Class ElasticSourceBuilder<T>

Type Parameters:
T - type of the output of the mapping function from SearchHit -> T

public final class ElasticSourceBuilder<T>
extends Object
Builder for Elasticsearch source which reads data from Elasticsearch and converts SearchHits using provided mapToItemFn


 BatchSource<String> source = new ElasticSourceBuilder<String>()
   .clientFn(() -> client(host, port))
   .searchRequestFn(() -> new SearchRequest("my-index"))

 BatchStage<String> stage = p.readFrom(source);
Requires clientFn(SupplierEx), searchRequestFn(SupplierEx) and mapToItemFn(FunctionEx).
  • Constructor Details

  • Method Details

    • build

      @Nonnull public BatchSource<T> build()
      Build Elasticsearch BatchSource with supplied parameters
      configured source which is to be used in the pipeline
    • clientFn

      @Nonnull public ElasticSourceBuilder<T> clientFn​(@Nonnull SupplierEx<org.elasticsearch.client.RestClientBuilder> clientFn)
      Set the client supplier function

      The connector uses the returned instance to access Elasticsearch. Also see ElasticClients for convenience factory methods.

      For example, to provide an authenticated client:

       builder.clientFn(() -> client(host, port, username, password))
      This parameter is required.
      clientFn - supplier function returning configured Elasticsearch REST client
    • searchRequestFn

      @Nonnull public ElasticSourceBuilder<T> searchRequestFn​(@Nonnull SupplierEx<> searchRequestFn)
      Set the search request supplier function

      The connector executes this search request to retrieve documents from Elasticsearch.

      For example, to create SearchRequest limited to an index `logs`:

       builder.searchRequestFn(() -> new SearchRequest("logs"))
      This parameter is required.
      searchRequestFn - search request supplier function
    • mapToItemFn

      @Nonnull public <T_NEW> ElasticSourceBuilder<T_NEW> mapToItemFn​(@Nonnull FunctionEx<? super,​T_NEW> mapToItemFn)
      Set the function to map SearchHit to a pipeline item

      For example, to map a SearchHit to a value of a field `productId`:

       builder.mapToItemFn(hit -> (String) hit.getSourceAsMap().get("productId"))
      This parameter is required.
      mapToItemFn - maps search hits to output items
    • optionsFn

      @Nonnull public ElasticSourceBuilder<T> optionsFn​(@Nonnull FunctionEx<? super org.elasticsearch.action.ActionRequest,​org.elasticsearch.client.RequestOptions> optionsFn)
      Set the function that provides RequestOptions

      It can either return a constant value or a value based on provided request.

      For example, use this to provide a custom authentication header:

       sourceBuilder.optionsFn((request) -> {
           RequestOptions.Builder builder = RequestOptions.DEFAULT.toBuilder();
           builder.addHeader("Authorization", "Bearer " + TOKEN);
      optionsFn - function that provides RequestOptions
      See Also:
      RequestOptions in Elastic documentation
    • enableSlicing

      @Nonnull public ElasticSourceBuilder<T> enableSlicing()
      Enable slicing

      Number of slices is equal to globalParallelism (localParallelism * numberOfNodes) when only slicing is enabled. When co-located reading is enabled as well then number of slices for particular node is equal to localParallelism.

      Use this option to read from multiple shards in parallel. It can also be used on single shard, but it may increase initial latency. See Elastic documentation for Sliced Scroll for details.

    • enableCoLocatedReading

      @Nonnull public ElasticSourceBuilder<T> enableCoLocatedReading()
      Enable co-located reading Jet cluster member must run exactly on the same nodes as Elastic cluster.
    • scrollKeepAlive

      @Nonnull public ElasticSourceBuilder<T> scrollKeepAlive​(@Nonnull String scrollKeepAlive)
      Set the keepAlive for Elastic search scroll

      The value must be in Elastic time unit format, e.g. 500ms for 500 milliseconds, 30s for 30 seconds, 5m for 5 minutes. See SearchRequest.scroll(String).

      scrollKeepAlive - keepAlive value, this must be high enough to process all results from a single scroll, default value 1m
    • retries

      @Nonnull public ElasticSourceBuilder<T> retries​(int retries)
      Number of retries the connector will do in addition to Elastic client retries Elastic client tries to connect to a node only once for each request. When a request fails the node is marked dead and is not retried again for the request. This causes problems with single node clusters or in a situation where whole cluster becomes unavailable at the same time (e.g. due to a network issue). The initial delay is 2s, increasing by factor of 2 with each retry (4s, 8s, 16s, ..).
      retries - number of retries, defaults to 5