summaryzb commented on code in PR #43280:
URL: https://github.com/apache/spark/pull/43280#discussion_r1350032071
##########
core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala:
##########
@@ -2580,7 +2580,8 @@ private[spark] class DAGScheduler(
// if the cluster manager explicitly tells us that the entire worker was
lost, then
// we know to unregister shuffle output. (Note that "worker" specifically
refers to the process
// from a Standalone cluster, where the shuffle service lives in the
Worker.)
- val fileLost = !sc.shuffleDriverComponents.supportsReliableStorage() &&
+ val fileLost = !(sc.shuffleDriverComponents.supportsReliableStorage() &&
+ sc.shuffleDriverComponents.supportsReliableStorage(execId)) &&
Review Comment:
No. If we use `(supportsReliableStorage() ||
supportsReliableStorage(execId))`, newer implementation will always return true
when use disaggregated shuffle cluster.
In old implementation `supportsReliableStorage(execId)` always get true
since not implemented yet, thus the same behavior as just call
`supportsReliableStorage()`
In newer implementation `supportsReliableStorage()` should direct return
boolean, if it got true then `supportsReliableStorage(execId)` do real jugement
--
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]
