[FLINK-21057] Streaming checkpointing with small interval leads app to hang

• Type: Bug
• Status: Open
• Priority: Not a Priority
• Resolution: Unresolved
• Affects Version/s: 1.11.3
• Component/s: Connectors / Kafka
• Labels: auto-deprioritized-major auto-deprioritized-minor

Environment

• streaming app
• flink cluster in standalone-job / application mode
• 1.11.3 Flink version
• jobmanager --> 1 instance
• taskmanager --> 1 instance
• parallelism --> 2

• Assignee: Unassigned
• Reporter: Nazar Volynets
• Created: 20/Jan/21 23:35
• Updated: 10/Nov/21 10:43

Description

There is a simple streaming app with enabled checkpointing:

• statebackend --> RockDB
• mode --> EXACTLY_ONCE

STRs:
1. Run Flink cluster in standalone-job / application mode (with embedded streaming app)
2. Get error
3. Wait 1 min
4. Stop Flink cluster
4. Repeat steps from 1 to 3 util error :

taskmanager

org.apache.kafka.common.KafkaException: Unexpected error in InitProducerIdResponse; Producer attempted an operation with an old epoch. Either there is a newer producer with the same transactionalId, or the producer's transaction has been expired by the broker.
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.producer.internals.TransactionManager$InitProducerIdHandler.handleResponse(TransactionManager.java:1352) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.producer.internals.TransactionManager$TxnRequestHandler.onComplete(TransactionManager.java:1260) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.ClientResponse.onComplete(ClientResponse.java:109) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.NetworkClient.completeResponses(NetworkClient.java:572) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.NetworkClient.poll(NetworkClient.java:564) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.producer.internals.Sender.maybeSendAndPollTransactionalRequest(Sender.java:414) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.producer.internals.Sender.runOnce(Sender.java:312) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at org.apache.kafka.clients.producer.internals.Sender.run(Sender.java:239) ~[?:?]
flink-kafka-mirror-maker-jobmanager     | 	at java.lang.Thread.run(Unknown Source) ~[?:?]

It is obvious

Please find below:

• streaming app code base (example)
• attached logs
  • jobmanager
  • taskmanager

Example

App

build.gradle (dependencies)

...
ext {
  ...
  javaVersion = '11'
  flinkVersion = '1.12.0'
  scalaBinaryVersion = '2.11'
  ...
}

dependencies {
  ...
  implementation "org.apache.flink:flink-streaming-java_${scalaBinaryVersion}:${flinkVersion}"
  implementation "org.apache.flink:flink-clients_${scalaBinaryVersion}:${flinkVersion}"
  implementation "org.apache.flink:flink-statebackend-rocksdb_${scalaBinaryVersion}:${flinkVersion}"
  ...
}

App

public static void main(String[] args) {
  ...
  StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

  env.setParallelism(2);

  env.enableCheckpointing(500);

  env.setStateBackend(new RocksDBStateBackend("file:///xxx/config/checkpoints/rocksdb", true));

  env.getCheckpointConfig().setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE);
  env.getCheckpointConfig().setMinPauseBetweenCheckpoints(1000);
  env.getCheckpointConfig().setCheckpointTimeout(600000);
  env.getCheckpointConfig().setMaxConcurrentCheckpoints(1);

  FlinkKafkaConsumer<Record> consumer = createConsumer();
  FlinkKafkaProducer<Record> producer = createProducer();

  env
    .addSource(consumer)
    .uid("kafka-consumer")
    .addSink(producer)
    .uid("kafka-producer")
  ;

  env.execute();
}

public static FlinkKafkaConsumer<Record> createConsumer() {
  ...
  Properties props = new Properties();
  props.setProperty(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "kafka-source-1:9091");
  ... // nothing special
  props.setProperty(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, "true");

  FlinkKafkaConsumer<Record> consumer = new FlinkKafkaConsumer<>("topic-1", new RecordKafkaDerSchema(), props);

  ... // RecordKafkaDerSchema --> custom schema is used to copy not only message body but message key too
  ... // SimpleStringSchema --> can be used instead to reproduce issue

  consumer.setStartFromGroupOffsets();
  consumer.setCommitOffsetsOnCheckpoints(true);

  return consumer;
}

public static FlinkKafkaProducer<Record> createProducer() {
  ...
  Properties props = new Properties();
  ...
  props.setProperty(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "kafka-target-1:9094");
  props.setProperty(ProducerConfig.MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION, "1");
  props.setProperty(ProducerConfig.ACKS_CONFIG, "all");
  props.setProperty(ProducerConfig.BATCH_SIZE_CONFIG, "16384");
  props.setProperty(ProducerConfig.LINGER_MS_CONFIG, "1000");
  props.setProperty(ProducerConfig.REQUEST_TIMEOUT_MS_CONFIG, "2000");
  props.setProperty(ProducerConfig.DELIVERY_TIMEOUT_MS_CONFIG, "9000");
  props.setProperty(ProducerConfig.ENABLE_IDEMPOTENCE_CONFIG, "true");
  props.setProperty(ProducerConfig.TRANSACTIONAL_ID_CONFIG, "xxx"); // ignored due to expected behaviour - https://issues.apache.org/jira/browse/FLINK-17691
  props.setProperty(ProducerConfig.TRANSACTION_TIMEOUT_CONFIG, "" + (15 * 60 * 1000)); // decreased from 1 hour to 15 mins; app is going to be restarted quickly
  ...
  FlinkKafkaProducer<Record> producer = new FlinkKafkaProducer<>("topic-1", new RecordKafkaSerSchema(true), props, FlinkKafkaProducer.Semantic.EXACTLY_ONCE);

  ... // RecordKafkaSerSchema --> custom schema is used to copy not only message body but message key too
  ... // SimpleStringSchema --> can be used instead to reproduce issue

  return producer;
}