[
https://issues.apache.org/jira/browse/FLINK-3332?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15143029#comment-15143029
]
ASF GitHub Bot commented on FLINK-3332:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/1620#discussion_r52630685
--- Diff:
flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/operators/GenericExactlyOnceSink.java
---
@@ -0,0 +1,173 @@
+/**
+ * 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
+ * <p/>
+ * http://www.apache.org/licenses/LICENSE-2.0
+ * <p/>
+ * 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 org.apache.flink.streaming.runtime.operators;
+
+import org.apache.flink.api.common.ExecutionConfig;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.tuple.Tuple;
+import org.apache.flink.api.java.typeutils.TypeExtractor;
+import org.apache.flink.core.memory.DataInputView;
+import org.apache.flink.runtime.state.AbstractStateBackend;
+import org.apache.flink.runtime.state.StateHandle;
+import org.apache.flink.streaming.api.operators.AbstractStreamOperator;
+import org.apache.flink.streaming.api.operators.OneInputStreamOperator;
+import org.apache.flink.streaming.api.watermark.Watermark;
+import org.apache.flink.streaming.runtime.streamrecord.StreamRecord;
+import org.apache.flink.streaming.runtime.tasks.StreamTaskState;
+
+import java.io.IOException;
+import java.io.Serializable;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+/**
+ * Generic Sink that emits its input elements into an arbitrary backend.
This sink is integrated with the checkpointing
+ * mechanism to provide exactly once semantics.
+ *
+ * Incoming records are stored within a {@link
org.apache.flink.runtime.state.AbstractStateBackend}, and only committed if a
+ * checkpoint is completed. Should a job fail while the data is being
committed, no exactly once guarantee can be made.
+ * @param <IN>
+ */
+public abstract class GenericExactlyOnceSink<IN extends Tuple> extends
AbstractStreamOperator<IN> implements OneInputStreamOperator<IN, IN> {
+ private AbstractStateBackend.CheckpointStateOutputView out;
+ private TypeSerializer<IN> serializer;
+ protected TypeInformation<IN> typeInfo;
+
+ private ExactlyOnceState state = new ExactlyOnceState();
+
+ /**
+ * Saves a handle in the state.
+ * @param checkpointId
+ * @throws IOException
+ */
+ private void saveHandleInState(final long checkpointId) throws
IOException {
+ //only add handle if a new OperatorState was created since the
last snapshot/notify
+ if (out != null) {
+ out.writeByte(0); //EOF-byte
+ StateHandle<DataInputView> handle =
out.closeAndGetHandle();
+ if (state.pendingHandles.containsKey(checkpointId)) {
+
state.pendingHandles.get(checkpointId).add(handle);
+ } else {
+ ArrayList<StateHandle<DataInputView>> list =
new ArrayList<>();
+ list.add(handle);
+ state.pendingHandles.put(checkpointId, list);
+ }
+ out = null;
+ }
+ }
+
+ @Override
+ public StreamTaskState snapshotOperatorState(final long checkpointId,
final long timestamp) throws Exception {
+ StreamTaskState taskState =
super.snapshotOperatorState(checkpointId, timestamp);
+ saveHandleInState(checkpointId);
+ taskState.setFunctionState(state);
+ return taskState;
+ }
+
+ @Override
+ public void restoreState(StreamTaskState state, long recoveryTimestamp)
throws Exception {
+ super.restoreState(state, recoveryTimestamp);
+ this.state = (ExactlyOnceState) state.getFunctionState();
+ out = null;
+ }
+
+ @Override
+ public void notifyOfCompletedCheckpoint(long checkpointId) throws
Exception {
--- End diff --
I guess I have missed something, but how do prevent the following from
happening: You've completed a checkpoint and written all the elements up to
this point to the external system. Now the processing continues, but before you
can complete the next checkpoint, the system crashes. As a result, your state
will be rolled back to the latest checkpoint. However, this means that you will
write all elements from the last checkpoint again if the next completed
checkpoint message arrives, won't you?
> Provide an exactly-once Cassandra connector
> -------------------------------------------
>
> Key: FLINK-3332
> URL: https://issues.apache.org/jira/browse/FLINK-3332
> Project: Flink
> Issue Type: Improvement
> Components: Streaming Connectors
> Reporter: Robert Metzger
> Assignee: Chesnay Schepler
>
> With FLINK-3311, we are adding a Cassandra connector to Flink.
> It would be good to also provide an "exactly-once" C* connector.
> I would like to first discuss how we are going to implement this in Flink.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
