor = fc1.listStatus(qualifiedPath(hPrefix, fc1));
int dirLen = 0;
for (; pathsItor.hasNext(); dirLen++) {
boolean found = false;
FileStatus stat = pathsItor.next();
for (int j = 0; j < dirs.length; j++) {
if (qualifiedPath(dirs[j], fc1).equals(stat.getPath())) {
found = true;
break;
}
}
Assert.assertTrue(stat.getPath() + " not found", found);
}
Assert.assertEquals(6, dirLen);
pathsItor = fc1.listStatus(qualifiedPath(dirs[0], fc1));
Assert.assertFalse(pathsItor.hasNext());
}
/**
* Method to go though the HDFS filesystem in a DFS to find all files
*
* <p>fs:FileSystem object from HDFS minDate: Oldest date for files to be backed up maxDate:Newest
* date for files to be backed up p:Path in HDFS to look for files pathList:Will be filled with
* all files in p hmTimestamps: hashmap of timestamps for later sorting
*/
public void checkDir(
FileSystem fs,
long minDate,
long maxDate,
Path p,
ArrayList<Path> pathList,
HashMap<Path, Long> hmTimestamps) {
long tmpDate;
FileStatus[] fStat;
try {
String sPath = p.toUri().getPath();
// If this is a directory
if (fs.getFileStatus(p).isDir()) {
// ignore certain directories
if ("dfstmp".equals(p.getName())
|| "tmp".equals(p.getName())
|| "jobtracker".equals(p.getName())
|| sPath.startsWith("/mapred")
|| "ops".equals(p.getName())
|| p.getName().startsWith("_distcp_logs")) {
return;
}
// dump the mkdir and chmod commands for this
// directory -- skip root directory only
{
FileStatus stat = fs.getFileStatus(p);
if (!sPath.equals("/")) {
m_wrMkdirs.println("hadoop fs -mkdir " + sPath);
}
m_wrChmods.println(
"hadoop fs -chown " + stat.getOwner() + ":" + stat.getGroup() + " " + sPath);
Short sh = new Short(stat.getPermission().toShort());
m_wrChmods.println(
"hadoop fs -chmod " + Long.toOctalString(sh.longValue()) + " " + sPath);
}
fStat = fs.listStatus(p);
// Do a recursive call to all elements
for (int i = 0; i < fStat.length; i++) {
checkDir(fs, minDate, maxDate, fStat[i].getPath(), pathList, hmTimestamps);
}
} else {
// If not a directory then we've found a file
// ignore crc files
if (p.getName().endsWith(".crc")) {
return;
}
// ignore other files
if (sPath.startsWith("/user/oozie/etl/workflows/")) {
return;
}
// try to get the table name from the path. There are
// various types of tables, from those replicated from
// another database to regular hive tables to
// partitioned hive tables. We use table names to
// both exclude some from the backup, and for the rest
// to dump out the schema and partition name.
if (m_ignoreTables != null && m_ignoreTables.doIgnoreFile(sPath)) {
m_nIgnoredTables++;
if (m_nIgnoredTables < 5) {
System.out.println("Skipping ignore-table file: " + sPath);
} else if (m_nIgnoredTables == 5) {
System.out.println("(...not showing other skipped tables...)");
}
return;
}
FileStatus stat = fs.getFileStatus(p);
tmpDate = stat.getModificationTime() / 1000;
// store the chmods/chowns for all files
m_wrChmods.println(
"hadoop fs -chown " + stat.getOwner() + ":" + stat.getGroup() + " " + sPath);
m_wrChmods.println("hadoop fs -chmod " + stat.getPermission().toShort() + " " + sPath);
// check dates. is it too young?
if (tmpDate < minDate) {
return;
}
// is the file too recent?
if (tmpDate > maxDate) {
// System.out.println("file too recent: " + sPath);
return;
}
// file timestamp is ok
pathList.add(p);
hmTimestamps.put(p, new Long(tmpDate));
// store info about total bytes neeed to backup
m_nTotalBytes += fs.getContentSummary(p).getLength();
}
} catch (IOException e) {
System.err.println("ERROR: could not open " + p + ": " + e);
// System.exit(1) ;
}
}