use of org.elasticsearch.action.bulk.BulkResponse in project nifi by apache.
the class PutElasticsearch5 method onTrigger.
@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) throws ProcessException {
synchronized (esClient) {
if (esClient.get() == null) {
super.setup(context);
}
}
final String id_attribute = context.getProperty(ID_ATTRIBUTE).getValue();
final int batchSize = context.getProperty(BATCH_SIZE).evaluateAttributeExpressions().asInteger();
final List<FlowFile> flowFiles = session.get(batchSize);
if (flowFiles.isEmpty()) {
return;
}
final ComponentLog logger = getLogger();
// Keep track of the list of flow files that need to be transferred. As they are transferred, remove them from the list.
List<FlowFile> flowFilesToTransfer = new LinkedList<>(flowFiles);
try {
final BulkRequestBuilder bulk = esClient.get().prepareBulk();
for (FlowFile file : flowFiles) {
final String index = context.getProperty(INDEX).evaluateAttributeExpressions(file).getValue();
final String docType = context.getProperty(TYPE).evaluateAttributeExpressions(file).getValue();
final String indexOp = context.getProperty(INDEX_OP).evaluateAttributeExpressions(file).getValue();
final Charset charset = Charset.forName(context.getProperty(CHARSET).evaluateAttributeExpressions(file).getValue());
final String id = file.getAttribute(id_attribute);
if (id == null) {
logger.warn("No value in identifier attribute {} for {}, transferring to failure", new Object[] { id_attribute, file });
flowFilesToTransfer.remove(file);
session.transfer(file, REL_FAILURE);
} else {
session.read(file, new InputStreamCallback() {
@Override
public void process(final InputStream in) throws IOException {
// For the bulk insert, each document has to be on its own line, so remove all CRLF
String json = IOUtils.toString(in, charset).replace("\r\n", " ").replace('\n', ' ').replace('\r', ' ');
if (indexOp.equalsIgnoreCase("index")) {
bulk.add(esClient.get().prepareIndex(index, docType, id).setSource(json.getBytes(charset)));
} else if (indexOp.equalsIgnoreCase("upsert")) {
bulk.add(esClient.get().prepareUpdate(index, docType, id).setDoc(json.getBytes(charset)).setDocAsUpsert(true));
} else if (indexOp.equalsIgnoreCase("update")) {
bulk.add(esClient.get().prepareUpdate(index, docType, id).setDoc(json.getBytes(charset)));
} else {
throw new IOException("Index operation: " + indexOp + " not supported.");
}
}
});
}
}
if (bulk.numberOfActions() > 0) {
final BulkResponse response = bulk.execute().actionGet();
if (response.hasFailures()) {
// Responses are guaranteed to be in order, remove them in reverse order
BulkItemResponse[] responses = response.getItems();
if (responses != null && responses.length > 0) {
for (int i = responses.length - 1; i >= 0; i--) {
final BulkItemResponse item = responses[i];
final FlowFile flowFile = flowFilesToTransfer.get(item.getItemId());
if (item.isFailed()) {
logger.warn("Failed to insert {} into Elasticsearch due to {}, transferring to failure", new Object[] { flowFile, item.getFailure().getMessage() });
session.transfer(flowFile, REL_FAILURE);
} else {
session.getProvenanceReporter().send(flowFile, response.remoteAddress().getAddress());
session.transfer(flowFile, REL_SUCCESS);
}
flowFilesToTransfer.remove(flowFile);
}
}
}
// Transfer any remaining flowfiles to success
for (FlowFile ff : flowFilesToTransfer) {
session.getProvenanceReporter().send(ff, response.remoteAddress().getAddress());
session.transfer(ff, REL_SUCCESS);
}
}
} catch (NoNodeAvailableException | ElasticsearchTimeoutException | ReceiveTimeoutTransportException | NodeClosedException exceptionToRetry) {
// Authorization errors and other problems are often returned as NoNodeAvailableExceptions without a
// traceable cause. However the cause seems to be logged, just not available to this caught exception.
// Since the error message will show up as a bulletin, we make specific mention to check the logs for
// more details.
logger.error("Failed to insert into Elasticsearch due to {}. More detailed information may be available in " + "the NiFi logs.", new Object[] { exceptionToRetry.getLocalizedMessage() }, exceptionToRetry);
session.transfer(flowFilesToTransfer, REL_RETRY);
context.yield();
} catch (Exception exceptionToFail) {
logger.error("Failed to insert into Elasticsearch due to {}, transferring to failure", new Object[] { exceptionToFail.getLocalizedMessage() }, exceptionToFail);
session.transfer(flowFilesToTransfer, REL_FAILURE);
context.yield();
}
}
use of org.elasticsearch.action.bulk.BulkResponse in project nifi by apache.
the class PutElasticsearch method onTrigger.
@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) throws ProcessException {
final ComponentLog logger = getLogger();
final String id_attribute = context.getProperty(ID_ATTRIBUTE).getValue();
final int batchSize = context.getProperty(BATCH_SIZE).evaluateAttributeExpressions().asInteger();
final List<FlowFile> flowFiles = session.get(batchSize);
if (flowFiles.isEmpty()) {
return;
}
// Keep track of the list of flow files that need to be transferred. As they are transferred, remove them from the list.
List<FlowFile> flowFilesToTransfer = new LinkedList<>(flowFiles);
try {
final BulkRequestBuilder bulk = esClient.get().prepareBulk();
if (authToken != null) {
bulk.putHeader("Authorization", authToken);
}
for (FlowFile file : flowFiles) {
final String index = context.getProperty(INDEX).evaluateAttributeExpressions(file).getValue();
final String docType = context.getProperty(TYPE).evaluateAttributeExpressions(file).getValue();
final String indexOp = context.getProperty(INDEX_OP).evaluateAttributeExpressions(file).getValue();
final Charset charset = Charset.forName(context.getProperty(CHARSET).evaluateAttributeExpressions(file).getValue());
final String id = file.getAttribute(id_attribute);
if (id == null) {
logger.error("No value in identifier attribute {} for {}, transferring to failure", new Object[] { id_attribute, file });
flowFilesToTransfer.remove(file);
session.transfer(file, REL_FAILURE);
} else {
session.read(file, new InputStreamCallback() {
@Override
public void process(final InputStream in) throws IOException {
String json = IOUtils.toString(in, charset).replace("\r\n", " ").replace('\n', ' ').replace('\r', ' ');
if (indexOp.equalsIgnoreCase("index")) {
bulk.add(esClient.get().prepareIndex(index, docType, id).setSource(json.getBytes(charset)));
} else if (indexOp.equalsIgnoreCase("upsert")) {
bulk.add(esClient.get().prepareUpdate(index, docType, id).setDoc(json.getBytes(charset)).setDocAsUpsert(true));
} else if (indexOp.equalsIgnoreCase("update")) {
bulk.add(esClient.get().prepareUpdate(index, docType, id).setDoc(json.getBytes(charset)));
} else {
throw new IOException("Index operation: " + indexOp + " not supported.");
}
}
});
}
}
final BulkResponse response = bulk.execute().actionGet();
if (response.hasFailures()) {
// Responses are guaranteed to be in order, remove them in reverse order
BulkItemResponse[] responses = response.getItems();
if (responses != null && responses.length > 0) {
for (int i = responses.length - 1; i >= 0; i--) {
final FlowFile flowFile = flowFilesToTransfer.get(i);
if (responses[i].isFailed()) {
logger.error("Failed to insert {} into Elasticsearch due to {}, transferring to failure", new Object[] { flowFile, responses[i].getFailure().getMessage() });
session.transfer(flowFile, REL_FAILURE);
} else {
session.getProvenanceReporter().send(flowFile, context.getProperty(HOSTS).evaluateAttributeExpressions().getValue() + "/" + responses[i].getIndex());
session.transfer(flowFile, REL_SUCCESS);
}
flowFilesToTransfer.remove(flowFile);
}
}
}
// Transfer any remaining flowfiles to success
flowFilesToTransfer.forEach(file -> {
session.transfer(file, REL_SUCCESS);
// Record provenance event
session.getProvenanceReporter().send(file, context.getProperty(HOSTS).evaluateAttributeExpressions().getValue() + "/" + context.getProperty(INDEX).evaluateAttributeExpressions(file).getValue());
});
} catch (NoNodeAvailableException | ElasticsearchTimeoutException | ReceiveTimeoutTransportException | NodeClosedException exceptionToRetry) {
// Authorization errors and other problems are often returned as NoNodeAvailableExceptions without a
// traceable cause. However the cause seems to be logged, just not available to this caught exception.
// Since the error message will show up as a bulletin, we make specific mention to check the logs for
// more details.
logger.error("Failed to insert into Elasticsearch due to {}. More detailed information may be available in " + "the NiFi logs.", new Object[] { exceptionToRetry.getLocalizedMessage() }, exceptionToRetry);
session.transfer(flowFilesToTransfer, REL_RETRY);
context.yield();
} catch (Exception exceptionToFail) {
logger.error("Failed to insert into Elasticsearch due to {}, transferring to failure", new Object[] { exceptionToFail.getLocalizedMessage() }, exceptionToFail);
session.transfer(flowFilesToTransfer, REL_FAILURE);
context.yield();
}
}
use of org.elasticsearch.action.bulk.BulkResponse in project fess by codelibs.
the class FessEsClient method deleteByQuery.
public int deleteByQuery(final String index, final String type, final QueryBuilder queryBuilder) {
final FessConfig fessConfig = ComponentUtil.getFessConfig();
SearchResponse response = client.prepareSearch(index).setTypes(type).setScroll(scrollForDelete).setSize(sizeForDelete).setFetchSource(new String[] { fessConfig.getIndexFieldId() }, null).setQuery(queryBuilder).setPreference(Constants.SEARCH_PREFERENCE_PRIMARY).execute().actionGet(fessConfig.getIndexScrollSearchTimeoutTimeout());
int count = 0;
String scrollId = response.getScrollId();
while (scrollId != null) {
final SearchHits searchHits = response.getHits();
final SearchHit[] hits = searchHits.getHits();
if (hits.length == 0) {
scrollId = null;
break;
}
final BulkRequestBuilder bulkRequest = client.prepareBulk();
for (final SearchHit hit : hits) {
bulkRequest.add(client.prepareDelete(index, type, hit.getId()));
}
count += hits.length;
final BulkResponse bulkResponse = bulkRequest.execute().actionGet(fessConfig.getIndexBulkTimeout());
if (bulkResponse.hasFailures()) {
throw new IllegalBehaviorStateException(bulkResponse.buildFailureMessage());
}
response = client.prepareSearchScroll(scrollId).setScroll(scrollForDelete).execute().actionGet(fessConfig.getIndexBulkTimeout());
scrollId = response.getScrollId();
}
return count;
}
use of org.elasticsearch.action.bulk.BulkResponse in project metron by apache.
the class ElasticsearchWriterTest method testSingleFailure.
@Test
public void testSingleFailure() throws Exception {
Tuple tuple1 = mock(Tuple.class);
BulkResponse response = mock(BulkResponse.class);
when(response.hasFailures()).thenReturn(true);
Exception e = new IllegalStateException();
BulkItemResponse itemResponse = buildBulkItemFailure(e);
when(response.iterator()).thenReturn(ImmutableList.of(itemResponse).iterator());
BulkWriterResponse expected = new BulkWriterResponse();
expected.addError(e, tuple1);
ElasticsearchWriter esWriter = new ElasticsearchWriter();
BulkWriterResponse actual = esWriter.buildWriteReponse(ImmutableList.of(tuple1), response);
assertEquals("Response should have one error and zero successes", expected, actual);
}
use of org.elasticsearch.action.bulk.BulkResponse in project nutch by apache.
the class TestElasticIndexWriter method setup.
@Before
public void setup() {
conf = NutchConfiguration.create();
conf.addResource("nutch-site-test.xml");
bulkRequestSuccessful = false;
clusterSaturated = false;
curNumFailures = 0;
maxNumFailures = 0;
Settings settings = Settings.builder().build();
ThreadPool threadPool = new ThreadPool(settings);
// customize the ES client to simulate responses from an ES cluster
client = new AbstractClient(settings, threadPool) {
@Override
public void close() {
}
@Override
protected <Request extends ActionRequest, Response extends ActionResponse, RequestBuilder extends ActionRequestBuilder<Request, Response, RequestBuilder>> void doExecute(Action<Request, Response, RequestBuilder> action, Request request, ActionListener<Response> listener) {
BulkResponse response = null;
if (clusterSaturated) {
// pretend the cluster is saturated
curNumFailures++;
if (curNumFailures >= maxNumFailures) {
// pretend the cluster is suddenly no longer saturated
clusterSaturated = false;
}
// respond with a failure
BulkItemResponse failed = new BulkItemResponse(0, OpType.INDEX, new BulkItemResponse.Failure("nutch", "index", "failure0", new EsRejectedExecutionException("saturated")));
response = new BulkResponse(new BulkItemResponse[] { failed }, 0);
} else {
// respond successfully
BulkItemResponse success = new BulkItemResponse(0, OpType.INDEX, new IndexResponse(new ShardId("nutch", UUID.randomUUID().toString(), 0), "index", "index0", 0, true));
response = new BulkResponse(new BulkItemResponse[] { success }, 0);
}
listener.onResponse((Response) response);
}
};
// customize the plugin to signal successful bulk operations
testIndexWriter = new ElasticIndexWriter() {
@Override
protected Client makeClient(Configuration conf) {
return client;
}
@Override
protected BulkProcessor.Listener bulkProcessorListener() {
return new BulkProcessor.Listener() {
@Override
public void afterBulk(long executionId, BulkRequest request, BulkResponse response) {
if (!response.hasFailures()) {
bulkRequestSuccessful = true;
}
}
@Override
public void afterBulk(long executionId, BulkRequest request, Throwable failure) {
}
@Override
public void beforeBulk(long executionId, BulkRequest request) {
}
};
}
};
}
Aggregations