Fehler im Shutdown-Code korrigiert: Shutdown von `EndlessConsumer` zu spät
[demos/kafka/training] / src / main / java / de / juplo / kafka / EndlessConsumer.java
index adebff1..3d154c2 100644 (file)
@@ -1,17 +1,15 @@
 package de.juplo.kafka;
 
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.apache.kafka.clients.consumer.ConsumerRecord;
-import org.apache.kafka.clients.consumer.ConsumerRecords;
-import org.apache.kafka.clients.consumer.KafkaConsumer;
+import org.apache.kafka.clients.consumer.*;
+import org.apache.kafka.common.TopicPartition;
+import org.apache.kafka.common.errors.RecordDeserializationException;
 import org.apache.kafka.common.errors.WakeupException;
-import org.apache.kafka.common.serialization.StringDeserializer;
 
 import javax.annotation.PreDestroy;
 import java.time.Duration;
-import java.util.Arrays;
-import java.util.Optional;
-import java.util.Properties;
+import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.locks.Condition;
@@ -20,67 +18,86 @@ import java.util.concurrent.locks.ReentrantLock;
 
 
 @Slf4j
-public class EndlessConsumer implements Runnable
+@RequiredArgsConstructor
+public class EndlessConsumer<K, V> implements ConsumerRebalanceListener, Runnable
 {
   private final ExecutorService executor;
-  private final String bootstrapServer;
-  private final String groupId;
+  private final PartitionStatisticsRepository repository;
   private final String id;
   private final String topic;
-  private final String autoOffsetReset;
+  private final Consumer<K, V> consumer;
+  private final java.util.function.Consumer<ConsumerRecord<K, V>> handler;
 
   private final Lock lock = new ReentrantLock();
   private final Condition condition = lock.newCondition();
   private boolean running = false;
   private Exception exception;
   private long consumed = 0;
-  private KafkaConsumer<String, String> consumer = null;
 
+  private final Map<Integer, Map<String, Long>> seen = new HashMap<>();
 
-  public EndlessConsumer(
-      ExecutorService executor,
-      String bootstrapServer,
-      String groupId,
-      String clientId,
-      String topic,
-      String autoOffsetReset)
+
+  @Override
+  public void onPartitionsRevoked(Collection<TopicPartition> partitions)
   {
-    this.executor = executor;
-    this.bootstrapServer = bootstrapServer;
-    this.groupId = groupId;
-    this.id = clientId;
-    this.topic = topic;
-    this.autoOffsetReset = autoOffsetReset;
+    partitions.forEach(tp ->
+    {
+      Integer partition = tp.partition();
+      Long newOffset = consumer.position(tp);
+      log.info(
+          "{} - removing partition: {}, offset of next message {})",
+          id,
+          partition,
+          newOffset);
+      Map<String, Long> removed = seen.remove(partition);
+      for (String key : removed.keySet())
+      {
+        log.info(
+            "{} - Seen {} messages for partition={}|key={}",
+            id,
+            removed.get(key),
+            partition,
+            key);
+      }
+      repository.save(new StatisticsDocument(partition, removed, consumer.position(tp)));
+    });
   }
 
   @Override
-  public void run()
+  public void onPartitionsAssigned(Collection<TopicPartition> partitions)
   {
-    try
+    partitions.forEach(tp ->
     {
-      Properties props = new Properties();
-      props.put("bootstrap.servers", bootstrapServer);
-      props.put("group.id", groupId);
-      props.put("client.id", id);
-      props.put("auto.offset.reset", autoOffsetReset);
-      props.put("key.deserializer", StringDeserializer.class.getName());
-      props.put("value.deserializer", StringDeserializer.class.getName());
+      Integer partition = tp.partition();
+      Long offset = consumer.position(tp);
+      log.info("{} - adding partition: {}, offset={}", id, partition, offset);
+      StatisticsDocument document =
+          repository
+              .findById(Integer.toString(partition))
+              .orElse(new StatisticsDocument(partition));
+      consumer.seek(tp, document.offset);
+      seen.put(partition, document.statistics);
+    });
+  }
 
-      this.consumer = new KafkaConsumer<>(props);
 
+  @Override
+  public void run()
+  {
+    try
+    {
       log.info("{} - Subscribing to topic {}", id, topic);
-      consumer.subscribe(Arrays.asList(topic));
+      consumer.subscribe(Arrays.asList(topic), this);
 
       while (true)
       {
-        ConsumerRecords<String, String> records =
+        ConsumerRecords<K, V> records =
             consumer.poll(Duration.ofSeconds(1));
 
         // Do something with the data...
         log.info("{} - Received {} messages", id, records.count());
-        for (ConsumerRecord<String, String> record : records)
+        for (ConsumerRecord<K, V> record : records)
         {
-          consumed++;
           log.info(
               "{} - {}: {}/{} - {}={}",
               id,
@@ -90,14 +107,48 @@ public class EndlessConsumer implements Runnable
               record.key(),
               record.value()
           );
+
+          handler.accept(record);
+
+          consumed++;
+
+          Integer partition = record.partition();
+          String key = record.key() == null ? "NULL" : record.key().toString();
+          Map<String, Long> byKey = seen.get(partition);
+
+          if (!byKey.containsKey(key))
+            byKey.put(key, 0l);
+
+          long seenByKey = byKey.get(key);
+          seenByKey++;
+          byKey.put(key, seenByKey);
         }
+
+        seen.forEach((partiton, statistics) -> repository.save(
+            new StatisticsDocument(
+                partiton,
+                statistics,
+                consumer.position(new TopicPartition(topic, partiton)))));
       }
     }
     catch(WakeupException e)
     {
-      log.info("{} - RIIING!", id);
+      log.info("{} - RIIING! Request to stop consumption - commiting current offsets!", id);
       shutdown();
     }
+    catch(RecordDeserializationException e)
+    {
+      TopicPartition tp = e.topicPartition();
+      long offset = e.offset();
+      log.error(
+          "{} - Could not deserialize  message on topic {} with offset={}: {}",
+          id,
+          tp,
+          offset,
+          e.getCause().toString());
+
+      shutdown(e);
+    }
     catch(Exception e)
     {
       log.error("{} - Unexpected error: {}", id, e.toString(), e);
@@ -105,8 +156,6 @@ public class EndlessConsumer implements Runnable
     }
     finally
     {
-      log.info("{} - Closing the KafkaConsumer", id);
-      consumer.close();
       log.info("{} - Consumer-Thread exiting", id);
     }
   }
@@ -121,9 +170,25 @@ public class EndlessConsumer implements Runnable
     lock.lock();
     try
     {
-      running = false;
-      exception = e;
-      condition.signal();
+      try
+      {
+        log.info("{} - Unsubscribing from topic {}", id, topic);
+        consumer.unsubscribe();
+      }
+      catch (Exception ue)
+      {
+        log.error(
+            "{} - Error while unsubscribing from topic {}: {}",
+            id,
+            topic,
+            ue.toString());
+      }
+      finally
+      {
+        running = false;
+        exception = e;
+        condition.signal();
+      }
     }
     finally
     {
@@ -131,6 +196,11 @@ public class EndlessConsumer implements Runnable
     }
   }
 
+  public Map<Integer, Map<String, Long>> getSeen()
+  {
+    return seen;
+  }
+
   public void start()
   {
     lock.lock();
@@ -173,22 +243,7 @@ public class EndlessConsumer implements Runnable
   public void destroy() throws ExecutionException, InterruptedException
   {
     log.info("{} - Destroy!", id);
-    try
-    {
-      stop();
-    }
-    catch (IllegalStateException e)
-    {
-      log.info("{} - Was already stopped", id);
-    }
-    catch (Exception e)
-    {
-      log.error("{} - Unexpected exception while trying to stop the consumer", id, e);
-    }
-    finally
-    {
-      log.info("{}: Consumed {} messages in total, exiting!", id, consumed);
-    }
+    log.info("{}: Consumed {} messages in total, exiting!", id, consumed);
   }
 
   public boolean running()