Using Bulk Processor
editUsing Bulk Processor
editThe BulkProcessor
class offers a simple interface to flush bulk operations automatically based on the number or size
of requests, or after a given period.
To use it, first create a BulkProcessor
instance:
import org.elasticsearch.action.bulk.BackoffPolicy; import org.elasticsearch.action.bulk.BulkProcessor; import org.elasticsearch.common.unit.ByteSizeUnit; import org.elasticsearch.common.unit.ByteSizeValue; import org.elasticsearch.core.TimeValue; BulkProcessor bulkProcessor = BulkProcessor.builder( client, new BulkProcessor.Listener() { @Override public void beforeBulk(long executionId, BulkRequest request) { ... } @Override public void afterBulk(long executionId, BulkRequest request, BulkResponse response) { ... } @Override public void afterBulk(long executionId, BulkRequest request, Throwable failure) { ... } }) .setBulkActions(10000) .setBulkSize(new ByteSizeValue(5, ByteSizeUnit.MB)) .setFlushInterval(TimeValue.timeValueSeconds(5)) .setConcurrentRequests(1) .setBackoffPolicy( BackoffPolicy.exponentialBackoff(TimeValue.timeValueMillis(100), 3)) .build();
Add your Elasticsearch client |
|
This method is called just before bulk is executed. You can for example see the numberOfActions with
|
|
This method is called after bulk execution. You can for example check if there was some failing requests
with |
|
This method is called when the bulk failed and raised a |
|
We want to execute the bulk every 10 000 requests |
|
We want to flush the bulk every 5mb |
|
We want to flush the bulk every 5 seconds whatever the number of requests |
|
Set the number of concurrent requests. A value of 0 means that only a single request will be allowed to be executed. A value of 1 means 1 concurrent request is allowed to be executed while accumulating new bulk requests. |
|
Set a custom backoff policy which will initially wait for 100ms, increase exponentially and retries up to three
times. A retry is attempted whenever one or more bulk item requests have failed with an |
By default, BulkProcessor
:
-
sets bulkActions to
1000
-
sets bulkSize to
5mb
- does not set flushInterval
- sets concurrentRequests to 1, which means an asynchronous execution of the flush operation.
- sets backoffPolicy to an exponential backoff with 8 retries and a start delay of 50ms. The total wait time is roughly 5.1 seconds.
Add requests
editThen you can simply add your requests to the BulkProcessor
:
bulkProcessor.add(new IndexRequest("twitter", "_doc", "1").source(/* your doc here */)); bulkProcessor.add(new DeleteRequest("twitter", "_doc", "2"));
Closing the Bulk Processor
editWhen all documents are loaded to the BulkProcessor
it can be closed by using awaitClose
or close
methods:
bulkProcessor.awaitClose(10, TimeUnit.MINUTES);
or
bulkProcessor.close();
Both methods flush any remaining documents and disable all other scheduled flushes, if they were scheduled by setting
flushInterval
. If concurrent requests were enabled, the awaitClose
method waits for up to the specified timeout for
all bulk requests to complete then returns true
; if the specified waiting time elapses before all bulk requests complete,
false
is returned. The close
method doesn’t wait for any remaining bulk requests to complete and exits immediately.
Using Bulk Processor in tests
editIf you are running tests with Elasticsearch and are using the BulkProcessor
to populate your dataset
you should better set the number of concurrent requests to 0
so the flush operation of the bulk will be executed
in a synchronous manner:
BulkProcessor bulkProcessor = BulkProcessor.builder(client, new BulkProcessor.Listener() { /* Listener methods */ }) .setBulkActions(10000) .setConcurrentRequests(0) .build(); // Add your requests bulkProcessor.add(/* Your requests */); // Flush any remaining requests bulkProcessor.flush(); // Or close the bulkProcessor if you don't need it anymore bulkProcessor.close(); // Refresh your indices client.admin().indices().prepareRefresh().get(); // Now you can start searching! client.prepareSearch().get();
Global Parameters
editGlobal parameters can be specified on the BulkRequest as well as BulkProcessor, similar to the REST API. These global parameters serve as defaults and can be overridden by local parameters specified on each sub request. Some parameters have to be set before any sub request is added - index, type - and you have to specify them during BulkRequest or BulkProcessor creation. Some are optional - pipeline, routing - and can be specified at any point before the bulk is sent.
try (BulkProcessor processor = initBulkProcessorBuilder(listener) .setGlobalIndex("tweets") .setGlobalType("_doc") .setGlobalRouting("routing") .setGlobalPipeline("pipeline_id") .build()) { processor.add(new IndexRequest() .source(XContentType.JSON, "user", "some user")); processor.add(new IndexRequest("blogs").id("1") .source(XContentType.JSON, "title", "some title")); }
global parameters from the BulkRequest will be applied on a sub request |
|
local pipeline parameter on a sub request will override global parameters from BulkRequest |