/**
* 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;
}
}
/** 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);
}
/**
* 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()]);
}
/**
* 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"));
}
/**
* Launch application for the dag represented by this client.
*
* @throws YarnException
* @throws IOException
*/
public void startApplication() throws YarnException, IOException {
Class<?>[] defaultClasses;
if (applicationType.equals(YARN_APPLICATION_TYPE)) {
// TODO restrict the security check to only check if security is enabled for webservices.
if (UserGroupInformation.isSecurityEnabled()) {
defaultClasses = DATATORRENT_SECURITY_CLASSES;
} else {
defaultClasses = DATATORRENT_CLASSES;
}
} else {
throw new IllegalStateException(applicationType + " is not a valid application type.");
}
LinkedHashSet<String> localJarFiles = findJars(dag, defaultClasses);
if (resources != null) {
localJarFiles.addAll(resources);
}
YarnClusterMetrics clusterMetrics = yarnClient.getYarnClusterMetrics();
LOG.info(
"Got Cluster metric info from ASM"
+ ", numNodeManagers="
+ clusterMetrics.getNumNodeManagers());
// GetClusterNodesRequest clusterNodesReq = Records.newRecord(GetClusterNodesRequest.class);
// GetClusterNodesResponse clusterNodesResp =
// rmClient.clientRM.getClusterNodes(clusterNodesReq);
// LOG.info("Got Cluster node info from ASM");
// for (NodeReport node : clusterNodesResp.getNodeReports()) {
// LOG.info("Got node report from ASM for"
// + ", nodeId=" + node.getNodeId()
// + ", nodeAddress" + node.getHttpAddress()
// + ", nodeRackName" + node.getRackName()
// + ", nodeNumContainers" + node.getNumContainers()
// + ", nodeHealthStatus" + node.getHealthReport());
// }
List<QueueUserACLInfo> listAclInfo = yarnClient.getQueueAclsInfo();
for (QueueUserACLInfo aclInfo : listAclInfo) {
for (QueueACL userAcl : aclInfo.getUserAcls()) {
LOG.info(
"User ACL Info for Queue"
+ ", queueName="
+ aclInfo.getQueueName()
+ ", userAcl="
+ userAcl.name());
}
}
// Get a new application id
YarnClientApplication newApp = yarnClient.createApplication();
appId = newApp.getNewApplicationResponse().getApplicationId();
// Dump out information about cluster capability as seen by the resource manager
int maxMem = newApp.getNewApplicationResponse().getMaximumResourceCapability().getMemory();
LOG.info("Max mem capabililty of resources in this cluster " + maxMem);
int amMemory = dag.getMasterMemoryMB();
if (amMemory > maxMem) {
LOG.info(
"AM memory specified above max threshold of cluster. Using max value."
+ ", specified="
+ amMemory
+ ", max="
+ maxMem);
amMemory = maxMem;
}
if (dag.getAttributes().get(LogicalPlan.APPLICATION_ID) == null) {
dag.setAttribute(LogicalPlan.APPLICATION_ID, appId.toString());
}
// Create launch context for app master
LOG.info("Setting up application submission context for ASM");
ApplicationSubmissionContext appContext = Records.newRecord(ApplicationSubmissionContext.class);
// set the application id
appContext.setApplicationId(appId);
// set the application name
appContext.setApplicationName(dag.getValue(LogicalPlan.APPLICATION_NAME));
appContext.setApplicationType(this.applicationType);
if (YARN_APPLICATION_TYPE.equals(this.applicationType)) {
// appContext.setMaxAppAttempts(1); // no retries until Stram is HA
}
// Set up the container launch context for the application master
ContainerLaunchContext amContainer = Records.newRecord(ContainerLaunchContext.class);
// Setup security tokens
// If security is enabled get ResourceManager and NameNode delegation tokens.
// Set these tokens on the container so that they are sent as part of application submission.
// This also sets them up for renewal by ResourceManager. The NameNode delegation rmToken
// is also used by ResourceManager to fetch the jars from HDFS and set them up for the
// application master launch.
if (UserGroupInformation.isSecurityEnabled()) {
Credentials credentials = new Credentials();
String tokenRenewer = conf.get(YarnConfiguration.RM_PRINCIPAL);
if (tokenRenewer == null || tokenRenewer.length() == 0) {
throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer");
}
// For now, only getting tokens for the default file-system.
FileSystem fs = StramClientUtils.newFileSystemInstance(conf);
try {
final Token<?> tokens[] = fs.addDelegationTokens(tokenRenewer, credentials);
if (tokens != null) {
for (Token<?> token : tokens) {
LOG.info("Got dt for " + fs.getUri() + "; " + token);
}
}
} finally {
fs.close();
}
addRMDelegationToken(tokenRenewer, credentials);
DataOutputBuffer dob = new DataOutputBuffer();
credentials.writeTokenStorageToStream(dob);
ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
amContainer.setTokens(fsTokens);
}
// set local resources for the application master
// local files or archives as needed
// In this scenario, the jar file for the application master is part of the local resources
Map<String, LocalResource> localResources = new HashMap<String, LocalResource>();
// copy required jar files to dfs, to be localized for containers
FileSystem fs = StramClientUtils.newFileSystemInstance(conf);
try {
Path appsBasePath =
new Path(StramClientUtils.getDTDFSRootDir(fs, conf), StramClientUtils.SUBDIR_APPS);
Path appPath = new Path(appsBasePath, appId.toString());
String libJarsCsv = copyFromLocal(fs, appPath, localJarFiles.toArray(new String[] {}));
LOG.info("libjars: {}", libJarsCsv);
dag.getAttributes().put(LogicalPlan.LIBRARY_JARS, libJarsCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.FILE, libJarsCsv, localResources, fs);
if (archives != null) {
String[] localFiles = archives.split(",");
String archivesCsv = copyFromLocal(fs, appPath, localFiles);
LOG.info("archives: {}", archivesCsv);
dag.getAttributes().put(LogicalPlan.ARCHIVES, archivesCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.ARCHIVE, archivesCsv, localResources, fs);
}
if (files != null) {
String[] localFiles = files.split(",");
String filesCsv = copyFromLocal(fs, appPath, localFiles);
LOG.info("files: {}", filesCsv);
dag.getAttributes().put(LogicalPlan.FILES, filesCsv);
LaunchContainerRunnable.addFilesToLocalResources(
LocalResourceType.FILE, filesCsv, localResources, fs);
}
dag.getAttributes().put(LogicalPlan.APPLICATION_PATH, appPath.toString());
if (dag.getAttributes().get(OperatorContext.STORAGE_AGENT) == null) {
/* which would be the most likely case */
Path checkpointPath = new Path(appPath, LogicalPlan.SUBDIR_CHECKPOINTS);
// use conf client side to pickup any proxy settings from dt-site.xml
dag.setAttribute(
OperatorContext.STORAGE_AGENT, new FSStorageAgent(checkpointPath.toString(), conf));
}
if (dag.getAttributes().get(LogicalPlan.CONTAINER_OPTS_CONFIGURATOR) == null) {
dag.setAttribute(
LogicalPlan.CONTAINER_OPTS_CONFIGURATOR, new BasicContainerOptConfigurator());
}
// Set the log4j properties if needed
if (!log4jPropFile.isEmpty()) {
Path log4jSrc = new Path(log4jPropFile);
Path log4jDst = new Path(appPath, "log4j.props");
fs.copyFromLocalFile(false, true, log4jSrc, log4jDst);
FileStatus log4jFileStatus = fs.getFileStatus(log4jDst);
LocalResource log4jRsrc = Records.newRecord(LocalResource.class);
log4jRsrc.setType(LocalResourceType.FILE);
log4jRsrc.setVisibility(LocalResourceVisibility.APPLICATION);
log4jRsrc.setResource(ConverterUtils.getYarnUrlFromURI(log4jDst.toUri()));
log4jRsrc.setTimestamp(log4jFileStatus.getModificationTime());
log4jRsrc.setSize(log4jFileStatus.getLen());
localResources.put("log4j.properties", log4jRsrc);
}
if (originalAppId != null) {
Path origAppPath = new Path(appsBasePath, this.originalAppId);
LOG.info("Restart from {}", origAppPath);
copyInitialState(origAppPath);
}
// push logical plan to DFS location
Path cfgDst = new Path(appPath, LogicalPlan.SER_FILE_NAME);
FSDataOutputStream outStream = fs.create(cfgDst, true);
LogicalPlan.write(this.dag, outStream);
outStream.close();
Path launchConfigDst = new Path(appPath, LogicalPlan.LAUNCH_CONFIG_FILE_NAME);
outStream = fs.create(launchConfigDst, true);
conf.writeXml(outStream);
outStream.close();
FileStatus topologyFileStatus = fs.getFileStatus(cfgDst);
LocalResource topologyRsrc = Records.newRecord(LocalResource.class);
topologyRsrc.setType(LocalResourceType.FILE);
topologyRsrc.setVisibility(LocalResourceVisibility.APPLICATION);
topologyRsrc.setResource(ConverterUtils.getYarnUrlFromURI(cfgDst.toUri()));
topologyRsrc.setTimestamp(topologyFileStatus.getModificationTime());
topologyRsrc.setSize(topologyFileStatus.getLen());
localResources.put(LogicalPlan.SER_FILE_NAME, topologyRsrc);
// Set local resource info into app master container launch context
amContainer.setLocalResources(localResources);
// Set the necessary security tokens as needed
// amContainer.setContainerTokens(containerToken);
// Set the env variables to be setup in the env where the application master will be run
LOG.info("Set the environment for the application master");
Map<String, String> env = new HashMap<String, String>();
// Add application jar(s) location to classpath
// At some point we should not be required to add
// the hadoop specific classpaths to the env.
// It should be provided out of the box.
// For now setting all required classpaths including
// the classpath to "." for the application jar(s)
// including ${CLASSPATH} will duplicate the class path in app master, removing it for now
// StringBuilder classPathEnv = new StringBuilder("${CLASSPATH}:./*");
StringBuilder classPathEnv = new StringBuilder("./*");
String classpath = conf.get(YarnConfiguration.YARN_APPLICATION_CLASSPATH);
for (String c :
StringUtils.isBlank(classpath)
? YarnConfiguration.DEFAULT_YARN_APPLICATION_CLASSPATH
: classpath.split(",")) {
if (c.equals("$HADOOP_CLIENT_CONF_DIR")) {
// SPOI-2501
continue;
}
classPathEnv.append(':');
classPathEnv.append(c.trim());
}
env.put("CLASSPATH", classPathEnv.toString());
// propagate to replace node managers user name (effective in non-secure mode)
env.put("HADOOP_USER_NAME", UserGroupInformation.getLoginUser().getUserName());
amContainer.setEnvironment(env);
// Set the necessary command to execute the application master
ArrayList<CharSequence> vargs = new ArrayList<CharSequence>(30);
// Set java executable command
LOG.info("Setting up app master command");
vargs.add(javaCmd);
if (dag.isDebug()) {
vargs.add("-agentlib:jdwp=transport=dt_socket,server=y,suspend=n");
}
// Set Xmx based on am memory size
// default heap size 75% of total memory
vargs.add("-Xmx" + (amMemory * 3 / 4) + "m");
vargs.add("-XX:+HeapDumpOnOutOfMemoryError");
vargs.add("-XX:HeapDumpPath=/tmp/dt-heap-" + appId.getId() + ".bin");
vargs.add("-Dhadoop.root.logger=" + (dag.isDebug() ? "DEBUG" : "INFO") + ",RFA");
vargs.add("-Dhadoop.log.dir=" + ApplicationConstants.LOG_DIR_EXPANSION_VAR);
vargs.add(String.format("-D%s=%s", StreamingContainer.PROP_APP_PATH, dag.assertAppPath()));
if (dag.isDebug()) {
vargs.add("-Dlog4j.debug=true");
}
String loggersLevel = conf.get(DTLoggerFactory.DT_LOGGERS_LEVEL);
if (loggersLevel != null) {
vargs.add(String.format("-D%s=%s", DTLoggerFactory.DT_LOGGERS_LEVEL, loggersLevel));
}
vargs.add(StreamingAppMaster.class.getName());
vargs.add("1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stdout");
vargs.add("2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stderr");
// Get final command
StringBuilder command = new StringBuilder(9 * vargs.size());
for (CharSequence str : vargs) {
command.append(str).append(" ");
}
LOG.info("Completed setting up app master command " + command.toString());
List<String> commands = new ArrayList<String>();
commands.add(command.toString());
amContainer.setCommands(commands);
// Set up resource type requirements
// For now, only memory is supported so we set memory requirements
Resource capability = Records.newRecord(Resource.class);
capability.setMemory(amMemory);
appContext.setResource(capability);
// Service data is a binary blob that can be passed to the application
// Not needed in this scenario
// amContainer.setServiceData(serviceData);
appContext.setAMContainerSpec(amContainer);
// Set the priority for the application master
Priority pri = Records.newRecord(Priority.class);
pri.setPriority(amPriority);
appContext.setPriority(pri);
// Set the queue to which this application is to be submitted in the RM
appContext.setQueue(queueName);
// Submit the application to the applications manager
// SubmitApplicationResponse submitResp = rmClient.submitApplication(appRequest);
// Ignore the response as either a valid response object is returned on success
// or an exception thrown to denote some form of a failure
String specStr =
Objects.toStringHelper("Submitting application: ")
.add("name", appContext.getApplicationName())
.add("queue", appContext.getQueue())
.add("user", UserGroupInformation.getLoginUser())
.add("resource", appContext.getResource())
.toString();
LOG.info(specStr);
if (dag.isDebug()) {
// LOG.info("Full submission context: " + appContext);
}
yarnClient.submitApplication(appContext);
} finally {
fs.close();
}
}
/**
* 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);
}
/**
* 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);
}
private void parseMetaData() throws IOException {
Text line = new Text();
long read;
FSDataInputStream in = null;
LineReader lin = null;
try {
in = fs.open(masterIndexPath);
FileStatus masterStat = fs.getFileStatus(masterIndexPath);
masterIndexTimestamp = masterStat.getModificationTime();
lin = new LineReader(in, getConf());
read = lin.readLine(line);
// the first line contains the version of the index file
String versionLine = line.toString();
String[] arr = versionLine.split(" ");
version = Integer.parseInt(arr[0]);
// make it always backwards-compatible
if (this.version > HarFileSystem.VERSION) {
throw new IOException(
"Invalid version " + this.version + " expected " + HarFileSystem.VERSION);
}
// each line contains a hashcode range and the index file name
String[] readStr;
while (read < masterStat.getLen()) {
int b = lin.readLine(line);
read += b;
readStr = line.toString().split(" ");
int startHash = Integer.parseInt(readStr[0]);
int endHash = Integer.parseInt(readStr[1]);
stores.add(
new Store(
Long.parseLong(readStr[2]), Long.parseLong(readStr[3]), startHash, endHash));
line.clear();
}
} catch (IOException ioe) {
LOG.warn("Encountered exception ", ioe);
throw ioe;
} finally {
IOUtils.cleanup(LOG, lin, in);
}
FSDataInputStream aIn = fs.open(archiveIndexPath);
try {
FileStatus archiveStat = fs.getFileStatus(archiveIndexPath);
archiveIndexTimestamp = archiveStat.getModificationTime();
LineReader aLin;
// now start reading the real index file
for (Store s : stores) {
read = 0;
aIn.seek(s.begin);
aLin = new LineReader(aIn, getConf());
while (read + s.begin < s.end) {
int tmp = aLin.readLine(line);
read += tmp;
String lineFeed = line.toString();
String[] parsed = lineFeed.split(" ");
parsed[0] = decodeFileName(parsed[0]);
archive.put(new Path(parsed[0]), new HarStatus(lineFeed));
line.clear();
}
}
} finally {
IOUtils.cleanup(LOG, aIn);
}
}
/** * 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; }
