Add support in Flink jobs for schema registry provider + avro deser schema #283
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -14,7 +14,7 @@ | |
| * limitations under the License. | ||
| */ | ||
|
|
||
| package ai.chronon.aggregator.stats | ||
|
|
||
| object EditDistance { | ||
|
|
||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,53 @@ | ||
| package ai.chronon.flink | ||
|
|
||
| import ai.chronon.online.TopicChecker | ||
| import ai.chronon.online.TopicInfo | ||
| import org.apache.flink.api.common.eventtime.WatermarkStrategy | ||
| import org.apache.flink.api.common.serialization.DeserializationSchema | ||
| import org.apache.flink.api.scala.createTypeInformation | ||
| import org.apache.flink.connector.kafka.source.KafkaSource | ||
| import org.apache.flink.connector.kafka.source.enumerator.initializer.OffsetsInitializer | ||
| import org.apache.flink.streaming.api.scala.DataStream | ||
| import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment | ||
| import org.apache.kafka.clients.consumer.OffsetResetStrategy | ||
| import org.apache.spark.sql.Row | ||
|
|
||
| class KafkaFlinkSource(kafkaBootstrap: Option[String], | ||
| deserializationSchema: DeserializationSchema[Row], | ||
| topicInfo: TopicInfo) | ||
| extends FlinkSource[Row] { | ||
|
|
||
| val bootstrap: String = | ||
| kafkaBootstrap.getOrElse( | ||
| topicInfo.params.getOrElse( | ||
| "bootstrap", | ||
| topicInfo.params("host") + topicInfo.params | ||
| .get("port") | ||
| .map(":" + _) | ||
| .getOrElse(throw new IllegalArgumentException("No bootstrap servers provided")) | ||
| )) | ||
|
|
||
| // confirm the topic exists | ||
| TopicChecker.topicShouldExist(topicInfo.name, bootstrap, topicInfo.params) | ||
|
|
||
|
Comment on lines
+30
to
+32
Contributor
There was a problem hiding this comment. Close AdminClient to avoid leaks. + try {
+ TopicChecker.topicShouldExist(topicInfo.name, bootstrap, topicInfo.params)
+ } finally {
+ // ensure resources are closed
+ }
|
||
| implicit val parallelism: Int = TopicChecker.getPartitions(topicInfo.name, bootstrap, topicInfo.params) | ||
|
|
||
| override def getDataStream(topic: String, groupByName: String)(env: StreamExecutionEnvironment, | ||
| parallelism: Int): DataStream[Row] = { | ||
| val kafkaSource = KafkaSource | ||
| .builder[Row]() | ||
| .setTopics(topicInfo.name) | ||
| .setGroupId(s"chronon-$groupByName") | ||
| // we might have a fairly large backlog to catch up on, so we choose to go with the latest offset when we're | ||
| // starting afresh | ||
| .setStartingOffsets(OffsetsInitializer.committedOffsets(OffsetResetStrategy.LATEST)) | ||
| .setValueOnlyDeserializer(deserializationSchema) | ||
| .setBootstrapServers(bootstrap) | ||
| .setProperties(TopicChecker.mapToJavaProperties(topicInfo.params)) | ||
| .build() | ||
|
|
||
| env | ||
| .fromSource(kafkaSource, WatermarkStrategy.noWatermarks(), s"Kafka source: $groupByName - ${topicInfo.name}") | ||
| .setParallelism(parallelism) | ||
| } | ||
| } | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,17 @@ | ||
| package ai.chronon.flink | ||
|
|
||
| import ai.chronon.online.TopicInfo | ||
| import org.apache.flink.api.common.serialization.DeserializationSchema | ||
| import org.apache.spark.sql.Encoder | ||
| import org.apache.spark.sql.Row | ||
|
|
||
| /** | ||
| * A SchemaProvider is responsible for providing the Encoder and DeserializationSchema for a given topic. | ||
| * This class handles looking up the schema and then based on the schema type (e.g. Avro, Protobuf) it will create | ||
| * the appropriate Encoder and DeserializationSchema. The Encoder is needed for SparkExpressionEval and the DeserializationSchema | ||
| * is needed to allow Flink's Kafka / other sources to crack open the Array[Byte] payloads. | ||
| * @param conf - Configuration for the SchemaProvider (we pick this up from the topicInfo param map) | ||
| */ | ||
| abstract class SchemaProvider(conf: Map[String, String]) { | ||
| def buildEncoderAndDeserSchema(topicInfo: TopicInfo): (Encoder[Row], DeserializationSchema[Row]) | ||
| } |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| @@ -0,0 +1,45 @@ | ||||||||||||||||||||||||||
| package ai.chronon.flink | ||||||||||||||||||||||||||
| import ai.chronon.online.TopicInfo | ||||||||||||||||||||||||||
| import io.confluent.kafka.schemaregistry.avro.AvroSchema | ||||||||||||||||||||||||||
| import io.confluent.kafka.schemaregistry.client.CachedSchemaRegistryClient | ||||||||||||||||||||||||||
| import io.confluent.kafka.schemaregistry.client.SchemaRegistryClient | ||||||||||||||||||||||||||
| import io.confluent.kafka.schemaregistry.client.rest.exceptions.RestClientException | ||||||||||||||||||||||||||
| import org.apache.flink.api.common.serialization.DeserializationSchema | ||||||||||||||||||||||||||
| import org.apache.spark.sql.Encoder | ||||||||||||||||||||||||||
| import org.apache.spark.sql.Row | ||||||||||||||||||||||||||
| import org.apache.spark.sql.avro.AvroDeserializationSupport | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| class SchemaRegistrySchemaProvider(conf: Map[String, String]) extends SchemaProvider(conf) { | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| private val schemaRegistryUrl: String = | ||||||||||||||||||||||||||
| conf.getOrElse("registry_url", throw new IllegalArgumentException("registry_url not set")) | ||||||||||||||||||||||||||
| private val CacheCapacity: Int = 10 | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| private val schemaRegistryClient: SchemaRegistryClient = buildSchemaRegistryClient(schemaRegistryUrl) | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| private[flink] def buildSchemaRegistryClient(registryUrl: String): SchemaRegistryClient = | ||||||||||||||||||||||||||
| new CachedSchemaRegistryClient(registryUrl, CacheCapacity) | ||||||||||||||||||||||||||
|
Comment on lines
+18
to
+21
Contributor
There was a problem hiding this comment. 🛠️ Refactor suggestion Add error handling for client initialization. Catch and wrap initialization errors, consider adding retry mechanism. - private val schemaRegistryClient: SchemaRegistryClient = buildSchemaRegistryClient(schemaRegistryUrl)
+ private val schemaRegistryClient: SchemaRegistryClient = try {
+ buildSchemaRegistryClient(schemaRegistryUrl)
+ } catch {
+ case e: Exception => throw new IllegalStateException(s"Failed to initialize schema registry client: ${e.getMessage}", e)
+ }📝 Committable suggestion
Suggested change
|
||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| override def buildEncoderAndDeserSchema(topicInfo: TopicInfo): (Encoder[Row], DeserializationSchema[Row]) = { | ||||||||||||||||||||||||||
| val subject = topicInfo.params.getOrElse("subject", s"${topicInfo.name}-value") | ||||||||||||||||||||||||||
| val parsedSchema = | ||||||||||||||||||||||||||
| try { | ||||||||||||||||||||||||||
| val metadata = schemaRegistryClient.getLatestSchemaMetadata(subject) | ||||||||||||||||||||||||||
| schemaRegistryClient.getSchemaById(metadata.getId) | ||||||||||||||||||||||||||
| } catch { | ||||||||||||||||||||||||||
| case e: RestClientException => | ||||||||||||||||||||||||||
| throw new IllegalArgumentException( | ||||||||||||||||||||||||||
| s"Failed to retrieve schema details from the registry. Status: ${e.getStatus}; Error code: ${e.getErrorCode}", | ||||||||||||||||||||||||||
| e) | ||||||||||||||||||||||||||
| case e: Exception => | ||||||||||||||||||||||||||
| throw new IllegalArgumentException("Error connecting to and requesting schema details from the registry", e) | ||||||||||||||||||||||||||
| } | ||||||||||||||||||||||||||
| // we currently only support Avro encoders | ||||||||||||||||||||||||||
| parsedSchema.schemaType() match { | ||||||||||||||||||||||||||
| case AvroSchema.TYPE => | ||||||||||||||||||||||||||
| val schema = parsedSchema.asInstanceOf[AvroSchema] | ||||||||||||||||||||||||||
| AvroDeserializationSupport.build(topicInfo.name, schema.canonicalString(), schemaRegistryWireFormat = true) | ||||||||||||||||||||||||||
| case _ => throw new IllegalArgumentException(s"Unsupported schema type: ${parsedSchema.schemaType()}") | ||||||||||||||||||||||||||
| } | ||||||||||||||||||||||||||
| } | ||||||||||||||||||||||||||
| } | ||||||||||||||||||||||||||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
wonder if we should do Array[Any] & schema separately - instead of Map[String, Any]. That we can do when we need to optimize later - since it is backwards compatible I think
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So we plug the watermark strategy downstream of the sparkexpr eval so that we can pick up the timestamp as the user defined it in their groupby source. Our spark eval output is map[string, any] which is what the type is here (we just reach in and pull out the 'ts' column). One thing to call out - the Map[String, Any] type pretty much stays for one operator hop. Immediately after the spark + watermark op we have a Avro codec operator which converts this to the appropriate event avro ready for wrting to the kv..