Github user tweise commented on a diff in the pull request:
https://github.com/apache/incubator-apex-malhar/pull/133#discussion_r48077190
--- Diff:
contrib/src/main/java/com/datatorrent/contrib/nifi/AbstractNiFiInputOperator.java
---
@@ -0,0 +1,162 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package com.datatorrent.contrib.nifi;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import org.apache.nifi.remote.Transaction;
+import org.apache.nifi.remote.TransferDirection;
+import org.apache.nifi.remote.client.SiteToSiteClient;
+import org.apache.nifi.remote.protocol.DataPacket;
+
+import com.datatorrent.api.Context;
+import com.datatorrent.api.InputOperator;
+
+/**
+ * This is the base implementation of a NiFi input operator.
+ * Subclasses should implement the methods which convert NiFi DataPackets
to tuples and emit them.
+ * <p>
+ * Ports:<br>
+ * <b>Input</b>: No input port<br>
+ * <b>Output</b>: Can have any number of output ports<br>
+ * <br>
+ * Properties:<br>
+ * None<br>
+ * <br>
+ * Compile time checks:<br>
+ * Classes derived from this have to implement the abstract methods
emitTuples(List<T> tuples)
+ * and createTuple(DataPacket dp)<br>
+ * <br>
+ * Run time checks:<br>
+ * None<br>
+ * <br>
+ * Benchmarks:<br>
+ * TBD<br>
+ * </p>
+ *
+ * @displayName Abstract NiFi Input
+ * @category Messaging
+ * @tags input operator
+ * @since 3.3.0
+ */
+
+public abstract class AbstractNiFiInputOperator<T> implements InputOperator
+{
+
+ private static final Logger LOGGER =
LoggerFactory.getLogger(AbstractNiFiInputOperator.class);
+
+ private SiteToSiteClient client;
+ private final SiteToSiteClient.Builder siteToSiteBuilder;
+
+ /**
+ * @param siteToSiteBuilder the builder for a NiFi SiteToSiteClient
+ */
+ public AbstractNiFiInputOperator(final SiteToSiteClient.Builder
siteToSiteBuilder)
+ {
+ this.siteToSiteBuilder = siteToSiteBuilder;
+ }
+
+ @Override
+ public void setup(Context.OperatorContext context)
+ {
+ this.client = siteToSiteBuilder.build();
+ }
+
+ @Override
+ public void teardown()
+ {
+ try {
+ client.close();
+ } catch (IOException e) {
+ LOGGER.error(e.getMessage(), e);
+ }
+ }
+
+ @Override
+ public void emitTuples()
+ {
+ try {
+ final Transaction transaction =
client.createTransaction(TransferDirection.RECEIVE);
+ if (transaction == null) {
+ LOGGER.warn("A transaction could not be created, returning...");
+ return;
+ }
+
+ DataPacket dataPacket = transaction.receive();
+ if (dataPacket == null) {
+ transaction.confirm();
+ transaction.complete();
+ LOGGER.debug("No data available to pull, returning and will try
again...");
+ return;
+ }
+
+ // read all of the available data packets and convert to the given
type
+ final List<T> tuples = new ArrayList<>();
+ do {
+ tuples.add(createTuple(dataPacket));
+ dataPacket = transaction.receive();
+ } while (dataPacket != null);
+
+ // confirm all of the expected data was received
+ transaction.confirm();
+
+ // delegate to sub-classes to emit the tuples
+ emitTuples(tuples);
+
+ // everything was successful so complete the transaction
+ transaction.complete();
+
--- End diff --
Doing the confirm here will lead to data loss when the operator fails. We
would need to retrieve the data into a fault tolerant storage before confirming
the transaction. This looks similar to JMS, where we cannot acknowledge the
message prior to having it stored. The place to store data would be endWindow.
Looks like the pattern from JSM applies here. Kafka and files are different
because they retain the data and we can simply go back to where the operator
left off.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
