public static void main(String[] args) throws IOException {
Path baseDir = null;
String localPath = null;
String preservePath = null;
String sIgnoreTablesFilename = null;
String sNoPreserveFilename = null;
String sDateString = null;
long size = 0;
// UNIX dates for right now
long now = new java.util.Date().getTime() / 1000;
long maxDate = now;
for (int i = 0; i < args.length; i++) {
if (args[i].equals("--hdfs-path")) {
baseDir = new Path(args[++i]);
continue;
}
if (args[i].equals("--local-path")) {
localPath = args[++i];
continue;
}
if (args[i].equals("--preserve-path")) {
preservePath = args[++i];
continue;
}
if (args[i].equals("--no-preserve")) {
sNoPreserveFilename = args[++i];
continue;
}
if (args[i].equals("--ignore-tables")) {
sIgnoreTablesFilename = args[++i];
continue;
}
if (args[i].equals("--sleep")) {
try {
m_nSleepSeconds = Integer.parseInt(args[++i]);
} catch (Exception e) {
System.err.println("ERROR: " + e.toString() + "\n");
usage();
}
continue;
}
if (args[i].equals("--dry-run")) {
m_bDryRun = true;
continue;
}
if (args[i].equals("--date")) {
sDateString = args[++i];
continue;
}
if (args[i].equals("--max-date")) {
maxDate = Long.parseLong(args[++i]);
continue;
}
if (args[i].equals("--max-bytes")) {
size = Long.parseLong(args[++i]);
continue;
}
System.err.println("ERROR: unknown arg " + args[i]);
usage();
}
if (baseDir == null || localPath == null || preservePath == null || sDateString == null) {
usage();
}
long minDate;
if ("yesterday".equals(sDateString)) {
// figure out yesterday's dates
Calendar cal = Calendar.getInstance();
cal.roll(Calendar.DAY_OF_YEAR, -1);
// yesterday midnight
cal.set(Calendar.HOUR_OF_DAY, 0);
cal.set(Calendar.MINUTE, 0);
cal.set(Calendar.SECOND, 0);
cal.set(Calendar.MILLISECOND, 0);
minDate = cal.getTimeInMillis() / 1000;
// yesterday end of day
cal.set(Calendar.HOUR_OF_DAY, 23);
cal.set(Calendar.MINUTE, 59);
cal.set(Calendar.SECOND, 59);
cal.set(Calendar.MILLISECOND, 999);
maxDate = cal.getTimeInMillis() / 1000;
} else if ("last-week".equals(sDateString)) {
minDate = maxDate - (7 * 24 * 60 * 60);
} else if ("last-day".equals(sDateString)) {
minDate = maxDate - (24 * 60 * 60);
} else {
// UNIX date since epoch of last backup
minDate = Long.parseLong(sDateString);
}
long tmpDate = 0;
BackupHdfs bak = new BackupHdfs();
// initialize the list of tables to ignore
if (sIgnoreTablesFilename != null) {
bak.initializeTablesToIgnore(sIgnoreTablesFilename);
}
// initialize list of files to not preserve
if (sNoPreserveFilename != null) {
bak.initializeNoPreserve(sNoPreserveFilename);
}
ArrayList<Path> pathList = new ArrayList<Path>(2000);
HashMap<Path, Long> hmTimestamps = new HashMap<Path, Long>();
Configuration conf = new Configuration();
FileSystem fs = FileSystem.get(conf);
// If the HDFS path is a dir continue
if (fs.getFileStatus(baseDir).isDir()) {
Calendar cal = Calendar.getInstance();
System.err.println("");
cal.setTimeInMillis(minDate * 1000);
System.err.println("min date = " + cal.getTime().toString());
cal.setTimeInMillis(maxDate * 1000);
System.err.println("max date = " + cal.getTime().toString());
System.err.println("");
System.err.println("Searching filesystem: " + baseDir.toUri().getPath());
bak.checkDir(fs, minDate, maxDate, baseDir, pathList, hmTimestamps);
System.err.println("");
System.err.println("Skipped " + m_nIgnoredTables + " files due to ignored tables");
System.err.println("");
System.err.println("Number of files to backup = " + pathList.size());
System.err.println("Total bytes to backup = " + prettyPrintBytes(m_nTotalBytes));
System.err.println("");
System.err.println("sorting list of files...");
Collections.sort(pathList, new DateComparator(hmTimestamps));
System.err.println("done");
System.err.println("");
System.err.println("starting backup...");
tmpDate = bak.backupFiles(localPath, preservePath, fs, pathList, size);
bak.closeFiles();
System.err.println("");
System.err.println("backup completed...");
}
if (tmpDate == 0) {
// If not size limit reached print out date for right now
System.out.println(maxDate);
} else {
// Print out date for last file backed up
System.err.println("Size limit reached.");
System.out.println(tmpDate);
}
System.exit(0);
}
public H2OHdfsInputStream(Path p, long offset, ProgressMonitor pmon) throws IOException {
super(offset, pmon);
_path = p;
_fs = FileSystem.get(p.toUri(), CONF);
setExpectedSz(_fs.getFileStatus(p).getLen());
open();
}
public static void addFolder2(Path p, ArrayList<String> keys, ArrayList<String> failed)
throws IOException {
FileSystem fs = FileSystem.get(p.toUri(), PersistHdfs.CONF);
if (!fs.exists(p)) {
failed.add("Path does not exist: '" + p.toString() + "'");
return;
}
addFolder2(fs, p, keys, failed);
}
// Loading/Writing ice to HDFS
PersistHdfs(URI uri) {
try {
_iceRoot = new Path(uri + "/ice" + H2O.SELF_ADDRESS.getHostAddress() + "-" + H2O.API_PORT);
// Make the directory as-needed
FileSystem fs = FileSystem.get(_iceRoot.toUri(), CONF);
fs.mkdirs(_iceRoot);
} catch (Exception e) {
throw Log.errRTExcept(e);
}
}
/**
* Add a file path to the current set of classpath entries. It adds the file to cache as well.
* Intended to be used by user code.
*
* @param file Path of the file to be added
* @param conf Configuration that contains the classpath setting
* @param fs FileSystem with respect to which {@code archivefile} should be interpreted.
*/
public static void addFileToClassPath(Path file, Configuration conf, FileSystem fs)
throws IOException {
String filepath = file.toUri().getPath();
String classpath = conf.get("mapred.job.classpath.files");
conf.set(
"mapred.job.classpath.files",
classpath == null ? filepath : classpath + System.getProperty("path.separator") + filepath);
URI uri = fs.makeQualified(file).toUri();
addCacheFile(uri, conf);
}
public static void addFolder(Path p, JsonArray succeeded, JsonArray failed) throws IOException {
FileSystem fs = FileSystem.get(p.toUri(), PersistHdfs.CONF);
if (!fs.exists(p)) {
JsonObject o = new JsonObject();
o.addProperty(Constants.FILE, p.toString());
o.addProperty(Constants.ERROR, "Path does not exist!");
failed.add(o);
return;
}
addFolder(fs, p, succeeded, failed);
}
/**
* Make a path relative with respect to a root path. absPath is always assumed to descend from
* root. Otherwise returned path is null.
*/
static String makeRelative(Path root, Path absPath) {
if (!absPath.isAbsolute()) {
throw new IllegalArgumentException("!absPath.isAbsolute(), absPath=" + absPath);
}
String p = absPath.toUri().getPath();
StringTokenizer pathTokens = new StringTokenizer(p, "/");
for (StringTokenizer rootTokens = new StringTokenizer(root.toUri().getPath(), "/");
rootTokens.hasMoreTokens(); ) {
if (!rootTokens.nextToken().equals(pathTokens.nextToken())) {
return null;
}
}
StringBuilder sb = new StringBuilder();
for (; pathTokens.hasMoreTokens(); ) {
sb.append(pathTokens.nextToken());
if (pathTokens.hasMoreTokens()) {
sb.append(Path.SEPARATOR);
}
}
return sb.length() == 0 ? "." : sb.toString();
}
/**
* Add an archive path to the current set of classpath entries. It adds the archive to cache as
* well. Intended to be used by user code.
*
* @param archive Path of the archive to be added
* @param conf Configuration that contains the classpath setting
* @param fs FileSystem with respect to which {@code archive} should be interpreted.
*/
public static void addArchiveToClassPath(Path archive, Configuration conf, FileSystem fs)
throws IOException {
String archivepath = archive.toUri().getPath();
String classpath = conf.get("mapred.job.classpath.archives");
conf.set(
"mapred.job.classpath.archives",
classpath == null
? archivepath
: classpath + System.getProperty("path.separator") + archivepath);
URI uri = fs.makeQualified(archive).toUri();
addCacheArchive(uri, conf);
}
private static URI addArchiveToClassPathHelper(Path archive, Configuration conf)
throws IOException {
String classpath = conf.get("mapred.job.classpath.archives");
// the scheme/authority use ':' as separator. put the unqualified path in classpath
String archivePath = archive.toUri().getPath();
conf.set(
"mapred.job.classpath.archives",
classpath == null
? archivePath
: classpath + System.getProperty("path.separator") + archivePath);
return archive.makeQualified(archive.getFileSystem(conf)).toUri();
}
/**
* Delete a local path with asyncDiskService if available, or otherwise synchronously with local
* file system.
*/
private static void deleteLocalPath(
MRAsyncDiskService asyncDiskService, LocalFileSystem fs, Path path) throws IOException {
boolean deleted = false;
if (asyncDiskService != null) {
// Try to delete using asyncDiskService
String localPathToDelete = path.toUri().getPath();
deleted = asyncDiskService.moveAndDeleteAbsolutePath(localPathToDelete);
if (!deleted) {
LOG.warn(
"Cannot find DistributedCache path "
+ localPathToDelete
+ " on any of the asyncDiskService volumes!");
}
}
if (!deleted) {
// If no asyncDiskService, we will delete the files synchronously
fs.delete(path, true);
}
LOG.info("Deleted path " + path);
}
/**
* 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);
}
public static void addFolder(Path p, JsonArray succeeded, JsonArray failed) throws IOException {
FileSystem fs = FileSystem.get(p.toUri(), PersistHdfs.CONF);
addFolder(fs, p, succeeded, failed);
}
/**
* 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) ;
}
}
/**
* 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;
}