[
https://issues.apache.org/jira/browse/STORM-1199?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15065148#comment-15065148
]
ASF GitHub Bot commented on STORM-1199:
---------------------------------------
Github user roshannaik commented on a diff in the pull request:
https://github.com/apache/storm/pull/936#discussion_r48087261
--- Diff:
external/storm-hdfs/src/main/java/org/apache/storm/hdfs/spout/HdfsSpout.java ---
@@ -0,0 +1,654 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ * <p/>
+ * http://www.apache.org/licenses/LICENSE-2.0
+ * <p/>
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.storm.hdfs.spout;
+
+import java.io.IOException;
+import java.lang.reflect.Constructor;
+import java.util.Collection;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Timer;
+import java.util.TimerTask;
+import java.util.concurrent.LinkedBlockingQueue;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import backtype.storm.Config;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.fs.FileSystem;
+import org.apache.hadoop.fs.Path;
+import org.apache.storm.hdfs.common.HdfsUtils;
+import org.apache.storm.hdfs.common.security.HdfsSecurityUtil;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import backtype.storm.spout.SpoutOutputCollector;
+import backtype.storm.task.TopologyContext;
+import backtype.storm.topology.OutputFieldsDeclarer;
+import backtype.storm.topology.base.BaseRichSpout;
+import backtype.storm.tuple.Fields;
+
+public class HdfsSpout extends BaseRichSpout {
+
+ private static final Logger LOG =
LoggerFactory.getLogger(HdfsSpout.class);
+
+ private Path sourceDirPath;
+ private Path archiveDirPath;
+ private Path badFilesDirPath;
+ private Path lockDirPath;
+
+ private int commitFrequencyCount = Configs.DEFAULT_COMMIT_FREQ_COUNT;
+ private int commitFrequencySec = Configs.DEFAULT_COMMIT_FREQ_SEC;
+ private int maxDuplicates = Configs.DEFAULT_MAX_DUPLICATES;
+ private int lockTimeoutSec = Configs.DEFAULT_LOCK_TIMEOUT;
+ private boolean clocksInSync = true;
+
+ private ProgressTracker tracker = new ProgressTracker();
+
+ private FileSystem hdfs;
+ private FileReader reader;
+
+ private SpoutOutputCollector collector;
+ HashMap<MessageId, List<Object> > inflight = new HashMap<>();
+ LinkedBlockingQueue<HdfsUtils.Pair<MessageId, List<Object>>> retryList =
new LinkedBlockingQueue<>();
+
+ private String inprogress_suffix = ".inprogress";
+ private String ignoreSuffix = ".ignore";
+
+ private Configuration hdfsConfig;
+ private String readerType;
+
+ private Map conf = null;
+ private FileLock lock;
+ private String spoutId = null;
+
+ HdfsUtils.Pair<Path,FileLock.LogEntry> lastExpiredLock = null;
+ private long lastExpiredLockTime = 0;
+
+ private long tupleCounter = 0;
+ private boolean ackEnabled = false;
+ private int acksSinceLastCommit = 0 ;
+ private final AtomicBoolean commitTimeElapsed = new AtomicBoolean(false);
+ private final Timer commitTimer = new Timer();
+ private boolean fileReadCompletely = false;
+
+ private String configKey = Configs.DEFAULT_HDFS_CONFIG_KEY; // key for
hdfs kerberos configs
+
+ public HdfsSpout() {
+ }
+
+ public Path getLockDirPath() {
+ return lockDirPath;
+ }
+
+ public SpoutOutputCollector getCollector() {
+ return collector;
+ }
+
+ public HdfsSpout withConfigKey(String configKey){
+ this.configKey = configKey;
+ return this;
+ }
+
+ public void nextTuple() {
+ LOG.debug("Next Tuple");
+ // 1) First re-emit any previously failed tuples (from retryList)
+ if (!retryList.isEmpty()) {
+ LOG.debug("Sending from retry list");
+ HdfsUtils.Pair<MessageId, List<Object>> pair = retryList.remove();
+ emitData(pair.getValue(), pair.getKey());
+ return;
+ }
+
+ if( ackEnabled && tracker.size()>=maxDuplicates ) {
+ LOG.warn("Waiting for more ACKs before generating new tuples. " +
+ "Progress tracker size has reached limit {}"
+ , maxDuplicates);
+ // Don't emit anything .. allow configured spout wait strategy to
kick in
+ return;
+ }
+
+ // 2) If no failed tuples, then send tuples from hdfs
+ while (true) {
+ try {
+ // 3) Select a new file if one is not open already
+ if (reader == null) {
+ reader = pickNextFile();
+ if (reader == null) {
+ LOG.info("Currently no new files to process under : " +
sourceDirPath);
+ return;
+ }
+ }
+
+ // 4) Read record from file, emit to collector and record progress
+ List<Object> tuple = reader.next();
+ if (tuple != null) {
+ fileReadCompletely= false;
+ ++tupleCounter;
+ MessageId msgId = new MessageId(tupleCounter,
reader.getFilePath(), reader.getFileOffset());
+ emitData(tuple, msgId);
+
+ if(!ackEnabled) {
+ ++acksSinceLastCommit; // assume message is immediately acked
in non-ack mode
+ commitProgress(reader.getFileOffset());
+ } else {
+ commitProgress(tracker.getCommitPosition());
+ }
+ return;
+ } else {
+ fileReadCompletely = true;
+ if(!ackEnabled) {
+ markFileAsDone(reader.getFilePath());
+ }
+ }
+ } catch (IOException e) {
+ LOG.error("I/O Error processing at file location " +
getFileProgress(reader), e);
+ // don't emit anything .. allow configured spout wait strategy to
kick in
+ return;
+ } catch (ParseException e) {
+ LOG.error("Parsing error when processing at file location " +
getFileProgress(reader) +
+ ". Skipping remainder of file.", e);
+ markFileAsBad(reader.getFilePath());
+ // note: Unfortunately not emitting anything here due to parse
error
+ // will trigger the configured spout wait strategy which is
unnecessary
+ }
+ }
+
+ }
+
+ // will commit progress into lock file if commit threshold is reached
+ private void commitProgress(FileOffset position) {
+ if ( lock!=null && canCommitNow() ) {
+ try {
+ lock.heartbeat(position.toString());
+ acksSinceLastCommit = 0;
+ commitTimeElapsed.set(false);
+ setupCommitElapseTimer();
+ } catch (IOException e) {
+ LOG.error("Unable to commit progress Will retry later.", e);
+ }
+ }
+ }
+
+ private void setupCommitElapseTimer() {
+ if(commitFrequencySec<=0)
+ return;
+ TimerTask timerTask = new TimerTask() {
+ @Override
+ public void run() {
+ commitTimeElapsed.set(false);
+ }
+ };
+ commitTimer.schedule(timerTask, commitFrequencySec * 1000);
+ }
+
+
+ private static String getFileProgress(FileReader reader) {
+ return reader.getFilePath() + " " + reader.getFileOffset();
+ }
+
+ private void markFileAsDone(Path filePath) {
+ fileReadCompletely = false;
+ try {
+ renameCompletedFile(reader.getFilePath());
+ } catch (IOException e) {
+ LOG.error("Unable to archive completed file" + filePath, e);
+ }
+ unlockAndCloseReader();
+
+ }
+
+ private void markFileAsBad(Path file) {
+ String fileName = file.toString();
+ String fileNameMinusSuffix = fileName.substring(0,
fileName.indexOf(inprogress_suffix));
+ String originalName = new Path(fileNameMinusSuffix).getName();
+ Path newFile = new Path( badFilesDirPath + Path.SEPARATOR +
originalName);
+
+ LOG.info("Moving bad file to " + newFile);
+ try {
+ if (!hdfs.rename(file, newFile) ) { // seems this can fail by
returning false or throwing exception
+ throw new IOException("Move failed for bad file: " + file); //
convert false ret value to exception
+ }
+ } catch (IOException e) {
+ LOG.warn("Error moving bad file: " + file + ". to destination : " +
newFile);
+ }
+
+ unlockAndCloseReader();
+ }
+
+ private void unlockAndCloseReader() {
+ reader.close();
+ reader = null;
+ try {
+ lock.release();
+ } catch (IOException e) {
+ LOG.error("Unable to delete lock file : " + this.lock.getLockFile(),
e);
+ }
+ lock = null;
+ }
+
+
+
+ protected void emitData(List<Object> tuple, MessageId id) {
+ LOG.debug("Emitting - {}", id);
+ this.collector.emit(tuple, id);
+ inflight.put(id, tuple);
+ }
+
+ public void open(Map conf, TopologyContext context,
SpoutOutputCollector collector) {
+ this.conf = conf;
+ final String FILE_SYSTEM = "filesystem";
+ LOG.info("Opening");
+ this.collector = collector;
+ this.hdfsConfig = new Configuration();
+ this.tupleCounter = 0;
+
+ for( Object k : conf.keySet() ) {
+ String key = k.toString();
+ if( ! FILE_SYSTEM.equalsIgnoreCase( key ) ) { // to support unit
test only
+ String val = conf.get(key).toString();
+ LOG.info("Config setting : " + key + " = " + val);
+ this.hdfsConfig.set(key, val);
+ }
+ else
+ this.hdfs = (FileSystem) conf.get(key);
+
+ if(key.equalsIgnoreCase(Configs.READER_TYPE)) {
+ readerType = conf.get(key).toString();
+ checkValidReader(readerType);
+ }
+ }
+
+ // - Hdfs configs
+ this.hdfsConfig = new Configuration();
+ Map<String, Object> map = (Map<String,
Object>)conf.get(this.configKey);
+ if(map != null){
+ for(String key : map.keySet()){
+ this.hdfsConfig.set(key, String.valueOf(map.get(key)));
+ }
+ }
+
+ try {
+ HdfsSecurityUtil.login(conf, hdfsConfig);
+ } catch (IOException e) {
+ LOG.error("Failed to open " + sourceDirPath);
+ throw new RuntimeException(e);
+ }
+
+ // -- source dir config
+ if ( !conf.containsKey(Configs.SOURCE_DIR) ) {
+ LOG.error(Configs.SOURCE_DIR + " setting is required");
+ throw new RuntimeException(Configs.SOURCE_DIR + " setting is
required");
+ }
+ this.sourceDirPath = new Path( conf.get(Configs.SOURCE_DIR).toString()
);
+
+ // -- archive dir config
+ if ( !conf.containsKey(Configs.ARCHIVE_DIR) ) {
+ LOG.error(Configs.ARCHIVE_DIR + " setting is required");
+ throw new RuntimeException(Configs.ARCHIVE_DIR + " setting is
required");
+ }
+ this.archiveDirPath = new Path(
conf.get(Configs.ARCHIVE_DIR).toString() );
+
+ try {
+ if(hdfs.exists(archiveDirPath)) {
+ if(! hdfs.isDirectory(archiveDirPath) ) {
+ LOG.error("Archive directory is a file. " + archiveDirPath);
+ throw new RuntimeException("Archive directory is a file. " +
archiveDirPath);
+ }
+ } else if(! hdfs.mkdirs(archiveDirPath) ) {
+ LOG.error("Unable to create archive directory. " + archiveDirPath);
+ throw new RuntimeException("Unable to create archive directory " +
archiveDirPath);
+ }
+ } catch (IOException e) {
+ LOG.error("Unable to create archive dir ", e);
+ throw new RuntimeException("Unable to create archive directory ", e);
+ }
+
+ // -- bad files dir config
+ if ( !conf.containsKey(Configs.BAD_DIR) ) {
+ LOG.error(Configs.BAD_DIR + " setting is required");
+ throw new RuntimeException(Configs.BAD_DIR + " setting is required");
+ }
--- End diff --
will do.
> Create HDFS Spout
> -----------------
>
> Key: STORM-1199
> URL: https://issues.apache.org/jira/browse/STORM-1199
> Project: Apache Storm
> Issue Type: New Feature
> Reporter: Roshan Naik
> Assignee: Roshan Naik
> Attachments: HDFSSpoutforStorm v2.pdf, HDFSSpoutforStorm.pdf,
> hdfs-spout.1.patch
>
>
> Create an HDFS spout so that Storm can suck in data from files in a HDFS
> directory
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
