You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "Till Rohrmann (Jira)" <ji...@apache.org> on 2022/01/05 16:13:00 UTC

[jira] [Created] (FLINK-25533) Preferred AllocationIDs are not respected when fulfilling pending slot requests

Till Rohrmann created FLINK-25533:
-------------------------------------

             Summary: Preferred AllocationIDs are not respected when fulfilling pending slot requests
                 Key: FLINK-25533
                 URL: https://issues.apache.org/jira/browse/FLINK-25533
             Project: Flink
          Issue Type: Bug
          Components: Runtime / Coordination
    Affects Versions: 1.14.2, 1.13.5, 1.15.0
            Reporter: Till Rohrmann
            Assignee: Till Rohrmann
             Fix For: 1.15.0


In order to make best use of local recovery, we have to forward the set of preferred allocations to the {{DeclarativeSlotPoolBridge}} where new slots are matched with pending slot requests. At the moment this is not the case and this means that whenever we try to recover locally while not having all slots available, we might do wrong scheduling decisions.

In order to improve the situation, I propose to forward the set of preferred allocations to the {{DeclarativeSlotPoolBridge}}.



--
This message was sent by Atlassian Jira
(v8.20.1#820001)