azagrebin commented on a change in pull request #12278:
URL: https://github.com/apache/flink/pull/12278#discussion_r435362084
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/jobmaster/slotpool/SlotPoolImpl.java
##########
@@ -648,26 +648,8 @@ boolean offerSlot(
slotOffer.getResourceProfile(),
taskManagerGateway);
- // check whether we have request waiting for this slot
- PendingRequest pendingRequest =
pendingRequests.removeKeyB(allocationID);
Review comment:
There is this `UnfulfillableSlotRequestException` which is still a fail
fast route if RM finds that a certain request profile cannot be fulfilled at
all with any existing slot and cannot be allocated. It is relevant for both
batch and streaming and bulk as I see. I do not know the whole background of
this. At first glance, this looks to me as an optimisation that complicates
things a bit at the moment. It is probably necessary to avoid timeout waiting
to cancel everything if it is already clear that allocation can never succeed.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
