internal static void DStreamTextFileSamples() { count = 0; string directory = SparkCLRSamples.Configuration.SampleDataLocation; string checkpointPath = Path.Combine(directory, "checkpoint"); SparkContext sc = SparkCLRSamples.SparkContext; var b = sc.Broadcast <int>(0); StreamingContext ssc = StreamingContext.GetOrCreate(checkpointPath, () => { StreamingContext context = new StreamingContext(sc, 2000); context.Checkpoint(checkpointPath); var lines = context.TextFileStream(Path.Combine(directory, "test")); lines = context.Union(lines, lines); var words = lines.FlatMap(l => l.Split(' ')); var pairs = words.Map(w => new KeyValuePair <string, int>(w, 1)); // since operations like ReduceByKey, Join and UpdateStateByKey are // separate dstream transformations defined in CSharpDStream.scala // an extra CSharpRDD is introduced in between these operations var wordCounts = pairs.ReduceByKey((x, y) => x + y); var join = wordCounts.Join(wordCounts, 2); var state = join.UpdateStateByKey <string, Tuple <int, int>, int>(new UpdateStateHelper(b).Execute); state.ForeachRDD((time, rdd) => { // there's chance rdd.Take conflicts with ssc.Stop if (stopFileServer) { return; } object[] taken = rdd.Take(10); Console.WriteLine("-------------------------------------------"); Console.WriteLine("Time: {0}", time); Console.WriteLine("-------------------------------------------"); foreach (object record in taken) { Console.WriteLine(record); } Console.WriteLine(); stopFileServer = count++ > 100; }); return(context); }); ssc.Start(); StartFileServer(directory, "words.txt", 100); ssc.AwaitTermination(); ssc.Stop(); }
internal static void DStreamDirectKafkaWithRepartitionSample() { count = 0; string directory = SparkCLRSamples.Configuration.SampleDataLocation; string checkpointPath = Path.Combine(directory, "checkpoint"); StreamingContext ssc = StreamingContext.GetOrCreate(checkpointPath, () => { var conf = new SparkConf(); SparkContext sc = new SparkContext(conf); StreamingContext context = new StreamingContext(sc, 2000L); context.Checkpoint(checkpointPath); var kafkaParams = new List <Tuple <string, string> > { new Tuple <string, string>("metadata.broker.list", brokers), new Tuple <string, string>("auto.offset.reset", "smallest") }; conf.Set("spark.mobius.streaming.kafka.numPartitions." + topic, partitions.ToString()); var dstream = KafkaUtils.CreateDirectStream(context, new List <string> { topic }, kafkaParams, Enumerable.Empty <Tuple <string, long> >()); dstream.ForeachRDD((time, rdd) => { long batchCount = rdd.Count(); int numPartitions = rdd.GetNumPartitions(); Console.WriteLine("-------------------------------------------"); Console.WriteLine("Time: {0}", time); Console.WriteLine("-------------------------------------------"); Console.WriteLine("Count: " + batchCount); Console.WriteLine("Partitions: " + numPartitions); // only first batch has data and is repartitioned into 10 partitions if (count++ == 0) { Assert.AreEqual(messages, batchCount); Assert.IsTrue(numPartitions >= partitions); } else { Assert.AreEqual(0, batchCount); Assert.IsTrue(numPartitions == 0); } }); return(context); }); ssc.Start(); ssc.AwaitTermination(); }
static void Main(string[] args) { var sparkContext = new SparkContext(new SparkConf().SetAppName("SparkCLREventHub Example")); var eventhubsParams = new Dictionary <string, string>() { { "eventhubs.policyname", "<policyname>" }, { "eventhubs.policykey", "<policykey>" }, { "eventhubs.namespace", "<namespace>" }, { "eventhubs.name", "<name>" }, { "eventhubs.partition.count", "<partitioncount>" }, { "eventhubs.consumergroup", "$default" }, { "eventhubs.checkpoint.dir", "<hdfs path to eventhub checkpoint dir>" }, { "eventhubs.checkpoint.interval", "<interval>" }, }; const int windowDurationInSecs = 5; const int slideDurationInSecs = 5; const string checkpointPath = "<hdfs path to spark checkpoint dir>"; //const string outputPath = "<hdfs path to output dir>"; const long slideDurationInMillis = 5000; StreamingContext sparkStreamingContext = StreamingContext.GetOrCreate(checkpointPath, () => { var ssc = new StreamingContext(sparkContext, slideDurationInMillis); ssc.Checkpoint(checkpointPath); var stream = EventHubsUtils.CreateUnionStream(ssc, eventhubsParams.Select(v => new Tuple <string, string>(v.Key, v.Value))); var countByLogLevelAndTime = stream .Map(bytes => Encoding.UTF8.GetString(bytes)) .Filter(line => line.Contains(",")) .Map(line => line.Split(',')) .Map(columns => new Tuple <string, int>(string.Format("{0},{1}", columns[0], columns[1]), 1)) .ReduceByKeyAndWindow((x, y) => x + y, (x, y) => x - y, windowDurationInSecs, slideDurationInSecs, 3) .Map(logLevelCountPair => string.Format("{0},{1}", logLevelCountPair.Item1, logLevelCountPair.Item2)); countByLogLevelAndTime.ForeachRDD(countByLogLevel => { //dimensionalCount.SaveAsTextFile(string.Format("{0}/{1}", outputPath, Guid.NewGuid())); var dimensionalCountCollection = countByLogLevel.Collect(); foreach (var dimensionalCountItem in dimensionalCountCollection) { Console.WriteLine(dimensionalCountItem); } }); return(ssc); }); sparkStreamingContext.Start(); sparkStreamingContext.AwaitTermination(); }
static void Main(string[] args) { var sparkContext = new SparkContext(new SparkConf().SetAppName("SparkCLRKafka Example")); const string topicName = "<topicName>"; var topicList = new List <string> { topicName }; var kafkaParams = new Dictionary <string, string> //refer to http://kafka.apache.org/documentation.html#configuration { { "metadata.broker.list", "<kafka brokers list>" }, { "auto.offset.reset", "smallest" } }; var perTopicPartitionKafkaOffsets = new Dictionary <string, long>(); const int windowDurationInSecs = 5; const int slideDurationInSecs = 5; const string checkpointPath = "<hdfs path to spark checkpoint directory>"; const string appOutputPath = "<hdfs path to app output directory>"; const long slideDurationInMillis = 5000; StreamingContext sparkStreamingContext = StreamingContext.GetOrCreate(checkpointPath, () => { var ssc = new StreamingContext(sparkContext, slideDurationInMillis); ssc.Checkpoint(checkpointPath); var stream = KafkaUtils.CreateDirectStream(ssc, topicList, kafkaParams.Select(v => new Tuple <string, string>(v.Key, v.Value)), perTopicPartitionKafkaOffsets.Select(v => new Tuple <string, long>(v.Key, v.Value))); var countByLogLevelAndTime = stream .Map(tuple => Encoding.UTF8.GetString(tuple.Item2)) .Filter(line => line.Contains(",")) .Map(line => line.Split(',')) .Map(columns => new Tuple <string, int>(string.Format("{0},{1}", columns[0], columns[1]), 1)) .ReduceByKeyAndWindow((x, y) => x + y, (x, y) => x - y, windowDurationInSecs, slideDurationInSecs, 3) .Map(logLevelCountPair => string.Format("{0},{1}", logLevelCountPair.Item1, logLevelCountPair.Item2)); countByLogLevelAndTime.ForeachRDD(countByLogLevel => { countByLogLevel.SaveAsTextFile(string.Format("{0}/{1}", appOutputPath, Guid.NewGuid())); foreach (var logCount in countByLogLevel.Collect()) { Console.WriteLine(logCount); } }); return(ssc); }); sparkStreamingContext.Start(); sparkStreamingContext.AwaitTermination(); }
public static void Process(string AppName, string CheckpointPath, Dictionary <string, string> kafkaParams) { var sparkContext = new SparkContext(new SparkConf().SetAppName(AppName)); var topicList = new List <string> { kafkaParams["topic"] }; var perTopicPartitionKafkaOffsets = new Dictionary <string, long>(); const long slideDurationInMillis = 1000; StreamingContext sparkStreamingContext = StreamingContext.GetOrCreate(CheckpointPath, () => { var ssc = new StreamingContext(sparkContext, slideDurationInMillis); var stream = KafkaUtils.CreateDirectStream(ssc, topicList, kafkaParams, perTopicPartitionKafkaOffsets); stream.Map(kvp => { if (kvp.Value != null) { return(Encoding.UTF8.GetString(kvp.Value)); } else { return(null); } } ).ForeachRDD(RDD => { foreach (string line in RDD.Collect()) { var message = JObject.Parse(line); var _id = message.SelectToken("docid").ToString(); // ======================= // TODO: Process message // ======================= } } ); ssc.Checkpoint(CheckpointPath); return(ssc); }); sparkStreamingContext.Start(); sparkStreamingContext.AwaitTermination(); }
static void Main(string[] args) { if (args.Length < 2) { Console.WriteLine("Usage: HdfsWordCount <checkpointDirectory> <inputDirectory>"); return; } string checkpointPath = args[0]; string inputDir = args[1]; StreamingContext ssc = StreamingContext.GetOrCreate(checkpointPath, () => { var sparkConf = new SparkConf(); sparkConf.SetAppName("HdfsWordCount"); var sc = new SparkContext(sparkConf); StreamingContext context = new StreamingContext(sc, 30000); context.Checkpoint(checkpointPath); var lines = context.TextFileStream(inputDir); var words = lines.FlatMap(l => l.Split(' ')); var pairs = words.Map(w => new KeyValuePair <string, int>(w, 1)); var wordCounts = pairs.ReduceByKey((x, y) => x + y); wordCounts.ForeachRDD((time, rdd) => { Console.WriteLine("-------------------------------------------"); Console.WriteLine("Time: {0}", time); Console.WriteLine("-------------------------------------------"); object[] taken = rdd.Take(10); foreach (object record in taken) { Console.WriteLine(record); } Console.WriteLine(); }); return(context); }); ssc.Start(); ssc.AwaitTermination(); ssc.Stop(); }
public override void Run(Lazy <SparkContext> sparkContext, int currentTimes, int totalTimes) { DeleteCheckPointDirectory(currentTimes); var options = Options as UnionTopicTestOptions; var streamingContext = StreamingContext.GetOrCreate(options.CheckPointDirectory, () => { var ssc = new StreamingContext(sparkContext.Value, options.BatchSeconds * 1000L); ssc.Checkpoint(options.CheckPointDirectory); var stream1 = KafkaUtils.CreateDirectStream(ssc, new List <string> { options.Topic1 }, kafkaParams, offsetsRange) .Map(line => new RowIdCountTime().Deserialize(line.Value)); var stream2 = KafkaUtils.CreateDirectStream(ssc, new List <string> { options.Topic2 }, kafkaParams, offsetsRange) .Map(line => new RowIdCountTime().Deserialize(line.Value)); var stream = stream1.Union(stream2); if (options.RePartition > 0) { stream = stream.Repartition(options.RePartition); } stream.ForeachRDD(rdd => { rdd.Foreach(idCount => { Console.WriteLine($"{NowMilli} {this.GetType().Name} : {idCount.ToString()}"); }); }); SaveStreamToFile(stream.Map(it => it.ToString())); return(ssc); }); streamingContext.Start(); WaitTerminationOrTimeout(streamingContext); }
public override void Run(Lazy <SparkContext> sparkContext, int currentTimes, int totalTimes) { DeleteCheckPointDirectory(currentTimes); var options = Options as WindowSlideTestOptions; var allBeginTime = DateTime.Now; var topicList = new List <string>(options.Topics.Split(";,".ToArray())); ParseKafkaParameters(); for (var k = 0; options.TestTimes <= 0 || k < options.TestTimes; k++) { var beginTime = DateTime.Now; //Logger.LogInfo("begin test[{0}]-{1} , sparkContext = {2}", k + 1, options.TestTimes > 0 ? options.TestTimes.ToString() : "infinite", sparkContext.Value); var streamingContext = StreamingContext.GetOrCreate(options.CheckPointDirectory, () => { var ssc = new StreamingContext(sparkContext.Value, options.BatchSeconds * 1000L); ssc.Checkpoint(options.CheckPointDirectory); var stream = KafkaUtils.CreateDirectStream(ssc, topicList, kafkaParams, offsetsRange) .Map(line => Encoding.UTF8.GetString(line.Value)); var pairs = stream.Map(new ParseKeyValueArray(options.ElementCount, options.ShowReceivedLines).Parse); var reducedStream = pairs.ReduceByKeyAndWindow( new ReduceHelper(options.CheckArrayAtFirst).Sum, new ReduceHelper(options.CheckArrayAtFirst).InverseSum, options.WindowSeconds, options.SlideSeconds ); reducedStream.ForeachRDD(new SumCountStatic().ForeachRDD <int[]>); SaveStreamToFile(reducedStream); return(ssc); }); streamingContext.Start(); WaitTerminationOrTimeout(streamingContext); } }
static void Main(string[] args) { var checkpointPath = ""; var sparkContext = new SparkContext(new SparkConf()); var slideDurationInMillis = 10; var topics = new List <string>(); var kafkaParams = new List <Tuple <string, string> >(); var perTopicPartitionKafkaOffsets = new List <Tuple <string, long> >(); var windowDurationInSecs = 10; var slideDurationInSecs = 10; StreamingContext sparkStreamingContext = StreamingContext.GetOrCreate(checkpointPath, () => { var ssc = new StreamingContext(sparkContext, slideDurationInMillis); ssc.Checkpoint(checkpointPath); var stream = KafkaUtils.CreateDirectStream(ssc, topics, kafkaParams, perTopicPartitionKafkaOffsets); var countByLogLevelAndTime = stream .Map(kvp => Encoding.UTF8.GetString(kvp.Item2)) .Filter(line => line.Contains(",")) .Map(line => line.Split(',')) .Map(columns => new Tuple <string, int>( string.Format("{0},{1}", columns[0], columns[1]), 1)) .ReduceByKeyAndWindow((x, y) => x + y, (x, y) => x - y, windowDurationInSecs, slideDurationInSecs, 3) .Map(logLevelCountPair => string.Format("{0},{1}", logLevelCountPair.Item1, logLevelCountPair.Item2)); countByLogLevelAndTime.ForeachRDD(countByLogLevel => { foreach (var logCount in countByLogLevel.Collect()) { Console.WriteLine(logCount); } }); return(ssc); }); sparkStreamingContext.Start(); sparkStreamingContext.AwaitTermination(); Console.WriteLine("Hello World!"); }
internal static void DStreamTextFileSample() { count = 0; string directory = SparkCLRSamples.Configuration.SampleDataLocation; string checkpointPath = Path.Combine(directory, "checkpoint"); SparkContext sc = SparkCLRSamples.SparkContext; var b = sc.Broadcast <int>(0); StreamingContext ssc = StreamingContext.GetOrCreate(checkpointPath, () => { StreamingContext context = new StreamingContext(sc, 2000L); // batch interval is in milliseconds context.Checkpoint(checkpointPath); var lines = context.TextFileStream(Path.Combine(directory, "test")); lines = context.Union(lines, lines); var words = lines.FlatMap(l => l.Split(' ')); var pairs = words.Map(w => new Tuple <string, int>(w, 1)); // since operations like ReduceByKey, Join and UpdateStateByKey are // separate dstream transformations defined in CSharpDStream.scala // an extra CSharpRDD is introduced in between these operations var wordCounts = pairs.ReduceByKey((x, y) => x + y); var join = wordCounts.Window(2, 2).Join(wordCounts, 2); var initialStateRdd = sc.Parallelize(new[] { new Tuple <string, int>("AAA", 88), new Tuple <string, int>("BBB", 88) }); var state = join.UpdateStateByKey(new UpdateStateHelper(b).Execute, initialStateRdd); state.ForeachRDD((time, rdd) => { // there's chance rdd.Take conflicts with ssc.Stop if (stopFileServer) { return; } object[] taken = rdd.Take(10); Console.WriteLine("-------------------------------------------"); Console.WriteLine("Time: {0}", time); Console.WriteLine("-------------------------------------------"); foreach (object record in taken) { Console.WriteLine(record); var countByWord = (Tuple <string, int>)record; Assert.AreEqual(countByWord.Item2, countByWord.Item1 == "The" || countByWord.Item1 == "lazy" || countByWord.Item1 == "dog" ? 92 : 88); } Console.WriteLine(); stopFileServer = true; }); return(context); }); StartFileServer(ssc, directory, "words.txt"); ssc.Start(); ssc.AwaitTermination(); }
internal static void DStreamMapWithStateSample() { string directory = SparkCLRSamples.Configuration.SampleDataLocation; string checkpointPath = Path.Combine(directory, "checkpoint"); StreamingContext ssc = StreamingContext.GetOrCreate(checkpointPath, () => { SparkContext sc = SparkCLRSamples.SparkContext; StreamingContext context = new StreamingContext(sc, 10000L); // batch interval is in milliseconds context.Checkpoint(checkpointPath); var lines = context.TextFileStream(Path.Combine(directory, "test1")); lines = context.Union(lines, lines); var words = lines.FlatMap(l => l.Split(' ')); var pairs = words.Map(w => new KeyValuePair <string, int>(w, 1)); var wordCounts = pairs.ReduceByKey((x, y) => x + y); var initialState = sc.Parallelize(new[] { new KeyValuePair <string, int>("NOT_A_WORD", 1024), new KeyValuePair <string, int>("dog", 10000), }, 1); var stateSpec = new StateSpec <string, int, int, KeyValuePair <string, int> >((word, count, state) => { if (state.IsTimingOut()) { Console.WriteLine("Found timing out word: {0}", word); return(new KeyValuePair <string, int>(word, state.Get())); } var sum = 0; if (state.Exists()) { sum = state.Get(); } state.Update(sum + count); Console.WriteLine("word: {0}, count: {1}", word, sum + count); return(new KeyValuePair <string, int>(word, sum + count)); }).NumPartitions(1).InitialState(initialState).Timeout(TimeSpan.FromSeconds(30)); var snapshots = wordCounts.MapWithState(stateSpec).StateSnapshots(); snapshots.ForeachRDD((double time, RDD <dynamic> rdd) => { Console.WriteLine("-------------------------------------------"); Console.WriteLine("Snapshots @ Time: {0}", time); Console.WriteLine("-------------------------------------------"); foreach (KeyValuePair <string, int> record in rdd.Collect()) { Console.WriteLine("[{0}, {1}]", record.Key, record.Value); } Console.WriteLine(); }); return(context); }); ssc.Start(); StartFileServer(directory, "words.txt", 100); ssc.AwaitTermination(); ssc.Stop(); }