wankunde commented on code in PR #37922: URL: https://github.com/apache/spark/pull/37922#discussion_r1059572807
########## common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/RemoteBlockPushResolver.java: ########## @@ -396,6 +403,56 @@ public void applicationRemoved(String appId, boolean cleanupLocalDirs) { } } + @Override + public void removeShuffleMerge(RemoveShuffleMerge msg) { + AppShuffleInfo appShuffleInfo = validateAndGetAppShuffleInfo(msg.appId); + if (appShuffleInfo.attemptId != msg.appAttemptId) { + throw new IllegalArgumentException( + String.format("The attempt id %s in this RemoveShuffleMerge message does not match " + + "with the current attempt id %s stored in shuffle service for application %s", + msg.appAttemptId, appShuffleInfo.attemptId, msg.appId)); + } + appShuffleInfo.shuffles.computeIfPresent(msg.shuffleId, (shuffleId, mergePartitionsInfo) -> { + boolean deleteCurrent = + msg.shuffleMergeId == DELETE_CURRENT_MERGED_SHUFFLE_ID || + msg.shuffleMergeId == mergePartitionsInfo.shuffleMergeId; + AppAttemptShuffleMergeId currentAppAttemptShuffleMergeId = + new AppAttemptShuffleMergeId( + msg.appId, msg.appAttemptId, msg.shuffleId, mergePartitionsInfo.shuffleMergeId); + AppAttemptShuffleMergeId appAttemptShuffleMergeId = new AppAttemptShuffleMergeId( + msg.appId, msg.appAttemptId, msg.shuffleId, msg.shuffleMergeId); + if(deleteCurrent) { + // request to clean up shuffle we are currently hosting + if (!mergePartitionsInfo.isFinalized()) { + submitCleanupTask(() -> { + closeAndDeleteOutdatedPartitions( + currentAppAttemptShuffleMergeId, mergePartitionsInfo.shuffleMergePartitions); + writeAppAttemptShuffleMergeInfoToDB(appAttemptShuffleMergeId); + }); + } else { + submitCleanupTask(() -> { + deleteMergedFiles(currentAppAttemptShuffleMergeId, + mergePartitionsInfo.getReduceIds()); + writeAppAttemptShuffleMergeInfoToDB(appAttemptShuffleMergeId); + mergePartitionsInfo.setReduceIds(new int[0]); + }); + } + } else if(msg.shuffleMergeId < mergePartitionsInfo.shuffleMergeId) { + throw new RuntimeException(String.format("Asked to remove old shuffle merged data for " + + "application %s shuffleId %s shuffleMergeId %s, but current shuffleMergeId %s ", + msg.appId, msg.shuffleId, msg.shuffleMergeId, mergePartitionsInfo.shuffleMergeId)); + } else if (msg.shuffleMergeId > mergePartitionsInfo.shuffleMergeId) { + // cleanup request for newer shuffle - remove the outdated data we have. + submitCleanupTask(() -> { + closeAndDeleteOutdatedPartitions( + currentAppAttemptShuffleMergeId, mergePartitionsInfo.shuffleMergePartitions); + writeAppAttemptShuffleMergeInfoToDB(appAttemptShuffleMergeId); Review Comment: For example, for `appAttemptShuffleMergeId_1 = AppAttemptShuffleMergeId(app_1, appAttempt_1, shuffle_1, shuffleMergeId_1)`, the main thread submits a cleaner task with `appAttemptShuffleMergeId_1`, and then `writeAppAttemptShuffleMergeInfoToDB(appAttemptShuffleMergeId_1)`. The cleaner thread will cleanup `appAttemptShuffleMergeId_1` from DB later, but we expect `appAttemptShuffleMergeId_1` is in the DB to keep consistent with `appShuffleInfo.shuffles`? -- 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: reviews-unsubscr...@spark.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org