advancedxy commented on code in PR #693:
URL: https://github.com/apache/incubator-uniffle/pull/693#discussion_r1150131617
##########
proto/src/main/proto/Rss.proto:
##########
@@ -488,4 +488,26 @@ message CancelDecommissionRequest {
message CancelDecommissionResponse {
StatusCode status = 1;
string retMsg = 2;
-}
\ No newline at end of file
+}
+// ShuffleManager service lives inside of compute-engine's application master,
which handles rss shuffle specific logic
+// per application.
+service ShuffleManager {
+ rpc reportShuffleFetchFailure (ReportShuffleFetchFailureRequest) returns
(ReportShuffleFetchFailureResponse);
+}
+
+message ReportShuffleFetchFailureRequest {
+ // appId normally should be omitted, it's used to avoid wrongly request
issued from remaining executors of another
+ // app which accidentally has the same shuffle manager port with this app.
+ string appId = 1;
+ int32 shuffleId = 2;
+ int32 stageAttemptId = 3;
+ int32 partitionId = 4;
+ string exception = 5;
+ string serverId = 6;
Review Comment:
Good idea. But I believe only the id part is needed in this context. To
reduce the data transferred from client to rss shuffle manager, I don't think
we need other fields.
--
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]
