Skip to content

Commit f80ea83

Browse files
samxbrsamxbr
committed
Handle individual doc parsing failure gracefully in bulk request with pipeline (elastic#138624)
(cherry picked from commit 933354b)
1 parent 13804a8 commit f80ea83

File tree

4 files changed

+190
-1
lines changed

4 files changed

+190
-1
lines changed

‎docs/changelog/138624.yaml‎

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,6 @@
1+
pr: 138624
2+
summary: Handle individual doc parsing failure in bulk request with pipeline
3+
area: Ingest Node
4+
type: bug
5+
issues:
6+
- 138445

‎server/src/internalClusterTest/java/org/elasticsearch/ingest/IngestClientIT.java‎

Lines changed: 86 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -426,6 +426,92 @@ public void testPipelineProcessorOnFailure() throws Exception {
426426
assertThat(inserted.get("readme"), equalTo("pipeline with id [3] is a bad pipeline"));
427427
}
428428

429+
public void testBulkRequestWithInvalidJsonAndPipeline() throws Exception {
430+
// Test that when a document with invalid JSON is in a bulk request with a pipeline,
431+
// the invalid document fails gracefully without causing the entire bulk request to fail.
432+
// This tests the fix for https://github.com/elastic/elasticsearch/issues/138445
433+
434+
createIndex("test_index");
435+
436+
putJsonPipeline(
437+
"test-pipeline",
438+
(builder, params) -> builder.field("description", "test pipeline")
439+
.startArray("processors")
440+
.startObject()
441+
.startObject("test")
442+
.endObject()
443+
.endObject()
444+
.endArray()
445+
);
446+
447+
// Create a bulk request with valid and invalid documents
448+
BulkRequest bulkRequest = new BulkRequest();
449+
450+
// Valid document
451+
IndexRequest validRequest = new IndexRequest("test_index").id("valid_doc");
452+
validRequest.source("{\"valid\":\"test\"}", XContentType.JSON);
453+
validRequest.setPipeline("test-pipeline");
454+
bulkRequest.add(validRequest);
455+
456+
// Invalid document with missing closing brace
457+
IndexRequest invalidRequest = new IndexRequest("test_index").id("invalid_doc");
458+
invalidRequest.source("{\"invalid\":\"json\"", XContentType.JSON);
459+
invalidRequest.setPipeline("test-pipeline");
460+
bulkRequest.add(invalidRequest);
461+
462+
// Invalid document with duplicate fields
463+
IndexRequest invalidRequest2 = new IndexRequest("test_index").id("invalid_doc2");
464+
invalidRequest2.source("{\"invalid\":\"json\", \"invalid\":\"json\"}", XContentType.JSON);
465+
invalidRequest2.setPipeline("test-pipeline");
466+
bulkRequest.add(invalidRequest2);
467+
468+
// Another valid document
469+
IndexRequest validRequest2 = new IndexRequest("test_index").id("valid_doc2");
470+
validRequest2.source("{\"valid\":\"test2\"}", XContentType.JSON);
471+
validRequest2.setPipeline("test-pipeline");
472+
bulkRequest.add(validRequest2);
473+
474+
BulkResponse response = client().bulk(bulkRequest).actionGet();
475+
476+
// The bulk request should succeed
477+
assertThat(response.hasFailures(), is(true));
478+
assertThat(response.getItems().length, equalTo(4));
479+
480+
// First document should succeed
481+
BulkItemResponse item0 = response.getItems()[0];
482+
assertThat(item0.isFailed(), is(false));
483+
assertThat(item0.getResponse().getId(), equalTo("valid_doc"));
484+
assertThat(item0.getResponse().getResult(), equalTo(DocWriteResponse.Result.CREATED));
485+
486+
// Second document should fail
487+
BulkItemResponse item1 = response.getItems()[1];
488+
assertThat(item1.isFailed(), is(true));
489+
assertThat(item1.getFailure().getStatus(), equalTo(org.elasticsearch.rest.RestStatus.BAD_REQUEST));
490+
assertThat(item1.getFailure().getCause(), instanceOf(IllegalArgumentException.class));
491+
492+
// Third document should fail
493+
BulkItemResponse item2 = response.getItems()[2];
494+
assertThat(item2.isFailed(), is(true));
495+
assertThat(item2.getFailure().getStatus(), equalTo(org.elasticsearch.rest.RestStatus.BAD_REQUEST));
496+
assertThat(item2.getFailure().getCause(), instanceOf(IllegalArgumentException.class));
497+
498+
// Fourth document should succeed
499+
BulkItemResponse item3 = response.getItems()[3];
500+
assertThat(item3.isFailed(), is(false));
501+
assertThat(item3.getResponse().getId(), equalTo("valid_doc2"));
502+
assertThat(item3.getResponse().getResult(), equalTo(DocWriteResponse.Result.CREATED));
503+
504+
// Verify that the valid documents were indexed
505+
assertThat(client().prepareGet("test_index", "valid_doc").get().isExists(), is(true));
506+
assertThat(client().prepareGet("test_index", "valid_doc2").get().isExists(), is(true));
507+
// Verify that the invalid documents were not indexed
508+
assertThat(client().prepareGet("test_index", "invalid_doc").get().isExists(), is(false));
509+
assertThat(client().prepareGet("test_index", "invalid_doc2").get().isExists(), is(false));
510+
511+
// cleanup
512+
deletePipeline("test-pipeline");
513+
}
514+
429515
public static class ExtendedIngestTestPlugin extends IngestTestPlugin {
430516

431517
@Override

‎server/src/main/java/org/elasticsearch/ingest/IngestService.java‎

Lines changed: 14 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -799,7 +799,20 @@ protected void doRun() {
799799
final XContentMeteringParserDecorator meteringParserDecorator = documentParsingProvider.newMeteringParserDecorator(
800800
indexRequest
801801
);
802-
final IngestDocument ingestDocument = newIngestDocument(indexRequest, meteringParserDecorator);
802+
final IngestDocument ingestDocument;
803+
try {
804+
ingestDocument = newIngestDocument(indexRequest, meteringParserDecorator);
805+
} catch (Exception e) {
806+
// Document parsing failed (e.g. invalid JSON). Handle this gracefully
807+
// by marking this document as failed and continuing with other documents.
808+
final long ingestTimeInNanos = System.nanoTime() - startTimeInNanos;
809+
totalMetrics.postIngest(ingestTimeInNanos);
810+
totalMetrics.ingestFailed();
811+
ref.close();
812+
i++;
813+
onFailure.accept(slot, e);
814+
continue;
815+
}
803816
final org.elasticsearch.script.Metadata originalDocumentMetadata = ingestDocument.getMetadata().clone();
804817
// the document listener gives us three-way logic: a document can fail processing (1), or it can
805818
// be successfully processed. a successfully processed document can be kept (2) or dropped (3).

‎server/src/test/java/org/elasticsearch/ingest/IngestServiceTests.java‎

Lines changed: 84 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -65,6 +65,7 @@
6565
import org.elasticsearch.test.MockLog;
6666
import org.elasticsearch.threadpool.ThreadPool;
6767
import org.elasticsearch.xcontent.XContentBuilder;
68+
import org.elasticsearch.xcontent.XContentParseException;
6869
import org.elasticsearch.xcontent.XContentType;
6970
import org.elasticsearch.xcontent.cbor.CborXContent;
7071
import org.junit.Before;
@@ -1572,6 +1573,89 @@ public void testBulkRequestExecutionWithFailures() throws Exception {
15721573
verify(completionHandler, times(1)).accept(Thread.currentThread(), null);
15731574
}
15741575

1576+
public void testBulkRequestExecutionWithInvalidJsonDocument() {
1577+
// Test that when a document with invalid JSON (e.g., duplicate keys) is in a bulk request with a pipeline,
1578+
// the invalid document fails gracefully without causing the entire bulk request to fail.
1579+
BulkRequest bulkRequest = new BulkRequest();
1580+
String pipelineId = "_id";
1581+
1582+
// Valid document that should succeed
1583+
IndexRequest validRequest = new IndexRequest("_index").id("valid").setPipeline(pipelineId).setFinalPipeline("_none");
1584+
validRequest.source(Requests.INDEX_CONTENT_TYPE, "field1", "value1");
1585+
validRequest.setListExecutedPipelines(true);
1586+
bulkRequest.add(validRequest);
1587+
1588+
// Invalid document with missing closing brace
1589+
String invalidJson = "{\"invalid\":\"json\"";
1590+
IndexRequest invalidRequest = new IndexRequest("_index").id("invalid").setPipeline(pipelineId).setFinalPipeline("_none");
1591+
invalidRequest.source(new BytesArray(invalidJson), XContentType.JSON);
1592+
bulkRequest.add(invalidRequest);
1593+
1594+
// Another valid document that should succeed
1595+
IndexRequest validRequest2 = new IndexRequest("_index").id("valid2").setPipeline(pipelineId).setFinalPipeline("_none");
1596+
validRequest2.source(Requests.INDEX_CONTENT_TYPE, "field2", "value2");
1597+
validRequest2.setListExecutedPipelines(true);
1598+
bulkRequest.add(validRequest2);
1599+
1600+
// Invalid document with duplicated keys
1601+
String invalidJson2 = "{\"@timestamp\":\"2024-06-01T00:00:00Z\",\"@timestamp\":\"2024-06-01T00:00:00Z\"}";
1602+
IndexRequest invalidRequest2 = new IndexRequest("_index").id("invalid").setPipeline(pipelineId).setFinalPipeline("_none");
1603+
invalidRequest2.source(new BytesArray(invalidJson2), XContentType.JSON);
1604+
bulkRequest.add(invalidRequest2);
1605+
1606+
final Processor processor = mock(Processor.class);
1607+
when(processor.getType()).thenReturn("mock");
1608+
when(processor.getTag()).thenReturn("mockTag");
1609+
doAnswer(args -> {
1610+
BiConsumer<IngestDocument, Exception> handler = args.getArgument(1);
1611+
handler.accept(RandomDocumentPicks.randomIngestDocument(random()), null);
1612+
return null;
1613+
}).when(processor).execute(any(), any());
1614+
1615+
IngestService ingestService = createWithProcessors(Map.of("mock", (factories, tag, description, config) -> processor));
1616+
PutPipelineRequest putRequest = new PutPipelineRequest(
1617+
"_id1",
1618+
new BytesArray("{\"processors\": [{\"mock\" : {}}]}"),
1619+
XContentType.JSON
1620+
);
1621+
ClusterState clusterState = ClusterState.builder(ClusterName.DEFAULT)
1622+
.build();
1623+
ClusterState previousClusterState = clusterState;
1624+
clusterState = executePut(putRequest, clusterState);
1625+
ingestService.applyClusterState(new ClusterChangedEvent("", clusterState, previousClusterState));
1626+
1627+
BiConsumer<Integer, Exception> requestItemErrorHandler = mock();
1628+
final BiConsumer<Thread, Exception> onCompletion = mock();
1629+
1630+
ingestService.executeBulkRequest(
1631+
4,
1632+
bulkRequest.requests(),
1633+
indexReq -> {},
1634+
(s) -> false,
1635+
(slot, targetIndex, e) -> fail("Should not redirect to failure store"),
1636+
requestItemErrorHandler,
1637+
onCompletion,
1638+
EsExecutors.DIRECT_EXECUTOR_SERVICE
1639+
);
1640+
1641+
// The invalid documents should fail with a parsing error
1642+
verify(requestItemErrorHandler).accept(
1643+
eq(1), // slot 1 is the invalid document
1644+
argThat(e -> e instanceof XContentParseException)
1645+
);
1646+
verify(requestItemErrorHandler).accept(
1647+
eq(3), // slot 3 is the other invalid document
1648+
argThat(e -> e instanceof XContentParseException)
1649+
);
1650+
1651+
// The bulk listener should still be called with success
1652+
verify(onCompletion).accept(any(), eq(null));
1653+
assertStats(ingestService.stats().totalStats(), 4, 2, 0);
1654+
// Verify that the valid documents were processed (they should have their pipelines executed)
1655+
assertThat(validRequest.getExecutedPipelines(), equalTo(List.of(pipelineId)));
1656+
assertThat(validRequest2.getExecutedPipelines(), equalTo(List.of(pipelineId)));
1657+
}
1658+
15751659
public void testExecuteFailureRedirection() throws Exception {
15761660
final CompoundProcessor processor = mockCompoundProcessor();
15771661
IngestService ingestService = createWithProcessors(

0 commit comments

Comments
 (0)