leaves12138 commented on code in PR #8857: URL: https://github.com/apache/paimon/pull/8857#discussion_r3655916909
########## paimon-common/src/main/java/org/apache/paimon/lookup/sort/db/AsyncLsmCompactor.java: ########## @@ -0,0 +1,129 @@ +/* + * 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.paimon.lookup.sort.db; + +import javax.annotation.Nullable; + +import java.io.IOException; +import java.util.concurrent.CancellationException; +import java.util.concurrent.ExecutionException; +import java.util.concurrent.ExecutorService; +import java.util.concurrent.Future; +import java.util.concurrent.TimeUnit; + +/** Coordinates asynchronous compaction, failure propagation, backpressure, and shutdown. */ +class AsyncLsmCompactor extends LsmCompactor { + + private final ExecutorService executor; + + @Nullable private Future<?> compactionFuture; + + AsyncLsmCompactor( + LsmLevels levels, + CompactorFactory compactorFactory, + int levelZeroFileCountTrigger, + UniversalCompactor.FileSupplier fileSupplier, + UniversalCompactor.FileDeleter fileDeleter, + ExecutorService executor) { + super(levels, compactorFactory, levelZeroFileCountTrigger, fileSupplier, fileDeleter); + this.executor = executor; + } + + @Override + void scheduleIfNeeded() throws IOException { + checkFailure(); + if (!needsCompaction() || compactionFuture != null) { Review Comment: Could we coordinate worker termination and scheduling so a flush cannot miss compaction while the current `Future` is about to complete? There is a race here: 1. The worker finishes a compaction round and its final `needsCompaction()` observes `false`. 2. Before that worker's `Future` completes, the foreground flush adds enough L0 files and calls `scheduleIfNeeded()`. 3. `needsCompaction()` is now `true`, but `compactionFuture != null`, so this method does not submit a new task. 4. The old worker then returns, leaving the newly flushed L0 files uncompacted until another write (and a close in the same window can miss it as well). I reproduced this deterministically by pausing the worker after its final false stability check, adding a new L0 file, calling `scheduleIfNeeded()`, and then releasing the worker. After awaiting the old future, the L0 file count remains `1` and no task is pending. At the backpressure threshold, `applyBackpressure()` can similarly await the old future and return without scheduling the required compaction. The running/pending state and the final stability check need a shared synchronization boundary, or an equivalent rescheduling handshake, rather than relying only on the non-null `Future` reference. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
