1996fanrui commented on code in PR #28001: URL: https://github.com/apache/flink/pull/28001#discussion_r3147383944
########## flink-tests/src/test/java/org/apache/flink/test/checkpointing/RecoveredStateFilteringLargeRecordITCase.java: ########## @@ -0,0 +1,225 @@ +/* + * 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.flink.test.checkpointing; + +import org.apache.flink.api.common.JobStatus; +import org.apache.flink.api.common.eventtime.WatermarkStrategy; +import org.apache.flink.api.common.typeinfo.Types; +import org.apache.flink.api.connector.source.lib.NumberSequenceSource; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiterStrategy; +import org.apache.flink.api.java.functions.KeySelector; +import org.apache.flink.configuration.CheckpointingOptions; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.configuration.ExternalizedCheckpointRetention; +import org.apache.flink.configuration.MemorySize; +import org.apache.flink.configuration.RestartStrategyOptions; +import org.apache.flink.configuration.StateRecoveryOptions; +import org.apache.flink.configuration.TaskManagerOptions; +import org.apache.flink.connector.datagen.source.DataGeneratorSource; +import org.apache.flink.core.execution.JobClient; +import org.apache.flink.runtime.memory.MemoryManager; +import org.apache.flink.runtime.minicluster.MiniCluster; +import org.apache.flink.runtime.testutils.CommonTestUtils; +import org.apache.flink.runtime.testutils.MiniClusterResourceConfiguration; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.functions.sink.v2.DiscardingSink; +import org.apache.flink.test.junit5.InjectMiniCluster; +import org.apache.flink.test.junit5.MiniClusterExtension; +import org.apache.flink.util.TestLoggerExtension; + +import org.junit.jupiter.api.Test; +import org.junit.jupiter.api.extension.ExtendWith; +import org.junit.jupiter.api.extension.RegisterExtension; +import org.junit.jupiter.api.io.TempDir; + +import javax.annotation.Nullable; + +import java.io.File; +import java.time.Duration; +import java.util.Collections; +import java.util.List; +import java.util.Random; +import java.util.concurrent.ThreadLocalRandom; + +import static org.apache.flink.configuration.RestartStrategyOptions.RestartStrategyType.NO_RESTART_STRATEGY; + +/** + * Integration test: recovery with filtering must keep at most one pre-filter source buffer in + * flight per task, even when records are large enough to span many source buffers. + * + * <p>Strategy: configure the network memory segment to Flink's minimum page size and generate + * records several times larger. Every record therefore forces the recovery loop to allocate many + * consecutive pre-filter buffers, exercising the serialized getBuffer/recycle cycle that must reuse + * a single heap segment. A sleeping downstream operator induces back-pressure on the keyBy shuffle + * so that inflight buffers actually accumulate and are captured by the unaligned checkpoint; + * otherwise the recovery path would have nothing to filter. + * + * <p>Filtering during recovery is gated by two config options. Both are pinned on explicitly in + * {@link #getEnv} so the test deterministically exercises the filtering handler. + * + * <p>If the one-at-a-time invariant were violated during recovery, the runtime check in {@code + * InputChannelRecoveredStateHandler.getPreFilterBuffer} would fail with {@link + * IllegalStateException} and the job would fail. Reaching a completed post-recovery checkpoint is + * therefore proof that the invariant held across all buffer cycles. + */ +@ExtendWith({TestLoggerExtension.class}) +class RecoveredStateFilteringLargeRecordITCase { Review Comment: > * Integration test: recovery with filtering must keep at most one pre-filter source buffer in > * flight per task, even when records are large enough to span many source buffers. This test covers the above case, it is green mean one reusable buffer is enough for pre-filtering even if one record is larger than buffer size. > Shouldn't it deadlock? Deadlocks only occur when buffer pool are limited; they are not always reproducible. I used to encounter this occasionally when running `UnalignedCheckpointRescaleITCase` in a loop. And it has 49 different types of jobs. also, I removed the allocation of unpooled segments in `InputChannelRecoveredStateHandler.getBuffer` locally, and try run it in a loop again to reproduce it. I will update it here if it happens. -- 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]
