List of usage examples for org.apache.hadoop.conf Configuration get
public String get(String name, String defaultValue)
name. From source file:com.asakusafw.runtime.stage.input.StageInputFormat.java
License:Apache License
/** * Returns the {@link SplitCombiner} class used in the current job. * @param context the current job context * @return the {@link SplitCombiner} class * @since 0.7.1//from w w w . ja v a2 s. com */ public static Class<? extends SplitCombiner> getSplitCombinerClass(JobContext context) { if (context == null) { throw new IllegalArgumentException("context must not be null"); //$NON-NLS-1$ } Configuration conf = context.getConfiguration(); String combinerType = conf.get(KEY_SPLIT_COMBINER, DEFAULT_SPLIT_COMBINER); if (JobCompatibility.isLocalMode(context) && combinerType.equals(DEFAULT_SPLIT_COMBINER)) { return ExtremeSplitCombiner.class; } Class<? extends SplitCombiner> defined = SPLIT_COMBINERS.get(combinerType); if (defined != null) { return defined; } try { return conf.getClassByName(combinerType).asSubclass(SplitCombiner.class); } catch (Exception e) { LOG.warn(MessageFormat.format("failed to load a combiner \"{0}\"", combinerType), e); return IdentitySplitCombiner.class; } }
From source file:com.asakusafw.runtime.util.hadoop.ConfigurationProviderTest.java
License:Apache License
private boolean isLoaded(Configuration c) { return c.get("testing.conf", "not added").equals("added"); }
From source file:com.bah.lucene.BlockCacheDirectoryFactoryV2.java
License:Apache License
public BlockCacheDirectoryFactoryV2(Configuration configuration, long totalNumberOfBytes) { final int fileBufferSizeInt = configuration.getInt(BLUR_SHARD_BLOCK_CACHE_V2_FILE_BUFFER_SIZE, 8192); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_FILE_BUFFER_SIZE, fileBufferSizeInt)); final int cacheBlockSizeInt = configuration.getInt(BLUR_SHARD_BLOCK_CACHE_V2_CACHE_BLOCK_SIZE, 8192); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_CACHE_BLOCK_SIZE, cacheBlockSizeInt)); final Map<String, Integer> cacheBlockSizeMap = new HashMap<String, Integer>(); for (Entry<String, String> prop : configuration) { String key = prop.getKey(); if (key.startsWith(BLUR_SHARD_BLOCK_CACHE_V2_CACHE_BLOCK_SIZE_PREFIX)) { String value = prop.getValue(); int cacheBlockSizeForFile = Integer.parseInt(value); String fieldType = key.substring(BLUR_SHARD_BLOCK_CACHE_V2_CACHE_BLOCK_SIZE_PREFIX.length()); cacheBlockSizeMap.put(fieldType, cacheBlockSizeForFile); LOG.info(/*from w w w . j a v a 2 s .co m*/ MessageFormat.format("{0}={1} for file type [{2}]", key, cacheBlockSizeForFile, fieldType)); } } final STORE store = STORE.valueOf(configuration.get(BLUR_SHARD_BLOCK_CACHE_V2_STORE, OFF_HEAP)); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_STORE, store)); final Set<String> cachingFileExtensionsForRead = getSet( configuration.get(BLUR_SHARD_BLOCK_CACHE_V2_READ_CACHE_EXT, DEFAULT_VALUE)); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_READ_CACHE_EXT, cachingFileExtensionsForRead)); final Set<String> nonCachingFileExtensionsForRead = getSet( configuration.get(BLUR_SHARD_BLOCK_CACHE_V2_READ_NOCACHE_EXT, DEFAULT_VALUE)); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_READ_NOCACHE_EXT, nonCachingFileExtensionsForRead)); final boolean defaultReadCaching = configuration.getBoolean(BLUR_SHARD_BLOCK_CACHE_V2_READ_DEFAULT, true); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_READ_DEFAULT, defaultReadCaching)); final Set<String> cachingFileExtensionsForWrite = getSet( configuration.get(BLUR_SHARD_BLOCK_CACHE_V2_WRITE_CACHE_EXT, DEFAULT_VALUE)); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_WRITE_CACHE_EXT, cachingFileExtensionsForWrite)); final Set<String> nonCachingFileExtensionsForWrite = getSet( configuration.get(BLUR_SHARD_BLOCK_CACHE_V2_WRITE_NOCACHE_EXT, DEFAULT_VALUE)); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_WRITE_NOCACHE_EXT, nonCachingFileExtensionsForWrite)); final boolean defaultWriteCaching = configuration.getBoolean(BLUR_SHARD_BLOCK_CACHE_V2_WRITE_DEFAULT, true); LOG.info(MessageFormat.format("{0}={1}", BLUR_SHARD_BLOCK_CACHE_V2_WRITE_DEFAULT, defaultWriteCaching)); Size fileBufferSize = new Size() { @Override public int getSize(CacheDirectory directory, String fileName) { return fileBufferSizeInt; } }; Size cacheBlockSize = new Size() { @Override public int getSize(CacheDirectory directory, String fileName) { String ext = getExt(fileName); Integer size = cacheBlockSizeMap.get(ext); if (size != null) { return size; } return cacheBlockSizeInt; } }; FileNameFilter readFilter = new FileNameFilter() { @Override public boolean accept(CacheDirectory directory, String fileName) { String ext = getExt(fileName); if (cachingFileExtensionsForRead.contains(ext)) { return true; } else if (nonCachingFileExtensionsForRead.contains(ext)) { return false; } return defaultReadCaching; } }; FileNameFilter writeFilter = new FileNameFilter() { @Override public boolean accept(CacheDirectory directory, String fileName) { String ext = getExt(fileName); if (cachingFileExtensionsForWrite.contains(ext)) { return true; } else if (nonCachingFileExtensionsForWrite.contains(ext)) { return false; } return defaultWriteCaching; } }; Quiet quiet = new Quiet() { @Override public boolean shouldBeQuiet(CacheDirectory directory, String fileName) { Thread thread = Thread.currentThread(); String name = thread.getName(); if (name.startsWith(SHARED_MERGE_SCHEDULER)) { return true; } return false; } }; _cache = new BaseCache(totalNumberOfBytes, fileBufferSize, cacheBlockSize, readFilter, writeFilter, quiet, store); }
From source file:com.basho.riak.hadoop.config.RiakConfig.java
License:Apache License
/** * Get all the riak locations from the passed {@link Configuration} * /* www. j av a2s. c om*/ * @param conf * the {@link Configuration} * @return an array of {@link RiakLocation} (may be empty, never null) */ public static RiakLocation[] getRiakLocatons(Configuration conf) { String locations = conf.get(LOCATIONS_PROPERTY, ""); StringTokenizer st = new StringTokenizer(locations, COMMA); List<RiakLocation> result = new ArrayList<RiakLocation>(); while (st.hasMoreTokens()) { result.add(RiakLocation.fromString(st.nextToken())); } return result.toArray(new RiakLocation[result.size()]); }
From source file:com.baynote.kafka.hadoop.KafkaInputFormat.java
License:Apache License
/** * Gets the Zookeeper root of Kafka set by {@link #setZkRoot(Job, String)}, defaulting to {@link #DEFAULT_ZK_ROOT} * if it has not been set./*from ww w . j a v a 2s.co m*/ * * @param conf * the job conf. * @return the Zookeeper root of Kafka. */ public static String getZkRoot(final Configuration conf) { return conf.get("kafka.zk.root", DEFAULT_ZK_ROOT); }
From source file:com.bigstep.datalake.DLFileSystem.java
License:Apache License
@Override public synchronized void initialize(URI uri, Configuration conf) throws IOException { super.initialize(uri, conf); uri = selectDatalakeEndpointURI(uri, conf); /* set user pattern based on configuration file */ UserParam.setUserPattern(conf.get(DFSConfigKeys.DFS_WEBHDFS_USER_PATTERN_KEY, DFSConfigKeys.DFS_WEBHDFS_USER_PATTERN_DEFAULT)); kerberosIdentity = initialiseKerberosIdentity(conf); this.shouldUseEncryption = conf.getBoolean(FS_DL_IMPL_SHOULD_USE_ENCRYPTION_CONFIG_NAME, false); if (this.shouldUseEncryption) { initialiseAesEncryption(conf);/*www . ja va2 s .c om*/ } this.homeDirectory = conf.get(FS_DL_IMPL_HOME_DIRECTORY); if (homeDirectory == null) throw new IOException( "The Datalake requires a home directory to be configured in the fs.dl.impl.homeDirectory configuration variable. This is in the form /data_lake/dlxxxx"); this.defaultEndpoint = conf.get(FS_DL_IMPL_DEFAULT_ENDPOINT); if (defaultEndpoint == null) throw new IOException( "The Datalake requires a default endpoint to be configured the fs.dl.impl.defaultEndpoint configuration variable. This is in the form /data_lake/dlxxxx"); URI defaultEndpointURI = URI.create(defaultEndpoint); String authority = uri.getAuthority() == null ? defaultEndpointURI.getAuthority() : uri.getAuthority(); this.baseUri = URI.create(uri.getScheme() + "://" + authority + this.homeDirectory); this.nnAddrs = resolveNNAddr(); LOG.debug("Created kerberosIdentity " + kerberosIdentity + " for " + this.baseUri); boolean isHA = HAUtil.isClientFailoverConfigured(conf, this.baseUri); boolean isLogicalUri = isHA && HAUtil.isLogicalUri(conf, this.baseUri); // In non-HA or non-logical URI case, the code needs to call // getCanonicalUri() in order to handle the case where no port is // specified in the URI this.tokenServiceName = isLogicalUri ? HAUtil.buildTokenServiceForLogicalUri(this.baseUri, getScheme()) : SecurityUtil.buildTokenService(getCanonicalUri()); if (!isHA) { this.retryPolicy = RetryUtils.getDefaultRetryPolicy(conf, DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_POLICY_ENABLED_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_POLICY_ENABLED_DEFAULT, DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_POLICY_SPEC_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_POLICY_SPEC_DEFAULT, SafeModeException.class); } else { int maxFailoverAttempts = conf.getInt(DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_MAX_ATTEMPTS_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_MAX_ATTEMPTS_DEFAULT); int maxRetryAttempts = conf.getInt(DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_MAX_ATTEMPTS_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_RETRY_MAX_ATTEMPTS_DEFAULT); int failoverSleepBaseMillis = conf.getInt(DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_SLEEPTIME_BASE_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_SLEEPTIME_BASE_DEFAULT); int failoverSleepMaxMillis = conf.getInt(DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_SLEEPTIME_MAX_KEY, DFSConfigKeys.DFS_HTTP_CLIENT_FAILOVER_SLEEPTIME_MAX_DEFAULT); this.retryPolicy = RetryPolicies.failoverOnNetworkException(RetryPolicies.TRY_ONCE_THEN_FAIL, maxFailoverAttempts, maxRetryAttempts, failoverSleepBaseMillis, failoverSleepMaxMillis); } this.workingDir = getHomeDirectory(); //Delegation tokens don't work with httpfs this.canRefreshDelegationToken = false; this.disallowFallbackToInsecureCluster = !conf.getBoolean( CommonConfigurationKeys.IPC_CLIENT_FALLBACK_TO_SIMPLE_AUTH_ALLOWED_KEY, CommonConfigurationKeys.IPC_CLIENT_FALLBACK_TO_SIMPLE_AUTH_ALLOWED_DEFAULT); this.delegationToken = null; this.defaultFilePermissions = Short .decode(conf.get(FS_DL_IMPL_DEFAULT_FILE_PERMISSIONS, this.DEFAULT_FILE_PERMISSIONS)); this.defaultUMask = Short.decode(conf.get(FS_DL_IMPL_DEFAULT_UMASK, this.DEFAULT_UMASK)); this.transportScheme = conf.get(FS_DL_IMPL_TRANSPORT_SCHEME_CONFIG_NAME, FS_DL_IMPL_DEFAULT_TRANSPORT_SCHEME); if (!checkJCE()) throw new IOException(JCE_ERROR); }
From source file:com.blackberry.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 w w . j ava2 s . com // 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 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 maxConcurrentMR = getConfOrEnv(conf, "MAX_CONCURRENT_MR", "-1"); String zkConnectString = getConfOrEnv(conf, "ZK_CONNECT_STRING"); String logdir = getConfOrEnv(conf, "logdriver.logdir.name"); boolean resetOrphanedJobs = Boolean.parseBoolean(getConfOrEnv(conf, "reset.orphaned.jobs", "true")); String rootDir = getConfOrEnv(conf, "service.root.dir"); String maxTotalMR = getConfOrEnv(conf, "MAX_TOTAL_MR", "-1"); boolean doMerge = true; boolean doArchive = true; boolean doDelete = true; 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 (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"; } // We can hang if this fails. So make sure we abort if it fails. 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; } // Create the LockUtil instance lockUtil = new LockUtil(zkConnectString); // Now it's safe to create our Job Runner JobRunner jobRunner = new JobRunner(Integer.parseInt(maxConcurrentMR), Integer.parseInt(maxTotalMR)); Thread jobRunnerThread = new Thread(jobRunner); jobRunnerThread.setName("JobRunner"); jobRunnerThread.setDaemon(false); jobRunnerThread.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 { // 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.isDirectory() && 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.isDirectory() && 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 jobProps = new Properties(); jobProps.load(new FileInputStream(mergeJobPropertiesFile)); jobProps.setProperty("jobType", "merge"); jobProps.setProperty("rootDir", rootDir); jobProps.setProperty("dcNumber", dcNumber); jobProps.setProperty("service", service); jobProps.setProperty("date", matchDate); jobProps.setProperty("hour", matchHour); jobProps.setProperty("component", matchComponent); jobProps.setProperty("user.name", userName); jobProps.setProperty("logdir", logdir); jobRunner.submit(jobProps); 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 jobProps = new Properties(); jobProps.load(new FileInputStream(filterJobPropertiesFile)); jobProps.setProperty("jobType", "filter"); jobProps.setProperty("rootDir", rootDir); jobProps.setProperty("dcNumber", dcNumber); jobProps.setProperty("service", service); jobProps.setProperty("date", matchDate); jobProps.setProperty("hour", matchHour); jobProps.setProperty("component", matchComponent); jobProps.setProperty("user.name", userName); jobProps.setProperty("logdir", logdir); // Check to see if we should just keep all or delete all here. // The filter file should be here String appPath = jobProps.getProperty("oozie.wf.application.path"); appPath = appPath.replaceFirst("\\$\\{.*?\\}", ""); Path filterFile = new Path( appPath + "/" + conf.get("filter.definition.file", 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/"; PathInfo pathInfo = new PathInfo(); pathInfo.setDcNumber(dcNumber); pathInfo.setService(service); pathInfo.setLogdir(logdir); pathInfo.setDate(matchDate); pathInfo.setHour(matchHour); pathInfo.setComponent(matchComponent); try { lockUtil.acquireWriteLock(lockUtil.getLockPath(pathInfo)); fs.mkdirs(new Path(destination)); for (FileStatus f : fs.listStatus(p)) { fs.rename(f.getPath(), new Path(destination)); } } finally { lockUtil.releaseWriteLock(lockUtil.getLockPath(pathInfo)); } } else if (filters.size() == 1 && filters.get(0) instanceof BoomFilterMapper.DropAllFilter) { LOG.info(" Dropping everything. {} :: {} {} {} {} {}", new Object[] { p.toString(), dcNumber, service, matchDate, matchHour, matchComponent }); PathInfo pathInfo = new PathInfo(); pathInfo.setDcNumber(dcNumber); pathInfo.setService(service); pathInfo.setLogdir(logdir); pathInfo.setDate(matchDate); pathInfo.setHour(matchHour); pathInfo.setComponent(matchComponent); try { lockUtil.acquireWriteLock(lockUtil.getLockPath(pathInfo)); fs.delete(p, true); } finally { lockUtil.releaseWriteLock(lockUtil.getLockPath(pathInfo)); } } else { LOG.info(" Run Filter/Archive job {} :: {} {} {} {} {}", new Object[] { p.toString(), dcNumber, service, matchDate, matchHour, matchComponent }); jobRunner.submit(jobProps); } } 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 ({})", p); if (resetOrphanedJobs) { String matchDate = matcher.group(1); String matchHour = matcher.group(2); String matchComponent = matcher.group(3); // Move everything from working/xxx/incoming/ to incoming/ PathInfo lockPathInfo = new PathInfo(logdir, rootDir + "/" + dcNumber + "/" + service + "/" + logdir + "/" + matchDate + "/" + matchHour + "/" + matchComponent); lockUtil.acquireWriteLock(lockUtil.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()); fs.delete(fileStatus.getPath(), true); } lockUtil.releaseWriteLock(lockUtil.getLockPath(lockPathInfo)); fs.delete(new Path(p.toUri().getPath()), true); } } 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.isDirectory()) { 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 = lockUtil.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 = lockUtil.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('/')); } } } } // Now that we're done, wait for the Oozie Runner to stop, and print the // results. LOG.info("Waiting for Oozie jobs to complete."); jobRunner.shutdown(); jobRunnerThread.join(); LOG.info("Job Stats : Started={} Succeeded={} failed={} errors={}", new Object[] { jobRunner.getStarted(), jobRunner.getSucceeded(), jobRunner.getFailed(), jobRunner.getErrors() }); lockUtil.close(); } catch (Exception e) { LOG.error("Unexpected exception caught.", e); return 1; } return 0; }
From source file:com.blackberry.logdriver.admin.LogMaintenance.java
License:Apache License
private String getConfOrEnv(Configuration conf, String propertyOrEnv) { String property = propertyOrEnv.toLowerCase().replaceAll("_", "."); String env = propertyOrEnv.toUpperCase().replaceAll("\\.", "_"); LOG.debug("Checking {}/{}", property, env); String result = conf.get(property, System.getenv(env)); LOG.info("Option {}/{} = {}", new Object[] { property, env, result }); return result; }
From source file:com.blackberry.logdriver.fs.FileManager.java
License:Apache License
public FileManager(Configuration conf) throws Exception { this.lockUtil = new LockUtil(conf); this.fs = FileSystem.get(conf); // the logdir can be set by a java property. logdir = conf.get("logdriver.logdir.name", "logs"); }
From source file:com.blackberry.logdriver.mapreduce.BinaryRecordWriter.java
License:Apache License
/** * Create a writer for the given BinaryOutputFormat and TaskAttemptContext. * /*from w ww . j a v a2 s . c o m*/ * @param outputFormat * @param context */ public BinaryRecordWriter(BinaryOutputFormat outputFormat, TaskAttemptContext context) { Configuration conf = context.getConfiguration(); String extension = conf.get("output.file.extension", ""); try { Path outputPath = outputFormat.getDefaultWorkFile(context, extension); FileSystem fs = FileSystem.get(conf); LOG.info("Creating output path: {}", outputPath); out = fs.create(outputPath, true); } catch (IOException e) { LOG.error("Error creating output file.", e); } }