[lake/tiering] add table dropped handling and nullable write results

fluss-common/src/main/java/org/apache/fluss/lake/writer/LakeWriter.java

@@ -20,6 +20,8 @@
 import org.apache.fluss.annotation.PublicEvolving;
 import org.apache.fluss.record.LogRecord;
 
+import javax.annotation.Nullable;
+
 import java.io.Closeable;
 import java.io.IOException;
 
@@ -43,8 +45,9 @@ public interface LakeWriter<WriteResult> extends Closeable {
     /**
      * Completes the writing process and returns the write result.
      *
-     * @return the write result
+     * @return the write result, or null if no data was written (empty write scenario)
      * @throws IOException if an I/O error occurs
      */
+    @Nullable
     WriteResult complete() throws IOException;
 }

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/committer/TieringCommitOperator.java

@@ -151,6 +151,17 @@ public void processElement(StreamRecord<TableBucketWriteResult<WriteResult>> str
                 collectTableAllBucketWriteResult(tableId);
 
         if (committableWriteResults != null) {
+            // Check if any result is cancelled (table was dropped)
+            boolean isCancelled =
+                    committableWriteResults.stream().anyMatch(TableBucketWriteResult::isCancelled);
+            if (isCancelled) {
+                LOG.info(
+                        "Skipping commit for dropped table {}, table path {}.",
+                        tableId,
+                        tableBucketWriteResult.tablePath());
+                collectedTableBucketWriteResults.remove(tableId);
+                return;
+            }
             try {
                 CommitResult commitResult =
                         commitWriteResults(

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/event/TieringTableDroppedEvent.java

@@ -0,0 +1,63 @@
+/*
+ * 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 org.apache.fluss.flink.tiering.event;
+
+import org.apache.flink.api.connector.source.SourceEvent;
+
+import java.util.Objects;
+
+/**
+ * SourceEvent used to notify TieringSourceReader that a table has been dropped or recreated, and
+ * all pending splits for this table should be skipped.
+ */
+public class TieringTableDroppedEvent implements SourceEvent {
+
+    private static final long serialVersionUID = 1L;
+
+    private final long tableId;
+
+    public TieringTableDroppedEvent(long tableId) {
+        this.tableId = tableId;
+    }
+
+    public long getTableId() {
+        return tableId;
+    }
+
+    @Override
+    public boolean equals(Object o) {
+        if (this == o) {
+            return true;
+        }
+        if (!(o instanceof TieringTableDroppedEvent)) {
+            return false;
+        }
+        TieringTableDroppedEvent that = (TieringTableDroppedEvent) o;
+        return tableId == that.tableId;
+    }
+
+    @Override
+    public int hashCode() {
+        return Objects.hashCode(tableId);
+    }
+
+    @Override
+    public String toString() {
+        return "TieringTableDroppedEvent{" + "tableId=" + tableId + '}';
+    }
+}

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/source/TableBucketWriteResult.java

@@ -30,6 +30,11 @@
  * that the write result is for, the end log offset of tiering, the total number of write results in
  * one round of tiering. It'll be passed to downstream committer operator to collect all the write
  * results of a table and do commit.
+ *
+ * <p>When {@code cancelled} is {@code true}, it indicates this result originates from a tiering
+ * round that was aborted (e.g., the table was dropped). In this case the {@link WriteResult} will
+ * always be {@code null} and the downstream committer should skip the commit and instead report the
+ * cancellation back to the coordinator.
  */
 public class TableBucketWriteResult<WriteResult> implements Serializable {
 
@@ -57,21 +62,26 @@ public class TableBucketWriteResult<WriteResult> implements Serializable {
     // for the round of tiering is finished
     private final int numberOfWriteResults;
 
+    // indicates whether this result is from a cancelled tiering (e.g., table was dropped)
+    private final boolean cancelled;
+
     public TableBucketWriteResult(
             TablePath tablePath,
             TableBucket tableBucket,
             @Nullable String partitionName,
             @Nullable WriteResult writeResult,
             long logEndOffset,
             long maxTimestamp,
-            int numberOfWriteResults) {
+            int numberOfWriteResults,
+            boolean cancelled) {
         this.tablePath = tablePath;
         this.tableBucket = tableBucket;
         this.partitionName = partitionName;
         this.writeResult = writeResult;
         this.logEndOffset = logEndOffset;
         this.maxTimestamp = maxTimestamp;
         this.numberOfWriteResults = numberOfWriteResults;
+        this.cancelled = cancelled;
     }
 
     public TablePath tablePath() {
@@ -103,4 +113,8 @@ public long logEndOffset() {
     public long maxTimestamp() {
         return maxTimestamp;
     }
+
+    public boolean isCancelled() {
+        return cancelled;
+    }
 }

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/source/TableBucketWriteResultSerializer.java

@@ -33,7 +33,10 @@ public class TableBucketWriteResultSerializer<WriteResult>
     private static final ThreadLocal<DataOutputSerializer> SERIALIZER_CACHE =
             ThreadLocal.withInitial(() -> new DataOutputSerializer(64));
 
-    private static final int CURRENT_VERSION = 1;
+    private static final int CURRENT_VERSION = 2;
+
+    // Version 1: original format without the cancelled flag
+    private static final int VERSION_1 = 1;
 
     private final org.apache.fluss.lake.serializer.SimpleVersionedSerializer<WriteResult>
             writeResultSerializer;
@@ -91,6 +94,9 @@ public byte[] serialize(TableBucketWriteResult<WriteResult> tableBucketWriteResu
         // serialize number of write results
         out.writeInt(tableBucketWriteResult.numberOfWriteResults());
 
+        // serialize cancelled flag
+        out.writeBoolean(tableBucketWriteResult.isCancelled());
+
         final byte[] result = out.getCopyOfBuffer();
         out.clear();
         return result;
@@ -99,7 +105,7 @@ public byte[] serialize(TableBucketWriteResult<WriteResult> tableBucketWriteResu
     @Override
     public TableBucketWriteResult<WriteResult> deserialize(int version, byte[] serialized)
             throws IOException {
-        if (version != CURRENT_VERSION) {
+        if (version != CURRENT_VERSION && version != VERSION_1) {
             throw new IOException("Unknown version " + version);
         }
         final DataInputDeserializer in = new DataInputDeserializer(serialized);
@@ -125,7 +131,9 @@ public TableBucketWriteResult<WriteResult> deserialize(int version, byte[] seria
         if (writeResultLength >= 0) {
             byte[] writeResultBytes = new byte[writeResultLength];
             in.readFully(writeResultBytes);
-            writeResult = writeResultSerializer.deserialize(version, writeResultBytes);
+            writeResult =
+                    writeResultSerializer.deserialize(
+                            writeResultSerializer.getVersion(), writeResultBytes);
         } else {
             writeResult = null;
         }
@@ -136,13 +144,18 @@ public TableBucketWriteResult<WriteResult> deserialize(int version, byte[] seria
         long maxTimestamp = in.readLong();
         // deserialize number of write results
         int numberOfWriteResults = in.readInt();
+
+        // deserialize cancelled flag (added in version 2; default to false for version 1)
+        boolean cancelled = (version >= CURRENT_VERSION) && in.readBoolean();
+
         return new TableBucketWriteResult<>(
                 tablePath,
                 tableBucket,
                 partitionName,
                 writeResult,
                 logEndOffset,
                 maxTimestamp,
-                numberOfWriteResults);
+                numberOfWriteResults,
+                cancelled);
     }
 }

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/source/TieringSourceFetcherManager.java

@@ -55,41 +55,61 @@ public TieringSourceFetcherManager(
     }
 
     public void markTableReachTieringMaxDuration(long tableId) {
+        LOG.info("Enqueueing handleTableReachTieringMaxDuration task for table {}", tableId);
+        enqueueTaskForTable(
+                tableId,
+                reader -> {
+                    LOG.debug(
+                            "Executing handleTableReachTieringMaxDuration in split reader for table {}",
+                            tableId);
+                    reader.handleTableReachTieringMaxDuration(tableId);
+                },
+                "handleTableReachTieringMaxDuration");
+    }
+
+    public void markTableDropped(long tableId) {
+        LOG.info("Enqueueing handleTableDropped task for table {}", tableId);
+        enqueueTaskForTable(
+                tableId,
+                reader -> {
+                    LOG.debug("Executing handleTableDropped in split reader for table {}", tableId);
+                    reader.handleTableDropped(tableId);
+                },
+                "handleTableDropped");
+    }
+
+    private void enqueueTaskForTable(
+            long tableId, Consumer<TieringSplitReader<WriteResult>> action, String actionDesc) {
+        SplitFetcher<TableBucketWriteResult<WriteResult>, TieringSplit> splitFetcher;
         if (!fetchers.isEmpty()) {
-            // The fetcher thread is still running. This should be the majority of the cases.
-            LOG.info("fetchers is not empty, marking tiering max duration for table {}", tableId);
-            fetchers.values()
-                    .forEach(
-                            splitFetcher ->
-                                    enqueueMarkTableReachTieringMaxDurationTask(
-                                            splitFetcher, tableId));
+            LOG.info("Fetchers are active, enqueueing {} task for table {}", actionDesc, tableId);
+            fetchers.values().forEach(f -> enqueueReaderTask(f, action));
         } else {
-            SplitFetcher<TableBucketWriteResult<WriteResult>, TieringSplit> splitFetcher =
-                    createSplitFetcher();
             LOG.info(
-                    "fetchers is empty, enqueue marking tiering max duration for table {}",
+                    "No active fetchers, creating new fetcher and enqueueing {} task for table {}",
+                    actionDesc,
                     tableId);
-            enqueueMarkTableReachTieringMaxDurationTask(splitFetcher, tableId);
+            splitFetcher = createSplitFetcher();
+            enqueueReaderTask(splitFetcher, action);
             startFetcher(splitFetcher);
         }
     }
 
-    private void enqueueMarkTableReachTieringMaxDurationTask(
+    @SuppressWarnings("unchecked")
+    private void enqueueReaderTask(
             SplitFetcher<TableBucketWriteResult<WriteResult>, TieringSplit> splitFetcher,
-            long reachTieringDeadlineTable) {
+            Consumer<TieringSplitReader<WriteResult>> action) {
         splitFetcher.enqueueTask(
                 new SplitFetcherTask() {
                     @Override
                     public boolean run() {
-                        ((TieringSplitReader<WriteResult>) splitFetcher.getSplitReader())
-                                .handleTableReachTieringMaxDuration(reachTieringDeadlineTable);
+                        action.accept(
+                                (TieringSplitReader<WriteResult>) splitFetcher.getSplitReader());
                         return true;
                     }
 
                     @Override
-                    public void wakeUp() {
-                        // do nothing
-                    }
+                    public void wakeUp() {}
                 });
     }
 }

fluss-flink/fluss-flink-common/src/main/java/org/apache/fluss/flink/tiering/source/TieringSourceReader.java

@@ -22,6 +22,7 @@
 import org.apache.fluss.client.Connection;
 import org.apache.fluss.flink.adapter.SingleThreadMultiplexSourceReaderBaseAdapter;
 import org.apache.fluss.flink.tiering.event.TieringReachMaxDurationEvent;
+import org.apache.fluss.flink.tiering.event.TieringTableDroppedEvent;
 import org.apache.fluss.flink.tiering.source.split.TieringSplit;
 import org.apache.fluss.flink.tiering.source.state.TieringSplitState;
 import org.apache.fluss.lake.writer.LakeTieringFactory;
@@ -125,9 +126,15 @@ public void handleSourceEvents(SourceEvent sourceEvent) {
             TieringReachMaxDurationEvent reachMaxDurationEvent =
                     (TieringReachMaxDurationEvent) sourceEvent;
             long tableId = reachMaxDurationEvent.getTableId();
-            LOG.info("Received reach max duration for table {}", tableId);
+            LOG.info("Received reach max duration event for table {}", tableId);
             ((TieringSourceFetcherManager<WriteResult>) splitFetcherManager)
                     .markTableReachTieringMaxDuration(tableId);
+        } else if (sourceEvent instanceof TieringTableDroppedEvent) {
+            TieringTableDroppedEvent tableDroppedEvent = (TieringTableDroppedEvent) sourceEvent;
+            long tableId = tableDroppedEvent.getTableId();
+            LOG.info("Received table dropped event for table {}", tableId);
+            ((TieringSourceFetcherManager<WriteResult>) splitFetcherManager)
+                    .markTableDropped(tableId);
         }
     }