Share
Embed
Streaming Apps and Poison Pills: Handle the Unexpected with Kafka Streams
👨🏾💻 https://github.com/DivLoic/xke-ratatouille
📜https://kafka-summit.org/sessions/streaming-apps-poison-pills-handle-unexpected-kafka-streams/
Loïc DIVAD
October 01, 2019
Transcript
79 slides
-
1 @loicmdivad @XebiaFr @loicmdivad @XebiaFr Streaming Apps and Poison Pills:
handle the unexpected with Kafka Streams -
2 @loicmdivad @XebiaFr @loicmdivad @XebiaFr Loïc DIVAD Data Engineer @XebiaFr
@loicmdivad dataxday.fr organizer -
3 @loicmdivad @XebiaFr Processor API: The dark side of Kafka
Streams XKE* - March 2018 XKE = Xebia Knowledge Exchange -
4 @loicmdivad @XebiaFr 4 @loicmdivad @XebiaFr > println(sommaire) Incoming records
may be corrupted, or cannot be handled by the serializer / deserializer. These records are referred to as “poison pills” 1. Log and Crash 2. Skip the Corrupted 3. Sentinel Value Pattern 4. Dead Letter Queue Pattern -
7 @loicmdivad @XebiaFr 7 @loicmdivad @XebiaFr Streaming App Poison Pills
1. Log and Crash - Breakfast 2. Skip the Corrupted - Lunch 3. Sentinel Value Pattern - Drink 4. Dead Letter Queue Pattern - Dinner -
10 @loicmdivad @XebiaFr Really old systems receive raw bytes directly
from message queues 10100110111010101 Exercise #1 - breakfast -
11 @loicmdivad @XebiaFr Really old systems receive raw bytes directly
from message queues With Kafka (Connect and Streams) we’d like to continuously transform these messages 10100110111010101 Kafka Connect Kafka Brokers Exercise #1 - breakfast -
12 @loicmdivad @XebiaFr Really old systems receive raw bytes directly
from message queues With Kafka (Connect and Streams) we’d like to continuously transform these messages But we need a deserializer with special decoder to understand each event What happens if we get a buggy implementation of the deserializer? 10100110111010101 Kafka Connect Kafka Brokers Kafka Streams Exercise #1 - breakfast -
13 @loicmdivad @XebiaFr The Tooling Team They will provide an
appropriate deserializer -
14 @loicmdivad @XebiaFr // Exercise #1: Breakfast sealed trait FoodOrder
case class Breakfast(lang: Lang, fruit: Fruit, liquid: Liquid, pastries: Vector[Pastry] = Vector.empty) extends FoodOrder -
15 @loicmdivad @XebiaFr // Exercise #1: Breakfast sealed trait FoodOrder
case class Breakfast(lang: Lang, fruit: Fruit, liquid: Liquid, pastries: Vector[Pastry] = Vector.empty) extends FoodOrder implicit lazy val BreakfastCodec: Codec[Breakfast] = new Codec[Breakfast] = ??? -
16 @loicmdivad @XebiaFr // Exercise #1: Breakfast sealed trait FoodOrder
case class Breakfast(lang: Lang, fruit: Fruit, liquid: Liquid, pastries: Vector[Pastry] = Vector.empty) extends FoodOrder implicit lazy val BreakfastCodec: Codec[Breakfast] = new Codec[Breakfast] = ??? class FoodOrderSerializer extends Serializer[FoodOrder] = ??? class FoodOrderDeserializer extends Deserializer[FoodOrder] = ??? -
17 @loicmdivad @XebiaFr // Exercise #1: Breakfast sealed trait FoodOrder
case class Breakfast(lang: Lang, fruit: Fruit, liquid: Liquid, pastries: Vector[Pastry] = Vector.empty) extends FoodOrder implicit lazy val BreakfastCodec: Codec[Breakfast] = new Codec[Breakfast] = ??? class FoodOrderSerializer extends Serializer[FoodOrder] = ??? class FoodOrderDeserializer extends Deserializer[FoodOrder] = ??? org.apache.kafka.common.serialization Take Away -
19 @loicmdivad @XebiaFr Log and Crash 2019-04-17 03:43:12 macbook-de-lolo [ERROR]
(LogAndFailExceptionHandler.java:39) - Exception caught during Deserialization, taskId: 0_0, topic: input-food-order, partition: 0, offset: 109 Exception in thread "answer-one-breakfast-0d808ce7-0ef1-44c6-808a-f594bc7fceae-StreamThread-1" org.apache.kafka.streams.errors.StreamsException: Deserialization exception handler is set to fail upon a deserialization error. If you would rather have the streaming pipeline continue after a deserialization error, please set the default.deserialization.exception.handler appropriately. at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:80) at org.apache.kafka.streams.processor.internals.RecordQueue.addRawRecords(RecordQueue.java:101) at org.apache.kafka.streams.processor.internals.PartitionGroup.addRawRecords(PartitionGroup.java:124) ... at org.apache.kafka.streams.processor.internals.StreamTask.addRecords(StreamTask.java:711) at org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:747) Caused by: java.lang.IllegalArgumentException: dishes: Insufficient number of elements: decoded 0 but should have decoded 268435712 at scodec.Attempt$Failure.require(Attempt.scala:108) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:22) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.common.serialization.Deserializer.deserialize(Deserializer.java:58) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.streams.processor.internals.SourceNode.deserializeValue(SourceNode.java:60) at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:66) -
20 @loicmdivad @XebiaFr Log and Crash 2019-04-17 03:43:12 macbook-de-lolo [ERROR]
(LogAndFailExceptionHandler.java:39) - Exception caught during Deserialization, taskId: 0_0, topic: exercise-breakfast, partition: 0, offset: 109 Exception in thread "answer-one-breakfast-0d808ce7-0ef1-44c6-808a-f594bc7fceae-StreamThread-1" org.apache.kafka.streams.errors.StreamsException: Deserialization exception handler is set to fail upon a deserialization error. If you would rather have the streaming pipeline continue after a deserialization error, please set the default.deserialization.exception.handler appropriately. at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:80) at org.apache.kafka.streams.processor.internals.RecordQueue.addRawRecords(RecordQueue.java:101) at org.apache.kafka.streams.processor.internals.PartitionGroup.addRawRecords(PartitionGroup.java:124) ... at org.apache.kafka.streams.processor.internals.StreamTask.addRecords(StreamTask.java:711) at org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:747) Caused by: java.lang.IllegalArgumentException: dishes: Insufficient number of elements: decoded 0 but should have decoded 268435712 at scodec.Attempt$Failure.require(Attempt.scala:108) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:22) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.common.serialization.Deserializer.deserialize(Deserializer.java:58) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.streams.processor.internals.SourceNode.deserializeValue(SourceNode.java:60) at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:66) -
21 @loicmdivad @XebiaFr |val frame1: Array[Byte] = Array(0x33, 0xd4, 0xfc,
0x00, 0x00, 0x00, 0x01, 0xa5) |val frame2: Array[Byte] = Array(0x44, 0xd2, 0xfe, 0x10, 0x02, 0x03, 0x01) -
22 @loicmdivad @XebiaFr |val frame1: Array[Byte] = Array( , 0xd4,
0xfc, 0x00, 0x00, 0x00, 0x01, 0xa5) |val frame2: Array[Byte] = Array( , 0xd2, 0xfe, 0x10, 0x02, 0x03, 0x01) -
23 @loicmdivad @XebiaFr |val frame1: Array[Byte] = Array( , 0xd4,
0xfc, 0x00, 0x00, 0x00, 0x01, 0xa5) |val frame2: Array[Byte] = Array( , 0xd2, 0xfe, 0x10, x2, 0x03, 0x01) |case class Meat(sausages: Int, bacons: Int, . . . ) -
24 @loicmdivad @XebiaFr ▼ Change consumer group ▼ Manually update
my offsets ▼ Reset my streaming app and set my auto reset to LATEST ▽ $ kafka-streams-application-reset ... ▼ Destroy the topic, no message = no poison pill ▽ $ kafka-topics --delete --topic ... ▼ My favourite <3 ▽ $ confluent destroy && confluent start Don’t Do ▼ Fill an issue and suggest a fix to the tooling team -
26 @loicmdivad @XebiaFr 26 @loicmdivad @XebiaFr Log and Crash Like
all consumers, Kafka Streams applications deserialize messages from the broker. The deserialization process can fail. It raises an exception that cannot be caught by our code. Buggy deserializers have to be fixed before the application restarts, by default ... -
28 @loicmdivad @XebiaFr // Exercise #2: Lunch sealed trait FoodOrder
case class Lunch(name: String, price: Double, `type`: LunchType) extends FoodOrder -
29 @loicmdivad @XebiaFr // Exercise #2: Lunch sealed trait FoodOrder
case class Lunch(name: String, price: Double, `type`: LunchType) extends FoodOrder • starter • main • dessert -
31 @loicmdivad @XebiaFr Skip the Corrupted 2019-04-17 03:43:12 macbook-de-lolo [ERROR]
(LogAndFailExceptionHandler.java:39) - Exception caught during Deserialization, taskId: 0_0, topic: exercise-breakfast, partition: 0, offset: 109 Exception in thread "answer-one-breakfast-0d808ce7-0ef1-44c6-808a-f594bc7fceae-StreamThread-1" org.apache.kafka.streams.errors.StreamsException: Deserialization exception handler is set to fail upon a deserialization error. If you would rather have the streaming pipeline continue after a deserialization error, please set the default.deserialization.exception.handler appropriately. at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:80) at org.apache.kafka.streams.processor.internals.RecordQueue.addRawRecords(RecordQueue.java:101) at org.apache.kafka.streams.processor.internals.PartitionGroup.addRawRecords(PartitionGroup.java:124) ... at org.apache.kafka.streams.processor.internals.StreamTask.addRecords(StreamTask.java:711) at org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:747) Caused by: java.lang.IllegalArgumentException: ... decoded 0 but should have decoded 268435712 at scodec.Attempt$Failure.require(Attempt.scala:108) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:22) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.common.serialization.Deserializer.deserialize(Deserializer.java:58) at fr.xebia.ldi.ratatouille.serde.BreakfastDeserializer.deserialize(BreakfastDeserializer.scala:15) at org.apache.kafka.streams.processor.internals.SourceNode.deserializeValue(SourceNode.java:60) at org.apache.kafka.streams.processor.internals.RecordDeserializer.deserialize(RecordDeserializer.java:66) -
32 @loicmdivad @XebiaFr 32 @loicmdivad @XebiaFr public class LogAndFailExceptionHandler implements
DeserializationExceptionHandler /* ... */ public class LogAndContinueExceptionHandler implements DeserializationExceptionHandler /* ... */ -
33 @loicmdivad @XebiaFr public class LogAndFailExceptionHandler implements DeserializationExceptionHandler /* ...
*/ public class LogAndContinueExceptionHandler implements DeserializationExceptionHandler /* ... */ public interface DeserializationExceptionHandler extends Configurable { DeserializationHandlerResponse handle(final ProcessorContext context, final ConsumerRecord<byte[], byte[]> record, final Exception exception); enum DeserializationHandlerResponse { CONTINUE(0, "CONTINUE"), FAIL(1, "FAIL"); /* ... */ } } } -
34 @loicmdivad @XebiaFr public class LogAndFailExceptionHandler implements DeserializationExceptionHandler /* ...
*/ public class LogAndContinueExceptionHandler implements DeserializationExceptionHandler /* ... */ public interface DeserializationExceptionHandler extends Configurable { DeserializationHandlerResponse handle(final ProcessorContext context, final ConsumerRecord<byte[], byte[]> record, final Exception exception); enum DeserializationHandlerResponse { CONTINUE(0, "CONTINUE"), FAIL(1, "FAIL"); /* ... */ } } } Take Away -
36 @loicmdivad @XebiaFr 36 @loicmdivad @XebiaFr The Exception Handler in
the call stack Powered by the Flow intelliJ plugin ➞ findtheflow.io -
37 @loicmdivad @XebiaFr 37 @loicmdivad @XebiaFr Powered by the Flow
intelliJ plugin ➞ findtheflow.io The Exception Handler in the call stack -
38 @loicmdivad @XebiaFr 38 @loicmdivad @XebiaFr Powered by the Flow
intelliJ plugin ➞ findtheflow.io The Exception Handler in the call stack -
39 @loicmdivad @XebiaFr 39 @loicmdivad @XebiaFr Powered by the Flow
intelliJ plugin ➞ findtheflow.io The Exception Handler in the call stack -
40 @loicmdivad @XebiaFr 40 @loicmdivad @XebiaFr Skip the Corrupted All
exceptions thrown by deserializers are caught by a DeserializationExceptionHandler A handler returns Fail or Continue You can implement your own Handler But the two handlers provided by the library are really basic… let’s explore other methods -
41 @loicmdivad @XebiaFr 41 @loicmdivad @XebiaFr All exceptions thrown by
deserializers are caught by a DeserializationExceptionHandler A handler returns Fail or Continue You can implement your own Handler But the two handlers provided by the library are really basic… let’s explore other methods Skip the Corrupted Take Away -
43 @loicmdivad @XebiaFr // Exercise #3: Drink sealed trait FoodOrder
case class Drink(name: String, quantity: Int, `type`: DrinkType, alcohol: Option[Double]) extends FoodOrder -
44 @loicmdivad @XebiaFr // Exercise #3: Drink sealed trait FoodOrder
case class Drink(name: String, quantity: Int, `type`: DrinkType, alcohol: Option[Double]) extends FoodOrder • wine • rhum • beer • champagne • ... -
45 @loicmdivad @XebiaFr We need to turn the deserialization process
into a pure transformation that cannot crash To do so, we will replace corrupted message by a sentinel value. It’s a special-purpose record (e.g: null, None, Json.Null, etc ...) Sentinel Value Pattern f: G → H G H -
46 @loicmdivad @XebiaFr We need to turn the deserialization process
into a pure transformation that cannot crash To do so, we will replace corrupted message by a sentinel value. It’s a special-purpose record (e.g: null, None, Json.Null, etc ...) This allows downstream processors to recognize and handle such sentinel values Sentinel Value Pattern f: G → H G H G H -
47 @loicmdivad @XebiaFr We need to turn the deserialization process
into a pure transformation that cannot crash To do so, we will replace corrupted message by a sentinel value. It’s a special-purpose record (e.g: null, None, Json.Null, etc ...) This allows downstream processors to recognize and handle such sentinel values With Kafka Streams this can be achieved by implementing a Deserializer Sentinel Value Pattern f: G → H G H G H null -
48 @loicmdivad @XebiaFr case object FoodOrderError extends FoodOrder class FoodOrderDeserializer
extends Deserializer[FoodOrder] = ??? -
49 @loicmdivad @XebiaFr case object FoodOrderError extends FoodOrder class FoodOrderDeserializer
extends Deserializer[FoodOrder] = ??? class SentinelValueDeserializer extends FoodOrderDeserializer { override def deserialize(topic: String, data: Array[Byte]): FoodOrder = Try(super.deserialize(topic, data)).getOrElse(FoodOrderErr) } -
51 @loicmdivad @XebiaFr class FoodOrderSentinelValueProcessor extends ValueTransformer[Json, Unit] { var
sensor: Sensor = _ var context: ProcessorContext = _ def metricName(stat: String): MetricName = ??? override def init(context: ProcessorContext): Unit = { this.context = context this.sensor = this.context.metrics.addSensor("sentinel-value", INFO) sensor.add(metricName("total"), new Total()) sensor.add(metricName("rate"), new Rate(TimeUnit.SECONDS, new Count())) } override def transform(value: Json): Unit = sensor.record() } -
54 @loicmdivad @XebiaFr 54 @loicmdivad @XebiaFr Sentinel Value Pattern By
implementing a custom serde we can create a safe Deserializer. Downstreams now receive a sentinel value indicating a deserialization error. Errors can then be treated correctly, example: monitoring the number of deserialization errors with a custom metric But we lost a lot of information about the error… let’s see a last method -
55 @loicmdivad @XebiaFr 55 @loicmdivad @XebiaFr Sentinel Value Pattern By
implementing a custom serde we can create a safe Deserializer. Downstreams now receive a sentinel value indicating a deserialization error. Errors can then be treated correctly, example: monitoring the number of deserialization errors with a custom metric But we lost a lot of information about the error… let’s see a last method Take Away -
57 @loicmdivad @XebiaFr // Exercise #4: Dinner sealed trait FoodOrder
case class Dinner(dish: Command, zone: String, moment: Moment, maybeClient: Option[Client]) extends FoodOrder -
58 @loicmdivad @XebiaFr Dead Letter Queue Pattern In this method
we will let the deserializer fail. For each failure we will send a message to a topic containing corrupted messages. Each message will have the original content of the input message (for reprocessing) and additional meta data about the failure. With Kafka Streams this can be achieved by implementing a DeserializationExceptionHandler Streaming APP dead letter queue input topic output topic -
59 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { override def
handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { } -
60 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { override def
handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { val producerRecord = new ProducerRecord(topic, /*same key, value and ts,*/ headers.asJava) producer.send(producerRecord, /* Producer Callback */ ) DeserializationHandlerResponse.CONTINUE } -
61 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { var topic:
String = _ var producer: KafkaProducer[Array[Byte], Array[Byte]] = _ override def configure(configs: util.Map[String, _]): Unit = ??? override def handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { val producerRecord = new ProducerRecord(topic, /*same key, value and ts,*/ headers.asJava) producer.send(producerRecord, /* Producer Callback */ ) DeserializationHandlerResponse.CONTINUE } -
62 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { var topic:
String = _ var producer: KafkaProducer[Array[Byte], Array[Byte]] = _ override def configure(configs: util.Map[String, _]): Unit = ??? override def handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { val headers = record.headers().toArray ++ Array[Header]( new RecordHeader("processing-time", ???), new RecordHeader("hexa-datetime", ???), new RecordHeader("error-message", ???), ... ) val producerRecord = new ProducerRecord(topic, /*same key, value and ts,*/ headers.asJava) producer.send(producerRecord, /* Producer Callback */ ) DeserializationHandlerResponse.CONTINUE } -
63 @loicmdivad @XebiaFr Fill the headers with some meta data
01061696e0016536f6d6500000005736f6d65206f Value message to hexa Restaurant description Event date and time Food order category -
64 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { var topic:
String = _ var producer: KafkaProducer[Array[Byte], Array[Byte]] = _ override def configure(configs: util.Map[String, _]): Unit = ??? override def handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { val headers = record.headers().toArray ++ Array[Header]( new RecordHeader("processing-time", ???), new RecordHeader("hexa-datetime", ???), new RecordHeader("error-message", ???), ... ) val producerRecord = new ProducerRecord(topic, /*same key, value and ts,*/ headers.asJava) producer.send(producerRecord, /* Producer Callback */ ) DeserializationHandlerResponse.CONTINUE } -
65 @loicmdivad @XebiaFr class DeadLetterQueueFoodExceptionHandler() extends DeserializationExceptionHandler { var topic:
String = _ var producer: KafkaProducer[Array[Byte], Array[Byte]] = _ override def configure(configs: util.Map[String, _]): Unit = ??? override def handle(context: ProcessorContext, record: ConsumerRecord[Array[Byte], Array[Byte]], exception: Exception): DeserializationHandlerResponse = { val headers = record.headers().toArray ++ Array[Header]( new RecordHeader("processing-time", ???), new RecordHeader("hexa-datetime", ???), new RecordHeader("error-message", ???), ... ) val producerRecord = new ProducerRecord(topic, /*same key, value and ts,*/ headers.asJava) producer.send(producerRecord, /* Producer Callback */ ) DeserializationHandlerResponse.CONTINUE } Take Away -
69 @loicmdivad @XebiaFr 69 @loicmdivad @XebiaFr Dead Letter Queue Pattern
You can provide your own implementation of DeserializationExceptionHandler. This lets you use the Producer API to write a corrupted record directly to a quarantine topic. Then you can manually analyse your corrupted records ⚠Warning: This approach have side effects that are invisible to the Kafka Streams runtime. -
70 @loicmdivad @XebiaFr 70 @loicmdivad @XebiaFr Dead Letter Queue Pattern
You can provide your own implementation of DeserializationExceptionHandler. This lets you use the Producer API to write a corrupted record directly to a quarantine topic. Then you can manually analyse your corrupted records ⚠Warning: This approach have side effects that are invisible to the Kafka Streams runtime. Take Away -
73 @loicmdivad @XebiaFr 73 @loicmdivad @XebiaFr Related Post Kafka Connect
Deep Dive – Error Handling and Dead Letter Queues - by Robin Moffatt Building Reliable Reprocessing and Dead Letter Queues with Apache Kafka - by Ning Xia Handling bad messages using Kafka's Streams API - answer by Matthias J. Sax -
74 @loicmdivad @XebiaFr 74 @loicmdivad @XebiaFr Conclusion When using Kafka,
deserialization is the responsibility of the clients. These internal errors are not easy to catch When it’s possible, use Avro + Schema Registry When it’s not possible, Kafka Streams applies techniques to deal with serde errors: - DLQ: By extending a ExceptionHandler - Sentinel Value: By extending a Deserializer -
76 @loicmdivad @XebiaFr 76 @loicmdivad @XebiaFr Images Photo by rawpixel
on Unsplash Photo by João Marcelo Martins on Unsplash Photo by Jordane Mathieu on Unsplash Photo by Brooke Lark on Unsplash Photo by Jakub Kapusnak on Unsplash Photo by Melissa Walker Horn on Unsplash Photo by Aneta Pawlik on Unsplash -
77 @loicmdivad @XebiaFr 77 @loicmdivad @XebiaFr With special thanks to
Robin M. Sylvain L. Giulia B. -
79 @loicmdivad @XebiaFr Pure HTML Akka Http Server Akka Actor
System Kafka Topic Exercise1 Exercise2 Me, clicking everywhere Akka Stream Kafka