0aa13832b5
### What changes were proposed in this pull request? Celeborn fetch chunk also should support check timeout #### Test case ``` executor instance 20 SQL: SELECT count(1) from (select /*+ REPARTITION(100) */ * from spark_auxiliary.t50g) tmp; --conf spark.celeborn.client.spark.shuffle.writer=sort \ --conf spark.celeborn.client.fetch.excludeWorkerOnFailure.enabled=true \ --conf spark.celeborn.client.push.timeout=10s \ --conf spark.celeborn.client.push.replicate.enabled=true \ --conf spark.celeborn.client.push.revive.maxRetries=10 \ --conf spark.celeborn.client.reserveSlots.maxRetries=10 \ --conf spark.celeborn.client.registerShuffle.maxRetries=3 \ --conf spark.celeborn.client.push.blacklist.enabled=true \ --conf spark.celeborn.client.blacklistSlave.enabled=true \ --conf spark.celeborn.client.fetch.timeout=30s \ --conf spark.celeborn.client.push.data.timeout=30s \ --conf spark.celeborn.client.push.limit.inFlight.timeout=600s \ --conf spark.celeborn.client.push.maxReqsInFlight=32 \ --conf spark.celeborn.client.shuffle.compression.codec=ZSTD \ --conf spark.celeborn.rpc.askTimeout=30s \ --conf spark.celeborn.client.rpc.reserveSlots.askTimeout=30s \ --conf spark.celeborn.client.shuffle.batchHandleChangePartition.enabled=true \ --conf spark.celeborn.client.shuffle.batchHandleCommitPartition.enabled=true \ --conf spark.celeborn.client.shuffle.batchHandleReleasePartition.enabled=true ``` Test with 3 worker and add a `Thread.sleep(100s)` before worker handle `ChunkFetchRequest` Before patch <img width="1783" alt="截屏2023-06-14 上午11 20 55" src="https://github.com/apache/incubator-celeborn/assets/46485123/182dff7d-a057-4077-8368-d1552104d206"> After patch <img width="1792" alt="image" src="https://github.com/apache/incubator-celeborn/assets/46485123/3c8b7933-8ace-426d-8e9f-04e0aabfac8e"> The log shows the fetch timeout checker workers ``` 23/06/14 11:14:54 ERROR WorkerPartitionReader: Fetch chunk 0 failed. org.apache.celeborn.common.exception.CelebornIOException: FETCH_DATA_TIMEOUT at org.apache.celeborn.common.network.client.TransportResponseHandler.failExpiredFetchRequest(TransportResponseHandler.java:147) at org.apache.celeborn.common.network.client.TransportResponseHandler.lambda$new$1(TransportResponseHandler.java:103) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) 23/06/14 11:14:54 WARN RssInputStream: Fetch chunk failed 1/6 times for location PartitionLocation[ id-epoch:35-0 host-rpcPort-pushPort-fetchPort-replicatePort:10.169.48.203-9092-9094-9093-9095 mode:MASTER peer:(host-rpcPort-pushPort-fetchPort-replicatePort:10.169.48.202-9092-9094-9093-9095) storage hint:StorageInfo{type=HDD, mountPoint='/mnt/ssd/0', finalResult=true, filePath=} mapIdBitMap:null], change to peer org.apache.celeborn.common.exception.CelebornIOException: Fetch chunk 0 failed. at org.apache.celeborn.client.read.WorkerPartitionReader$1.onFailure(WorkerPartitionReader.java:98) at org.apache.celeborn.common.network.client.TransportResponseHandler.failExpiredFetchRequest(TransportResponseHandler.java:146) at org.apache.celeborn.common.network.client.TransportResponseHandler.lambda$new$1(TransportResponseHandler.java:103) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) Caused by: org.apache.celeborn.common.exception.CelebornIOException: FETCH_DATA_TIMEOUT at org.apache.celeborn.common.network.client.TransportResponseHandler.failExpiredFetchRequest(TransportResponseHandler.java:147) ... 8 more 23/06/14 11:14:54 INFO SortBasedShuffleWriter: Memory used 72.0 MB ``` ### Why are the changes needed? ### Does this PR introduce _any_ user-facing change? ### How was this patch tested? Closes #1587 from AngersZhuuuu/CELEBORN-676. Authored-by: Angerszhuuuu <angers.zhu@gmail.com> Signed-off-by: Angerszhuuuu <angers.zhu@gmail.com> |
||
|---|---|---|
| .. | ||
| assets/logo | ||
| configuration | ||
| celeborn_ratis_shell.md | ||
| cluster_planning.md | ||
| deploy_on_k8s.md | ||
| deploy.md | ||
| migration.md | ||
| monitoring.md | ||
| README.md | ||
| upgrade.md | ||
| hide | license | ||
|---|---|---|---|
|
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. |