Example usage for org.apache.hadoop.conf Configuration addResource

List of usage examples for org.apache.hadoop.conf Configuration addResource

Introduction

In this page you can find the example usage for org.apache.hadoop.conf Configuration addResource.

Prototype

public void addResource(Configuration conf) 

Source Link

Document

Add a configuration resource.

Usage

From source file:com.odiago.flumebase.util.AppUtils.java

License:Apache License

/**
 * @return a Configuration with the appropriate flumebase-site.xml file added.
 * Should be used as the base Configuration for the program.
 *//*from   ww w  . j  av  a 2s.co m*/
public static Configuration initConfResources() {
    String flumebaseConfFile = new File(getAppConfDir(), "flumebase-site.xml").toString();
    LOG.debug("Initializing configuration from " + flumebaseConfFile);
    Configuration conf = new Configuration();
    conf.addResource(new Path(flumebaseConfFile));
    return conf;
}

From source file:com.pivotal.hawq.mapreduce.MapReduceClusterDriver.java

License:Apache License

public MapReduceClusterDriver() {
    Configuration conf = new Configuration();
    conf.addResource(new Path(HADOOP_HOME + "/etc/hadoop/hdfs-site.xml"));
    conf.addResource(new Path(HADOOP_HOME + "/etc/hadoop/core-site.xml"));

    setConf(conf);/*  ww w .j av  a2  s.c  om*/
}

From source file:com.rim.logdriver.admin.LogMaintenance.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf();
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }//from   w  ww  .j  ava  2  s. c  o m

    // For some reason, Oozie needs some options to be set in system instead of
    // in the confiuration. So copy the configs over.
    {
        Iterator<Entry<String, String>> i = conf.iterator();
        while (i.hasNext()) {
            Entry<String, String> next = i.next();
            System.setProperty(next.getKey(), next.getValue());
        }
    }

    if (args.length < 3) {
        printUsage();
        return 1;
    }

    String userName = args[0];
    String dcNumber = args[1];
    String service = args[2];
    String date = null;
    String hour = null;
    if (args.length >= 4) {
        date = args[3];
    }
    if (args.length >= 5) {
        hour = args[4];
    }

    // Set from environment variables
    oozieUrl = getConfOrEnv(conf, "OOZIE_URL");
    String mergeJobPropertiesFile = getConfOrEnv(conf, "MERGEJOB_CONF");
    String filterJobPropertiesFile = getConfOrEnv(conf, "FILTERJOB_CONF");
    String daysBeforeArchive = getConfOrEnv(conf, "DAYS_BEFORE_ARCHIVE");
    String daysBeforeDelete = getConfOrEnv(conf, "DAYS_BEFORE_DELETE");
    String maxConcurrentMergeJobs = getConfOrEnv(conf, "MAX_CONCURRENT_MERGE_JOBS");
    String maxConcurrentFilterJobs = getConfOrEnv(conf, "MAX_CONCURRENT_FILTER_JOBS");
    String zkConnectString = getConfOrEnv(conf, "ZK_CONNECT_STRING");
    String logdir = getConfOrEnv(conf, "logdriver.logdir.name");
    boolean resetOrphanedJobs = Boolean.parseBoolean(getConfOrEnv(conf, "reset.orphaned.jobs"));
    String rootDir = getConfOrEnv(conf, "service.root.dir");

    boolean doMerge = true;
    boolean doArchive = true;
    boolean doDelete = true;

    if (oozieUrl == null) {
        LOG.info("OOZIE_URL is not set.  Not merging or archiving.");
        doMerge = false;
        doArchive = false;
    }
    if (zkConnectString == null) {
        LOG.error("ZK_CONNECT_STRING is not set.  Exiting.");
        return 1;
    }
    if (mergeJobPropertiesFile == null) {
        LOG.info("MERGEJOB_CONF is not set.  Not merging.");
        doMerge = false;
    }
    if (filterJobPropertiesFile == null) {
        LOG.info("FILTERJOB_CONF is not set.  Not archiving.");
        doArchive = false;
    }
    if (daysBeforeArchive == null) {
        LOG.info("DAYS_BEFORE_ARCHIVE is not set.  Not archiving.");
        doArchive = false;
    }
    if (doArchive && Integer.parseInt(daysBeforeArchive) < 0) {
        LOG.info("DAYS_BEFORE_ARCHIVE is negative.  Not archiving.");
        doArchive = false;
    }
    if (daysBeforeDelete == null) {
        LOG.info("DAYS_BEFORE_DELETE is not set.  Not deleting.");
        doDelete = false;
    }
    if (doDelete && Integer.parseInt(daysBeforeDelete) < 0) {
        LOG.info("DAYS_BEFORE_DELETE is negative.  Not deleting.");
        doDelete = false;
    }
    if (maxConcurrentMergeJobs == null) {
        LOG.info("MAX_CONCURRENT_MERGE_JOBS is not set.  Using default value of -1.");
        maxConcurrentMergeJobs = "-1";
    }
    if (maxConcurrentFilterJobs == null) {
        LOG.info("MAX_CONCURRENT_FILTER_JOBS is not set.  Using default value of -1.");
        maxConcurrentMergeJobs = "-1";
    }
    if (logdir == null) {
        LOG.info("LOGDRIVER_LOGDIR_NAME is not set.  Using default value of 'logs'.");
        logdir = "logs";
    }
    if (rootDir == null) {
        LOG.info("SERVICE_ROOT_DIR is not set.  Using default value of 'service'.");
        rootDir = "/service";
    }

    // Now it's safe to create our Oozie Runners.
    OozieRunner mergeOozieRunner = new OozieRunner(oozieUrl, Integer.parseInt(maxConcurrentMergeJobs));
    Thread mergeOozieRunnerThread = new Thread(mergeOozieRunner);
    mergeOozieRunnerThread.setName("OozieRunner - Merge");
    mergeOozieRunnerThread.setDaemon(false);
    mergeOozieRunnerThread.start();

    OozieRunner filterOozieRunner = new OozieRunner(oozieUrl, Integer.parseInt(maxConcurrentFilterJobs));
    Thread filterOozieRunnerThread = new Thread(filterOozieRunner);
    filterOozieRunnerThread.setName("OozieRunner - Filter");
    filterOozieRunnerThread.setDaemon(false);
    filterOozieRunnerThread.start();

    // Figure out what date we start filters on.
    String filterCutoffDate = "";
    if (doArchive) {
        Calendar cal = Calendar.getInstance();
        cal.add(Calendar.DAY_OF_MONTH, Integer.parseInt("-" + daysBeforeArchive));
        filterCutoffDate = String.format("%04d%02d%02d%02d", cal.get(Calendar.YEAR),
                (cal.get(Calendar.MONTH) + 1), cal.get(Calendar.DAY_OF_MONTH), cal.get(Calendar.HOUR_OF_DAY));
        LOG.info("Archiving logs from before {}", filterCutoffDate);
    }
    String deleteCutoffDate = "";
    if (doDelete) {
        Calendar cal = Calendar.getInstance();
        cal.add(Calendar.DAY_OF_MONTH, Integer.parseInt("-" + daysBeforeDelete));
        deleteCutoffDate = String.format("%04d%02d%02d%02d", cal.get(Calendar.YEAR),
                (cal.get(Calendar.MONTH) + 1), cal.get(Calendar.DAY_OF_MONTH), cal.get(Calendar.HOUR_OF_DAY));
        LOG.info("Deleting logs from before {}", deleteCutoffDate);
    }

    long now = System.currentTimeMillis();

    // Various exceptions have been popping up here. So make sure I catch them
    // all.
    try {
        // We can hang if this fails. So make sure we abort if it fails.
        FileSystem fs = null;
        try {
            fs = FileSystem.get(conf);
            fs.exists(new Path("/")); // Test if it works.
        } catch (IOException e) {
            LOG.error("Error getting filesystem.", e);
            return 1;
        }
        // We'll need an Oozie client to check on orphaned directories.
        oozieClient = getOozieClient();

        // LockUtils are used in a couple of places
        LockUtil lu = new LockUtil(zkConnectString);

        // Patterns to recognize hour, day and incoming directories, so that they
        // can be processed.
        Pattern datePathPattern;
        Pattern hourPathPattern;
        Pattern incomingPathPattern;
        Pattern dataPathPattern;
        Pattern archivePathPattern;
        Pattern workingPathPattern;
        if (hour != null) {
            datePathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")");
            hourPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/("
                    + Pattern.quote(hour) + ")");
            incomingPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/("
                    + Pattern.quote(hour) + ")/([^/]+)/incoming");
            dataPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/("
                    + Pattern.quote(hour) + ")/([^/]+)/data");
            archivePathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/("
                    + Pattern.quote(hour) + ")/([^/]+)/archive");
            workingPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/("
                    + Pattern.quote(hour) + ")/([^/]+)/working/([^/]+)_(\\d+)");
        } else if (date != null) {
            datePathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")");
            hourPathPattern = Pattern
                    .compile(rootDir + "/" + Pattern.quote(dcNumber) + "/" + Pattern.quote(service) + "/"
                            + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/(\\d{2})");
            incomingPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date)
                    + ")/(\\d{2})/([^/]+)/incoming");
            dataPathPattern = Pattern
                    .compile(rootDir + "/" + Pattern.quote(dcNumber) + "/" + Pattern.quote(service) + "/"
                            + Pattern.quote(logdir) + "/(" + Pattern.quote(date) + ")/(\\d{2})/([^/]+)/data");
            archivePathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date)
                    + ")/(\\d{2})/([^/]+)/archive");
            workingPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(" + Pattern.quote(date)
                    + ")/(\\d{2})/([^/]+)/working/([^/]+)_(\\d+)");
        } else {
            datePathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(\\d{8})");
            hourPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(\\d{8})/(\\d{2})");
            incomingPathPattern = Pattern
                    .compile(rootDir + "/" + Pattern.quote(dcNumber) + "/" + Pattern.quote(service) + "/"
                            + Pattern.quote(logdir) + "/(\\d{8})/(\\d{2})/([^/]+)/incoming");
            dataPathPattern = Pattern.compile(rootDir + "/" + Pattern.quote(dcNumber) + "/"
                    + Pattern.quote(service) + "/" + Pattern.quote(logdir) + "/(\\d{8})/(\\d{2})/([^/]+)/data");
            archivePathPattern = Pattern
                    .compile(rootDir + "/" + Pattern.quote(dcNumber) + "/" + Pattern.quote(service) + "/"
                            + Pattern.quote(logdir) + "/(\\d{8})/(\\d{2})/([^/]+)/archive");
            workingPathPattern = Pattern
                    .compile(rootDir + "/" + Pattern.quote(dcNumber) + "/" + Pattern.quote(service) + "/"
                            + Pattern.quote(logdir) + "/(\\d{8})/(\\d{2})/([^/]+)/working/([^/]+)_(\\d+)");
        }

        // Do a depth first search of the directory, processing anything that
        // looks
        // interesting along the way
        Deque<Path> paths = new ArrayDeque<Path>();
        Path rootPath = new Path(rootDir + "/" + dcNumber + "/" + service + "/" + logdir + "/");
        paths.push(rootPath);

        while (paths.size() > 0) {
            Path p = paths.pop();
            LOG.debug("{}", p.toString());

            if (!fs.exists(p)) {
                continue;
            }

            FileStatus dirStatus = fs.getFileStatus(p);
            FileStatus[] children = fs.listStatus(p);
            boolean addChildren = true;

            boolean old = dirStatus.getModificationTime() < now - WAIT_TIME;
            LOG.debug("    Was last modified {}ms ago", now - dirStatus.getModificationTime());

            if (!old) {
                LOG.debug("    Skipping, since it's not old enough.");

            } else if ((!rootPath.equals(p)) && (children.length == 0
                    || (children.length == 1 && children[0].getPath().getName().equals(READY_MARKER)))) {
                // old and no children? Delete!
                LOG.info("    Deleting empty directory {}", p.toString());
                fs.delete(p, true);

            } else {
                Matcher matcher = datePathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    LOG.debug("Checking date directory");

                    // If this is already done, then skip it. So only process if it
                    // doesn't exist.
                    if (fs.exists(new Path(p, READY_MARKER)) == false) {
                        // Check each subdirectory. If they all have ready markers, then I
                        // guess we're ready.
                        boolean ready = true;
                        for (FileStatus c : children) {
                            if (c.isDir() && fs.exists(new Path(c.getPath(), READY_MARKER)) == false) {
                                ready = false;
                                break;
                            }
                        }

                        if (ready) {
                            fs.createNewFile(new Path(p, READY_MARKER));
                        }
                    }
                }

                matcher = hourPathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    LOG.debug("Checking hour directory");

                    // If this is already done, then skip it. So only process if it
                    // doesn't exist.
                    if (fs.exists(new Path(p, READY_MARKER)) == false) {
                        // Check each subdirectory. If they all have ready markers, then I
                        // guess we're ready.
                        boolean ready = true;
                        for (FileStatus c : children) {
                            if (c.isDir() && fs.exists(new Path(c.getPath(), READY_MARKER)) == false) {
                                ready = false;
                                break;
                            }
                        }

                        if (ready) {
                            fs.createNewFile(new Path(p, READY_MARKER));
                        }
                    }
                }

                // Check to see if we have to run a merge
                matcher = incomingPathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    LOG.debug("Checking incoming directory");
                    String matchDate = matcher.group(1);
                    String matchHour = matcher.group(2);
                    String matchComponent = matcher.group(3);

                    String timestamp = matchDate + matchHour;

                    if (doDelete && timestamp.compareTo(deleteCutoffDate) < 0) {
                        LOG.info("Deleting old directory: {}", p);
                        fs.delete(p, true);
                        addChildren = false;
                    } else if (doMerge) {

                        // old, looks right, and has children? Run it!
                        boolean hasMatchingChildren = false;
                        boolean subdirTooYoung = false;

                        for (FileStatus child : children) {
                            if (!hasMatchingChildren) {
                                FileStatus[] grandchildren = fs.listStatus(child.getPath());
                                for (FileStatus gc : grandchildren) {
                                    if (VALID_FILE.matcher(gc.getPath().getName()).matches()) {
                                        hasMatchingChildren = true;
                                        break;
                                    }
                                }
                            }
                            if (!subdirTooYoung) {
                                if (child.getModificationTime() >= now - WAIT_TIME) {
                                    subdirTooYoung = true;
                                    LOG.debug("    Subdir {} is too young.", child.getPath());
                                }
                            }
                        }

                        if (!hasMatchingChildren) {
                            LOG.debug("    No files match the expected pattern ({})", VALID_FILE.pattern());
                        }

                        if (hasMatchingChildren && !subdirTooYoung) {
                            LOG.info("    Run Merge job {} :: {} {} {} {} {}", new Object[] { p.toString(),
                                    dcNumber, service, matchDate, matchHour, matchComponent });

                            Properties oozieJobProps = new Properties();
                            oozieJobProps.load(new FileInputStream(mergeJobPropertiesFile));

                            oozieJobProps.setProperty("rootDir", rootDir);
                            oozieJobProps.setProperty("dcNumber", dcNumber);
                            oozieJobProps.setProperty("service", service);
                            oozieJobProps.setProperty("date", matchDate);
                            oozieJobProps.setProperty("hour", matchHour);
                            oozieJobProps.setProperty("component", matchComponent);
                            oozieJobProps.setProperty("user.name", userName);
                            oozieJobProps.setProperty("logdir", logdir);

                            mergeOozieRunner.submit(oozieJobProps);

                            addChildren = false;
                        }
                    }
                }

                // Check to see if we need to run a filter and archive
                matcher = dataPathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    String matchDate = matcher.group(1);
                    String matchHour = matcher.group(2);
                    String matchComponent = matcher.group(3);

                    String timestamp = matchDate + matchHour;

                    if (doDelete && timestamp.compareTo(deleteCutoffDate) < 0) {
                        LOG.info("Deleting old directory: {}", p);
                        fs.delete(p, true);
                        addChildren = false;
                    } else if (doArchive && timestamp.compareTo(filterCutoffDate) < 0) {

                        Properties oozieJobProps = new Properties();
                        oozieJobProps.load(new FileInputStream(filterJobPropertiesFile));

                        oozieJobProps.setProperty("rootDir", rootDir);
                        oozieJobProps.setProperty("dcNumber", dcNumber);
                        oozieJobProps.setProperty("service", service);
                        oozieJobProps.setProperty("date", matchDate);
                        oozieJobProps.setProperty("hour", matchHour);
                        oozieJobProps.setProperty("component", matchComponent);
                        oozieJobProps.setProperty("user.name", userName);
                        oozieJobProps.setProperty("logdir", logdir);

                        // Check to see if we should just keep all or delete all here.
                        // The filter file should be here
                        String appPath = oozieJobProps.getProperty("oozie.wf.application.path");
                        appPath = appPath.replaceFirst("\\$\\{.*?\\}", "");
                        Path filterFile = new Path(appPath + "/" + service + ".yaml");
                        LOG.info("Filter file is {}", filterFile);
                        if (fs.exists(filterFile)) {
                            List<BoomFilterMapper.Filter> filters = BoomFilterMapper.loadFilters(matchComponent,
                                    fs.open(filterFile));

                            if (filters == null) {
                                LOG.warn(
                                        "    Got null when getting filters.  Not processing. {} :: {} {} {} {} {}",
                                        new Object[] { p.toString(), dcNumber, service, matchDate, matchHour,
                                                matchComponent });
                            } else if (filters.size() == 0) {
                                LOG.warn("    Got no filters.  Not processing. {} :: {} {} {} {} {}",
                                        new Object[] { p.toString(), dcNumber, service, matchDate, matchHour,
                                                matchComponent });
                            } else if (filters.size() == 1
                                    && filters.get(0) instanceof BoomFilterMapper.KeepAllFilter) {
                                LOG.info("    Keeping everything. {} :: {} {} {} {} {}",
                                        new Object[] { p.toString(), dcNumber, service, matchDate, matchHour,
                                                matchComponent });
                                // Move files from data to archive
                                // delete it all!
                                String destination = rootDir + "/" + dcNumber + "/" + service + "/" + logdir
                                        + "/" + matchDate + "/" + matchHour + "/" + matchComponent
                                        + "/archive/";

                                String[] moveArgs = { zkConnectString, dcNumber, service, matchDate, matchHour,
                                        matchComponent, "move " + p.toUri().getPath() + " " + destination };
                                ToolRunner.run(new Configuration(), new LockedFs(), moveArgs);
                            } else if (filters.size() == 1
                                    && filters.get(0) instanceof BoomFilterMapper.DropAllFilter) {
                                LOG.info("    Dropping everything. {} :: {} {} {} {} {}",
                                        new Object[] { p.toString(), dcNumber, service, matchDate, matchHour,
                                                matchComponent });
                                // delete it all!
                                String[] delArgs = { zkConnectString, dcNumber, service, matchDate, matchHour,
                                        matchComponent, "delete " + p.toUri().getPath() };
                                ToolRunner.run(new Configuration(), new LockedFs(), delArgs);
                            } else {
                                LOG.info("    Run Filter/Archive job {} :: {} {} {} {} {}",
                                        new Object[] { p.toString(), dcNumber, service, matchDate, matchHour,
                                                matchComponent });
                                filterOozieRunner.submit(oozieJobProps);
                            }
                        } else {
                            LOG.warn("Skipping filter job, since no filter file exists");
                        }

                        addChildren = false;
                    }
                }

                matcher = archivePathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    String matchDate = matcher.group(1);
                    String matchHour = matcher.group(2);

                    String timestamp = matchDate + matchHour;

                    if (doDelete && timestamp.compareTo(deleteCutoffDate) < 0) {
                        LOG.info("Deleting old directory: {}", p);
                        fs.delete(p, true);
                        addChildren = false;
                    }
                }

                matcher = workingPathPattern.matcher(p.toUri().getPath());
                if (matcher.matches()) {
                    LOG.info("  Matches working pattern");
                    if (resetOrphanedJobs) {
                        String matchDate = matcher.group(1);
                        String matchHour = matcher.group(2);
                        String matchComponent = matcher.group(3);
                        String matchOozieJobId = matcher.group(4);

                        // Check to see what's up with the oozie job. If it's still
                        // running,
                        // we don't want to touch it.
                        Status status = null;
                        try {
                            WorkflowJob jobInfo = oozieClient.getJobInfo(matchOozieJobId);
                            status = jobInfo.getStatus();
                        } catch (OozieClientException e) {
                            if (e.getMessage() != null && e.getMessage().contains("Job does not exist")) {
                                LOG.info("Oozie job not found.  Proceeding as though job was failed.", e);
                                status = Status.FAILED;
                            } else {
                                LOG.error("Oozie client error.  Not Proceeding.", e);
                            }
                        }
                        LOG.info("  Oozie job status is {}", status);
                        if (status != null && status != Status.RUNNING && status != Status.PREP
                                && status != Status.SUSPENDED) {
                            // Move everything from working/xxx/incoming/ to incoming/
                            PathInfo lockPathInfo = new PathInfo(rootDir + "/" + dcNumber + "/" + service + "/"
                                    + logdir + "/" + matchDate + "/" + matchHour + "/" + matchComponent);
                            lu.acquireWriteLock(lu.getLockPath(lockPathInfo));

                            FileStatus[] fileStatuses = fs
                                    .listStatus(new Path(p.toUri().getPath() + "/incoming/"));
                            if (fileStatuses != null) {
                                for (FileStatus fileStatus : fileStatuses) {
                                    Path toPath = new Path(fileStatus.getPath().getParent().getParent()
                                            .getParent().getParent(),
                                            "incoming/" + fileStatus.getPath().getName());

                                    LOG.info("  Moving data from {} to {}", fileStatus.getPath(), toPath);
                                    LOG.info("    mkdir {}", toPath);
                                    fs.mkdirs(toPath);

                                    Path fromDir = new Path(p.toUri().getPath(),
                                            "incoming/" + fileStatus.getPath().getName());
                                    LOG.info("    moving from {}", fromDir);
                                    FileStatus[] files = fs.listStatus(fromDir);
                                    if (files == null || files.length == 0) {
                                        LOG.info("    Nothing to move from  {}", fromDir);
                                    } else {
                                        for (FileStatus f : files) {
                                            LOG.info("    rename {} {}", f.getPath(),
                                                    new Path(toPath, f.getPath().getName()));
                                            fs.rename(f.getPath(), new Path(toPath, f.getPath().getName()));
                                        }
                                    }

                                    LOG.info("    rm {}", fileStatus.getPath().getParent().getParent());
                                    fs.delete(fileStatus.getPath().getParent().getParent(), true);
                                }

                                lu.releaseWriteLock(lu.getLockPath(lockPathInfo));

                            }
                        }
                    }

                    addChildren = false;
                }
            }

            // Add any children which are directories to the stack.
            if (addChildren) {
                for (int i = children.length - 1; i >= 0; i--) {
                    FileStatus child = children[i];
                    if (child.isDir()) {
                        paths.push(child.getPath());
                    }
                }
            }
        }

        // Since we may have deleted a bunch of directories, delete any unused
        // locks
        // from ZooKeeper.
        {
            LOG.info("Checking for unused locks in ZooKeeper");
            String scanPath = rootDir + "/" + dcNumber + "/" + service + "/" + logdir;
            if (date != null) {
                scanPath += "/" + date;
                if (hour != null) {
                    scanPath += "/" + hour;
                }
            }

            List<LockInfo> lockInfo = lu.scan(scanPath);

            for (LockInfo li : lockInfo) {
                // Check if the lock path still exists in HDFS. If it doesn't, then
                // delete it from ZooKeeper.
                String path = li.getPath();
                String hdfsPath = path.substring(LockUtil.ROOT.length());
                if (!fs.exists(new Path(hdfsPath))) {
                    ZooKeeper zk = lu.getZkClient();

                    while (!path.equals(LockUtil.ROOT)) {
                        try {
                            zk.delete(path, -1);
                        } catch (KeeperException.NotEmptyException e) {
                            // That's fine. just stop trying then.
                            break;
                        } catch (Exception e) {
                            LOG.error("Caught exception trying to delete from ZooKeeper.", e);
                            break;
                        }
                        LOG.info("Deleted from ZooKeeper: {}", path);
                        path = path.substring(0, path.lastIndexOf('/'));
                    }

                }
            }
        }
        lu.close();

        // Now that we're done, wait for the Oozie Runner to stop, and print the
        // results.
        LOG.info("Waiting for Oozie jobs to complete.");
        mergeOozieRunner.shutdown();
        mergeOozieRunnerThread.join();
        LOG.info("Oozie Job Stats : Merge  : Started={} Succeeded={} failed={} errors={}",
                new Object[] { mergeOozieRunner.getStarted(), mergeOozieRunner.getSucceeded(),
                        mergeOozieRunner.getFailed(), mergeOozieRunner.getErrors() });

        filterOozieRunner.shutdown();
        filterOozieRunnerThread.join();
        LOG.info("Oozie Job Stats : Filter : Started={} Succeeded={} failed={} errors={}",
                new Object[] { filterOozieRunner.getStarted(), filterOozieRunner.getSucceeded(),
                        filterOozieRunner.getFailed(), filterOozieRunner.getErrors() });

    } catch (Exception e) {
        LOG.error("Unexpected exception caught.", e);
        return 1;
    }

    return 0;
}

From source file:com.rim.logdriver.sawmill.Sawmill.java

License:Apache License

public void run(String[] args) {
    if (args.length < 1) {
        System.out.println("Usage: " + this.getClass().getSimpleName() + " <config.properties>");
        System.exit(1);/*w ww.j a v a  2s.  c o  m*/
    }

    LOG.info("Starting {}", Sawmill.class.getSimpleName());

    // First arg is the config
    String configFile = args[0];

    // Load configuration.
    Properties conf = new Properties();
    try {
        conf.load(new FileInputStream(configFile));
    } catch (FileNotFoundException e) {
        LOG.error("Config file not found.", e);
        System.exit(1);
    } catch (Throwable t) {
        LOG.error("Error reading config file.", t);
        System.exit(1);
    }

    // Parse the configuration.

    // Load in any Hadoop config files.
    Configuration hConf = new Configuration();
    {
        String[] hadoopConfs = Configs.hadoopConfigPaths.getArray(conf);
        for (String confPath : hadoopConfs) {
            hConf.addResource(new Path(confPath));
        }
        // Also, don't shut down my FileSystem automatically!!!
        hConf.setBoolean("fs.automatic.close", false);
        for (Entry<Object, Object> e : System.getProperties().entrySet()) {
            if (e.getValue() instanceof Integer) {
                hConf.setInt(e.getKey().toString(), (Integer) e.getValue());
            } else if (e.getValue() instanceof Long) {
                hConf.setLong(e.getKey().toString(), (Long) e.getValue());
            } else {
                hConf.set(e.getKey().toString(), e.getValue().toString());
            }
        }
    }

    // Ensure that UserGroupInformation is set up, and knows if security is
    // enabled.
    UserGroupInformation.setConfiguration(hConf);

    // Kerberos credentials. If these are not present, then it just won't try to
    // authenticate.
    String kerbConfPrincipal = Configs.kerberosPrincipal.get(conf);
    String kerbKeytab = Configs.kerberosKeytab.get(conf);
    Authenticator.getInstance().setKerbConfPrincipal(kerbConfPrincipal);
    Authenticator.getInstance().setKerbKeytab(kerbKeytab);

    // Check out the number of threads for workers, and creater the threadpools
    // for both workers and stats updates.
    int threadCount = Configs.threadpoolSize.getInteger(conf);
    final ScheduledExecutorService executor = Executors.newScheduledThreadPool(threadCount);

    // Get the MBean server
    MBeanServer mbs = ManagementFactory.getPlatformMBeanServer();

    // Set up the Mina Exception Monitor
    ExceptionMonitor.setInstance(new ExceptionLoggerExceptionMonitor());

    // For each port->output mapping, create a path (listener, queue, worker).
    // List<DataPath> paths = new ArrayList<DataPath>();
    final List<IoAcceptor> acceptors = new ArrayList<IoAcceptor>();
    final List<Writer> writers = new ArrayList<Writer>();
    {
        String[] pathStrings = Configs.paths.getArray(conf);
        for (String p : pathStrings) {
            Properties pathConf = Util.subProperties(conf, "path." + p);

            String name = Configs.name.get(pathConf);
            if (name == null) {
                LOG.info("Path has no name.  Using {}", p);
                name = p;
            }
            LOG.info("[{}] Configuring path {}", name, name);

            // Check the properties for this specific instance
            Integer maxLineLength = Configs.tcpMaxLineLength.getInteger(pathConf);
            if (maxLineLength == null) {
                maxLineLength = Configs.defaultTcpMaxLineLength.getInteger(conf);
            }
            LOG.info("[{}] Maximum line length is {}", name, maxLineLength);

            InetAddress bindAddress = null;
            try {
                String address = Configs.bindAddress.get(pathConf);
                bindAddress = InetAddress.getByName(address);
            } catch (UnknownHostException e) {
                LOG.error("[{}] Error getting bindAddress from string {}",
                        new Object[] { name, pathConf.getProperty("bindAddress") }, e);
            }

            Integer port = Configs.port.getInteger(pathConf);
            if (port == null) {
                LOG.error("[{}] Port not set.  Skipping this path.", name);
                continue;
            }

            int queueLength = Configs.queueCapacity.getInteger(pathConf);

            // Set up the actual processing chain
            IoAcceptor acceptor = new NioSocketAcceptor();
            SocketSessionConfig sessionConfig = (SocketSessionConfig) acceptor.getSessionConfig();
            sessionConfig.setReuseAddress(true);
            acceptors.add(acceptor);

            String charsetName = Configs.charset.getString(pathConf);
            Charset charset = null;
            try {
                charset = Charset.forName(charsetName);
            } catch (UnsupportedCharsetException e) {
                LOG.error("[{}] Charset '{}' is not supported.  Defaulting to UTF-8.", name, charsetName);
                charset = Charset.forName("UTF-8");
            }
            LOG.info("[{}] Using character set {}", name, charset.displayName());
            TextLineCodecFactory textLineCodecFactory = new TextLineCodecFactory(charset, LineDelimiter.UNIX,
                    LineDelimiter.AUTO);
            textLineCodecFactory.setDecoderMaxLineLength(maxLineLength);
            acceptor.getFilterChain().addLast("textLineCodec", new ProtocolCodecFilter(textLineCodecFactory));

            int numBuckets = Configs.outputBuckets.getInteger(pathConf);
            if (numBuckets > 1) {
                // Set up mulitple writers for one MultiEnqueueHandler
                @SuppressWarnings("unchecked")
                BlockingQueue<String>[] queues = new BlockingQueue[numBuckets];

                for (int i = 0; i < numBuckets; i++) {
                    BlockingQueue<String> queue = new ArrayBlockingQueue<String>(queueLength);
                    queues[i] = queue;

                    // Set up the processor on the other end.
                    Writer writer = new Writer();
                    writer.setName(name);
                    writer.setConfig(pathConf);
                    writer.setHadoopConf(hConf);
                    writer.setQueue(queue);
                    writer.init();

                    // Set up MBean for the Writer
                    {
                        ObjectName mbeanName = null;
                        try {
                            mbeanName = new ObjectName(Writer.class.getPackage().getName() + ":type="
                                    + Writer.class.getSimpleName() + " [" + i + "]" + ",name=" + name);
                        } catch (MalformedObjectNameException e) {
                            LOG.error("[{}] Error creating MBean name.", name, e);
                        } catch (NullPointerException e) {
                            LOG.error("[{}] Error creating MBean name.", name, e);
                        }
                        try {
                            mbs.registerMBean(writer, mbeanName);
                        } catch (InstanceAlreadyExistsException e) {
                            LOG.error("[{}] Error registering MBean name.", name, e);
                        } catch (MBeanRegistrationException e) {
                            LOG.error("[{}] Error registering MBean name.", name, e);
                        } catch (NotCompliantMBeanException e) {
                            LOG.error("[{}] Error registering MBean name.", name, e);
                        }
                    }

                    executor.scheduleWithFixedDelay(writer, 0, 100, TimeUnit.MILLISECONDS);
                    writers.add(writer);
                }

                MultiEnqueueHandler handler = new MultiEnqueueHandler(queues);
                acceptor.setHandler(handler);

                // Set up MBean for the MultiEnqueueHandler
                {
                    ObjectName mbeanName = null;
                    try {
                        mbeanName = new ObjectName(MultiEnqueueHandler.class.getPackage().getName() + ":type="
                                + MultiEnqueueHandler.class.getSimpleName() + ",name=" + name);
                    } catch (MalformedObjectNameException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    } catch (NullPointerException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    }
                    try {
                        mbs.registerMBean(handler, mbeanName);
                    } catch (InstanceAlreadyExistsException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (MBeanRegistrationException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (NotCompliantMBeanException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    }
                }
            } else {
                BlockingQueue<String> queue = new ArrayBlockingQueue<String>(queueLength);

                // Set up the processor on the other end.
                Writer writer = new Writer();
                writer.setName(name);
                writer.setConfig(pathConf);
                writer.setHadoopConf(hConf);
                writer.setQueue(queue);
                writer.init();

                // Set up MBean for the Writer
                {
                    ObjectName mbeanName = null;
                    try {
                        mbeanName = new ObjectName(Writer.class.getPackage().getName() + ":type="
                                + Writer.class.getSimpleName() + ",name=" + name);
                    } catch (MalformedObjectNameException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    } catch (NullPointerException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    }
                    try {
                        mbs.registerMBean(writer, mbeanName);
                    } catch (InstanceAlreadyExistsException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (MBeanRegistrationException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (NotCompliantMBeanException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    }
                }

                executor.scheduleWithFixedDelay(writer, 0, 100, TimeUnit.MILLISECONDS);
                writers.add(writer);

                EnqueueHandler handler = new EnqueueHandler(queue);
                acceptor.setHandler(handler);

                // Set up MBean for the EnqueueHandler
                {
                    ObjectName mbeanName = null;
                    try {
                        mbeanName = new ObjectName(EnqueueHandler.class.getPackage().getName() + ":type="
                                + EnqueueHandler.class.getSimpleName() + ",name=" + name);
                    } catch (MalformedObjectNameException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    } catch (NullPointerException e) {
                        LOG.error("[{}] Error creating MBean name.", name, e);
                    }
                    try {
                        mbs.registerMBean(handler, mbeanName);
                    } catch (InstanceAlreadyExistsException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (MBeanRegistrationException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    } catch (NotCompliantMBeanException e) {
                        LOG.error("[{}] Error registering MBean name.", name, e);
                    }
                }
            }

            acceptor.getSessionConfig().setReadBufferSize(Configs.tcpReadBufferSize.getInteger(pathConf));
            acceptor.getSessionConfig().setIdleTime(IdleStatus.BOTH_IDLE, 5);

            while (true) {
                try {
                    acceptor.bind(new InetSocketAddress(bindAddress, port));
                } catch (IOException e) {
                    LOG.error("Error binding to {}:{}.  Retrying...", bindAddress, port);

                    try {
                        Thread.sleep(2000);
                    } catch (InterruptedException e1) {
                        // nothing
                    }

                    continue;
                }

                break;
            }

        }
    }

    // Register a shutdown hook..
    Runtime.getRuntime().addShutdownHook(new Thread() {
        public void run() {
            LOG.info("Shutting down");

            LOG.info("Unbinding and disposing of all IoAcceptors");
            for (IoAcceptor acceptor : acceptors) {
                acceptor.unbind();
                acceptor.dispose(true);
            }

            LOG.info("Shutting down worker threadpools.  This could take a little while.");
            executor.shutdown();
            try {
                executor.awaitTermination(10, TimeUnit.MINUTES);
            } catch (InterruptedException e) {
                LOG.error("Interrupted waiting for writer threadpool termination.", e);
            }
            if (!executor.isTerminated()) {
                LOG.error("Threadpool did not terminate cleanly.");
            }

            LOG.info("Cleaning out any remaining messages from the queues.");
            List<Thread> threads = new ArrayList<Thread>();
            for (final Writer writer : writers) {
                Runnable r = new Runnable() {
                    @Override
                    public void run() {
                        try {
                            writer.runAndClose();
                        } catch (Throwable t) {
                            LOG.error("Error shutting down writer [{}]", writer.getName(), t);
                        }
                    }
                };
                Thread t = new Thread(r);
                t.setDaemon(false);
                t.start();
                threads.add(t);
            }

            for (Thread t : threads) {
                try {
                    t.join();
                } catch (InterruptedException e) {
                    LOG.error("Interrupted waiting for thread to finish.");
                }
            }

            LOG.info("Closing filesystems.");
            try {
                FileSystem.closeAll();
            } catch (Throwable t) {
                LOG.error("Error closing filesystems.", t);
            }

            LOG.info("Finished shutting down cleanly.");
        }
    });
}

From source file:com.rim.logdriver.util.Cat.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf(); // Configuration processed by ToolRunner
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }//w ww  .  ja v  a2s.c  o m

    FileSystem fs = FileSystem.get(conf);

    // The command line options
    List<Path> paths = new ArrayList<Path>();
    Path outputDir = null;

    // Load input files from the command line
    if (args.length < 2) {
        System.out.println("usage: [genericOptions] input [input ...] output");
        System.exit(1);
    }

    // Get the files we need from the command line.
    for (int i = 0; i < args.length - 1; i++) {
        for (FileStatus f : fs.globStatus(new Path(args[i]))) {
            paths.add(f.getPath());
        }
    }
    outputDir = new Path(args[args.length - 1]);

    Job job = new Job(conf);
    Configuration jobConf = job.getConfiguration();

    job.setJarByClass(Cat.class);
    jobConf.setIfUnset("mapred.job.name", "Cat Files");

    // To propagate credentials within Oozie
    if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {
        jobConf.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));
    }

    // Good output separators include things that are unsupported by XML. So we
    // just send the byte value of the character through. The restriction here
    // is that it can't be more than 1 byte when UTF-8 encoded, since it will be
    // read by Pig which only deals with single byte separators.
    {
        String outputSeparator = jobConf.get("logdriver.output.field.separator", DEFAULT_OUTPUT_SEPARATOR);
        byte[] bytes = outputSeparator.getBytes(UTF_8);
        if (bytes.length != 1) {
            LOG.error("The output separator must be a single byte in UTF-8.");
            return 1;
        }

        jobConf.set("logdriver.output.field.separator", Byte.toString(bytes[0]));
    }

    job.setInputFormatClass(BoomInputFormat.class);
    job.setMapperClass(CatMapper.class);
    job.setMapOutputKeyClass(Text.class);
    job.setMapOutputValueClass(NullWritable.class);

    job.setNumReduceTasks(0);

    job.setOutputFormatClass(TextOutputFormat.class);
    TextOutputFormat.setOutputPath(job, outputDir);

    for (Path path : paths) {
        BoomInputFormat.addInputPath(job, path);
    }

    // Run the job.
    if (conf.getBoolean("job.wait", DEFAULT_WAIT_JOB)) {
        return job.waitForCompletion(true) ? 0 : 1;
    } else {
        job.submit();
        return 0;
    }
}

From source file:com.rim.logdriver.util.FastSearch.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf(); // Configuration processed by ToolRunner
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }//from  w ww . j a va 2 s.  c om

    FileSystem fs = FileSystem.get(conf);

    // The command line options
    String searchString = null;
    List<Path> paths = new ArrayList<Path>();
    Path outputDir = null;

    // Load input files from the command line
    if (args.length < 3) {
        System.out.println("usage: [genericOptions] searchString input [input ...] output");
        System.exit(1);
    }

    // Get the files we need from the command line.
    searchString = args[0];
    for (int i = 1; i < args.length - 1; i++) {
        for (FileStatus f : fs.globStatus(new Path(args[i]))) {
            paths.add(f.getPath());
        }
    }
    outputDir = new Path(args[args.length - 1]);

    Job job = new Job(conf);
    Configuration jobConf = job.getConfiguration();

    job.setJarByClass(FastSearch.class);
    jobConf.setIfUnset("mapred.job.name", "Search Files");

    // To propagate credentials within Oozie
    if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {
        jobConf.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));
    }

    // Good output separators include things that are unsupported by XML. So we
    // just send the byte value of the character through. The restriction here
    // is that it can't be more than 1 byte when UTF-8 encoded, since it will be
    // read by Pig which only deals with single byte separators.
    {
        String outputSeparator = jobConf.get("logdriver.output.field.separator", DEFAULT_OUTPUT_SEPARATOR);
        byte[] bytes = outputSeparator.getBytes(UTF_8);
        if (bytes.length != 1) {
            LOG.error("The output separator must be a single byte in UTF-8.");
            return 1;
        }

        jobConf.set("logdriver.output.field.separator", Byte.toString(bytes[0]));
    }

    jobConf.set("logdriver.search.string", Base64.encodeBase64String(searchString.getBytes("UTF-8")));

    job.setInputFormatClass(AvroBlockInputFormat.class);
    job.setMapperClass(SearchMapper.class);
    job.setMapOutputKeyClass(Text.class);
    job.setMapOutputValueClass(NullWritable.class);

    job.setNumReduceTasks(0);

    // And set the output as usual
    job.setOutputFormatClass(TextOutputFormat.class);
    TextOutputFormat.setOutputPath(job, outputDir);
    for (Path path : paths) {
        AvroBlockInputFormat.addInputPath(job, path);
    }

    // Run the job.
    if (conf.getBoolean("job.wait", DEFAULT_WAIT_JOB)) {
        return job.waitForCompletion(true) ? 0 : 1;
    } else {
        job.submit();
        return 0;
    }
}

From source file:com.rim.logdriver.util.Grep.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf(); // Configuration processed by ToolRunner
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }//  w w  w.  j av  a2  s . c  o m

    FileSystem fs = FileSystem.get(conf);

    // The command line options
    String regex = null;
    List<Path> paths = new ArrayList<Path>();
    Path outputDir = null;

    // Load input files from the command line
    if (args.length < 3) {
        System.out.println("usage: [genericOptions] regex input [input ...] output");
        System.exit(1);
    }

    // Get the files we need from the command line.
    regex = args[0];
    for (int i = 1; i < args.length - 1; i++) {
        for (FileStatus f : fs.globStatus(new Path(args[i]))) {
            paths.add(f.getPath());
        }
    }
    outputDir = new Path(args[args.length - 1]);

    Job job = new Job(conf);
    Configuration jobConf = job.getConfiguration();

    job.setJarByClass(Grep.class);
    jobConf.setIfUnset("mapred.job.name", "Grep Files");

    // To propagate credentials within Oozie
    if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {
        jobConf.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));
    }

    // Good output separators include things that are unsupported by XML. So we
    // just send the byte value of the character through. The restriction here
    // is that it can't be more than 1 byte when UTF-8 encoded, since it will be
    // read by Pig which only deals with single byte separators.
    {
        String outputSeparator = jobConf.get("logdriver.output.field.separator", DEFAULT_OUTPUT_SEPARATOR);
        byte[] bytes = outputSeparator.getBytes(UTF_8);
        if (bytes.length != 1) {
            LOG.error("The output separator must be a single byte in UTF-8.");
            return 1;
        }

        jobConf.set("logdriver.output.field.separator", Byte.toString(bytes[0]));
    }

    jobConf.set("logdriver.grep.regex", Base64.encodeBase64String(regex.getBytes("UTF-8")));

    job.setInputFormatClass(BoomInputFormat.class);
    job.setMapperClass(GrepMapper.class);
    job.setMapOutputKeyClass(Text.class);
    job.setMapOutputValueClass(NullWritable.class);

    job.setNumReduceTasks(0);

    // And set the output as usual
    job.setOutputFormatClass(TextOutputFormat.class);
    TextOutputFormat.setOutputPath(job, outputDir);
    for (Path path : paths) {
        BoomInputFormat.addInputPath(job, path);
    }

    // Run the job.
    if (conf.getBoolean("job.wait", DEFAULT_WAIT_JOB)) {
        return job.waitForCompletion(true) ? 0 : 1;
    } else {
        job.submit();
        return 0;
    }

}

From source file:com.rim.logdriver.util.MultiSearch.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf(); // Configuration processed by ToolRunner
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }/*from www.j ava 2  s .  com*/

    FileSystem fs = FileSystem.get(conf);

    // The command line options
    String searchStringDir = null;
    List<Path> paths = new ArrayList<Path>();
    Path outputDir = null;

    // Load input files from the command line
    if (args.length < 3) {
        System.out.println("usage: [genericOptions] searchStringDirectory input [input ...] output");
        System.exit(1);
    }

    // Get the files we need from the command line.
    searchStringDir = args[0];
    // We are going to be reading all the files in this directory a lot. So
    // let's up the replication factor by a lot so that they're easy to read.
    for (FileStatus f : fs.listStatus(new Path(searchStringDir))) {
        fs.setReplication(f.getPath(), (short) 16);
    }

    for (int i = 1; i < args.length - 1; i++) {
        for (FileStatus f : fs.globStatus(new Path(args[i]))) {
            paths.add(f.getPath());
        }
    }

    outputDir = new Path(args[args.length - 1]);

    Job job = new Job(conf);
    Configuration jobConf = job.getConfiguration();

    job.setJarByClass(MultiSearch.class);
    jobConf.setIfUnset("mapred.job.name", "MultiSearch");

    // To propagate credentials within Oozie
    if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {
        jobConf.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));
    }

    // Good output separators include things that are unsupported by XML. So we
    // just send the byte value of the character through. The restriction here
    // is that it can't be more than 1 byte when UTF-8 encoded, since it will be
    // read by Pig which only deals with single byte separators.
    {
        String outputSeparator = jobConf.get("logdriver.output.field.separator", DEFAULT_OUTPUT_SEPARATOR);
        byte[] bytes = outputSeparator.getBytes(UTF_8);
        if (bytes.length != 1) {
            LOG.error("The output separator must be a single byte in UTF-8.");
            return 1;
        }

        jobConf.set("logdriver.output.field.separator", Byte.toString(bytes[0]));
    }

    jobConf.set("logdriver.search.string.dir", searchStringDir);

    // This search is generally too fast to make good use of 128MB blocks, so
    // let's set the value to 256MB (if it's not set already)
    if (jobConf.get("mapred.max.split.size") == null) {
        jobConf.setLong("mapred.max.split.size", 256 * 1024 * 1024);
    }

    job.setInputFormatClass(AvroBlockInputFormat.class);
    job.setMapperClass(SearchMapper.class);
    job.setMapOutputKeyClass(Text.class);
    job.setMapOutputValueClass(NullWritable.class);

    job.setNumReduceTasks(0);

    job.setOutputFormatClass(TextOutputFormat.class);
    TextOutputFormat.setOutputPath(job, outputDir);
    for (Path path : paths) {
        AvroBlockInputFormat.addInputPath(job, path);
    }

    // Run the job.
    if (conf.getBoolean("job.wait", DEFAULT_WAIT_JOB)) {
        return job.waitForCompletion(true) ? 0 : 1;
    } else {
        job.submit();
        return 0;
    }
}

From source file:com.rim.logdriver.util.Search.java

License:Apache License

@Override
public int run(String[] args) throws Exception {
    Configuration conf = getConf(); // Configuration processed by ToolRunner
    // If run by Oozie, then load the Oozie conf too
    if (System.getProperty("oozie.action.conf.xml") != null) {
        conf.addResource(new URL("file://" + System.getProperty("oozie.action.conf.xml")));
    }//from  w ww .  ja v a  2 s  .c om

    FileSystem fs = FileSystem.get(conf);

    // The command line options
    String searchString = null;
    List<Path> paths = new ArrayList<Path>();
    Path outputDir = null;

    // Load input files from the command line
    if (args.length < 3) {
        System.out.println("usage: [genericOptions] searchString input [input ...] output");
        System.exit(1);
    }

    // Get the files we need from the command line.
    searchString = args[0];
    for (int i = 1; i < args.length - 1; i++) {
        for (FileStatus f : fs.globStatus(new Path(args[i]))) {
            paths.add(f.getPath());
        }
    }
    outputDir = new Path(args[args.length - 1]);

    Job job = new Job(conf);
    Configuration jobConf = job.getConfiguration();

    job.setJarByClass(Search.class);
    jobConf.setIfUnset("mapred.job.name", "Search Files");

    // To propagate credentials within Oozie
    if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {
        jobConf.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));
    }

    // Good output separators include things that are unsupported by XML. So we
    // just send the byte value of the character through. The restriction here
    // is that it can't be more than 1 byte when UTF-8 encoded, since it will be
    // read by Pig which only deals with single byte separators.
    {
        String outputSeparator = jobConf.get("logdriver.output.field.separator", DEFAULT_OUTPUT_SEPARATOR);
        byte[] bytes = outputSeparator.getBytes(UTF_8);
        if (bytes.length != 1) {
            LOG.error("The output separator must be a single byte in UTF-8.");
            return 1;
        }

        jobConf.set("logdriver.output.field.separator", Byte.toString(bytes[0]));
    }

    jobConf.set("logdriver.search.string", searchString);

    job.setInputFormatClass(BoomInputFormat.class);
    job.setMapperClass(SearchMapper.class);
    job.setMapOutputKeyClass(Text.class);
    job.setMapOutputValueClass(NullWritable.class);

    job.setNumReduceTasks(0);

    // And set the output as usual
    job.setOutputFormatClass(TextOutputFormat.class);
    TextOutputFormat.setOutputPath(job, outputDir);
    for (Path path : paths) {
        BoomInputFormat.addInputPath(job, path);
    }

    // Run the job.
    if (conf.getBoolean("job.wait", DEFAULT_WAIT_JOB)) {
        return job.waitForCompletion(true) ? 0 : 1;
    } else {
        job.submit();
        return 0;
    }
}

From source file:com.rockstor.util.RockConfiguration.java

License:Apache License

public static Configuration addRockStorResources(Configuration conf) {
    conf.addResource("rockstor-default.xml");
    conf.addResource("rockstor-site.xml");
    LOG.debug("rockstor.bin.home: " + conf.getResource("./"));
    LOG.debug("rockstor.default.config.path: " + conf.getResource("rockstor-default.xml"));
    LOG.debug("rockstor.user.config.path: " + conf.getResource("rockstor-site.xml"));
    checkDefaultsVersion(conf);/*from  w  w w .jav a  2 s .c  om*/
    return conf;
}