yabola commented on code in PR #37922:
URL: https://github.com/apache/spark/pull/37922#discussion_r1052111525
##########
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);
Review Comment:
Do we really need to write shuffleMergeInfo in DB? It seems that it will be
[reloaded from here]
(https://github.com/apache/spark/blob/master/common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/RemoteBlockPushResolver.java#L1002)
##########
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());
Review Comment:
I think if mergePartitionsInfo is not finalized,
`mergePartitionsInfo.getReduceIds()` will be empty. How about cleaning the
current unfinalized partitions
`AppShuffleMergePartitionsInfo.shuffleMergePartitions`?
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
