/**
* 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
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* 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 storm.starter.bolt;
import backtype.storm.Config;
import backtype.storm.task.OutputCollector;
import backtype.storm.task.TopologyContext;
import backtype.storm.topology.OutputFieldsDeclarer;
import backtype.storm.topology.base.BaseRichBolt;
import backtype.storm.tuple.Fields;
import backtype.storm.tuple.Tuple;
import backtype.storm.tuple.Values;
import org.apache.log4j.Logger;
import storm.starter.tools.NthLastModifiedTimeTracker;
import storm.starter.tools.SlidingWindowCounter;
import storm.starter.util.TupleHelpers;
import java.util.HashMap;
import java.util.Map;
import java.util.Map.Entry;
/**
* This bolt performs rolling counts of incoming objects, i.e. sliding window based counting.
* <p/>
* The bolt is configured by two parameters, the length of the sliding window in seconds (which influences the output
* data of the bolt, i.e. how it will count objects) and the emit frequency in seconds (which influences how often the
* bolt will output the latest window counts). For instance, if the window length is set to an equivalent of five
* minutes and the emit frequency to one minute, then the bolt will output the latest five-minute sliding window every
* minute.
* <p/>
* The bolt emits a rolling count tuple per object, consisting of the object itself, its latest rolling count, and the
* actual duration of the sliding window. The latter is included in case the expected sliding window length (as
* configured by the user) is different from the actual length, e.g. due to high system load. Note that the actual
* window length is tracked and calculated for the window, and not individually for each object within a window.
* <p/>
* Note: During the startup phase you will usually observe that the bolt warns you about the actual sliding window
* length being smaller than the expected length. This behavior is expected and is caused by the way the sliding window
* counts are initially "loaded up". You can safely ignore this warning during startup (e.g. you will see this warning
* during the first ~ five minutes of startup time if the window length is set to five minutes).
*/
public class RollingCountBolt extends BaseRichBolt {
private static final long serialVersionUID = 5537727428628598519L;
private static final Logger LOG = Logger.getLogger(RollingCountBolt.class);
private static final int NUM_WINDOW_CHUNKS = 5;
private static final int DEFAULT_SLIDING_WINDOW_IN_SECONDS = NUM_WINDOW_CHUNKS * 60;
private static final int DEFAULT_EMIT_FREQUENCY_IN_SECONDS = DEFAULT_SLIDING_WINDOW_IN_SECONDS / NUM_WINDOW_CHUNKS;
private static final String WINDOW_LENGTH_WARNING_TEMPLATE =
"Actual window length is %d seconds when it should be %d seconds"
+ " (you can safely ignore this warning during the startup phase)";
private final SlidingWindowCounter<Object> counter;
private final int windowLengthInSeconds;
private final int emitFrequencyInSeconds;
private OutputCollector collector;
private NthLastModifiedTimeTracker lastModifiedTracker;
public RollingCountBolt() {
this(DEFAULT_SLIDING_WINDOW_IN_SECONDS, DEFAULT_EMIT_FREQUENCY_IN_SECONDS);
}
public RollingCountBolt(int windowLengthInSeconds, int emitFrequencyInSeconds) {
this.windowLengthInSeconds = windowLengthInSeconds;
this.emitFrequencyInSeconds = emitFrequencyInSeconds;
counter = new SlidingWindowCounter<Object>(deriveNumWindowChunksFrom(this.windowLengthInSeconds,
this.emitFrequencyInSeconds));
}
private int deriveNumWindowChunksFrom(int windowLengthInSeconds, int windowUpdateFrequencyInSeconds) {
return windowLengthInSeconds / windowUpdateFrequencyInSeconds;
}
@SuppressWarnings("rawtypes")
@Override
public void prepare(Map stormConf, TopologyContext context, OutputCollector collector) {
this.collector = collector;
lastModifiedTracker = new NthLastModifiedTimeTracker(deriveNumWindowChunksFrom(this.windowLengthInSeconds,
this.emitFrequencyInSeconds));
}
@Override
public void execute(Tuple tuple) {
if (TupleHelpers.isTickTuple(tuple)) {
LOG.debug("Received tick tuple, triggering emit of current window counts");
emitCurrentWindowCounts();
}
else {
countObjAndAck(tuple);
}
}
private void emitCurrentWindowCounts() {
Map<Object, Long> counts = counter.getCountsThenAdvanceWindow();
int actualWindowLengthInSeconds = lastModifiedTracker.secondsSinceOldestModification();
lastModifiedTracker.markAsModified();
if (actualWindowLengthInSeconds != windowLengthInSeconds) {
LOG.warn(String.format(WINDOW_LENGTH_WARNING_TEMPLATE, actualWindowLengthInSeconds, windowLengthInSeconds));
}
emit(counts, actualWindowLengthInSeconds);
}
private void emit(Map<Object, Long> counts, int actualWindowLengthInSeconds) {
for (Entry<Object, Long> entry : counts.entrySet()) {
Object obj = entry.getKey();
Long count = entry.getValue();
collector.emit(new Values(obj, count, actualWindowLengthInSeconds));
}
}
private void countObjAndAck(Tuple tuple) {
Object obj = tuple.getValue(0);
counter.incrementCount(obj);
collector.ack(tuple);
}
@Override
public void declareOutputFields(OutputFieldsDeclarer declarer) {
declarer.declare(new Fields("obj", "count", "actualWindowLengthInSeconds"));
}
@Override
public Map<String, Object> getComponentConfiguration() {
Map<String, Object> conf = new HashMap<String, Object>();
conf.put(Config.TOPOLOGY_TICK_TUPLE_FREQ_SECS, emitFrequencyInSeconds);
return conf;
}
}