@Override
public int run(String[] args) throws Exception {
CommandLine cmd = getCommand(args);
DistCpOptions options = getDistCpOptions(cmd);
Configuration conf = this.getConf();
// inject wf configs
Path confPath = new Path("file:///" + System.getProperty("oozie.action.conf.xml"));
LOG.info(confPath + " found conf ? " + confPath.getFileSystem(conf).exists(confPath));
conf.addResource(confPath);
String falconFeedStorageType = cmd.getOptionValue("falconFeedStorageType").trim();
Storage.TYPE feedStorageType = Storage.TYPE.valueOf(falconFeedStorageType);
DistCp distCp =
(feedStorageType == Storage.TYPE.FILESYSTEM)
? new CustomReplicator(conf, options)
: new DistCp(conf, options);
LOG.info("Started DistCp");
distCp.execute();
if (feedStorageType == Storage.TYPE.FILESYSTEM) {
executePostProcessing(options); // this only applies for FileSystem Storage.
}
LOG.info("Completed DistCp");
return 0;
}
private static void invokeCopy(Configuration config, String[] parsedArgs) {
try {
log.info(
"Running DistCp with arguments ["
+ StringUtils.arrayToCommaDelimitedString(parsedArgs)
+ "]");
DistCpOptions inputOptions = OptionsParser.parse(parsedArgs);
org.apache.hadoop.tools.DistCp distCp =
new org.apache.hadoop.tools.DistCp(config, inputOptions);
distCp.execute();
} catch (Exception e) {
throw new HadoopException("Error running DistCp job", e);
}
}
private void updatePermissions(FileStatus src, FileStatus dst) throws IOException {
if (preserve_status) {
DistCp.updatePermissions(src, dst, preseved, destFileSys);
}
}
@Override
public boolean runDistCp(Path src, Path dst, Configuration conf) throws IOException {
DistCpOptions options = new DistCpOptions(Collections.singletonList(src), dst);
options.setSyncFolder(true);
options.setSkipCRC(true);
options.preserve(FileAttribute.BLOCKSIZE);
// Creates the command-line parameters for distcp
String[] params = {"-update", "-skipcrccheck", src.toString(), dst.toString()};
try {
conf.setBoolean("mapred.mapper.new-api", true);
DistCp distcp = new DistCp(conf, options);
// HIVE-13704 states that we should use run() instead of execute() due to a hadoop known issue
// added by HADOOP-10459
if (distcp.run(params) == 0) {
return true;
} else {
return false;
}
} catch (Exception e) {
throw new IOException("Cannot execute DistCp process: " + e, e);
} finally {
conf.setBoolean("mapred.mapper.new-api", false);
}
}