[
https://issues.apache.org/jira/browse/FLINK-6434?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16000369#comment-16000369
]
Till Rohrmann commented on FLINK-6434:
--------------------------------------
You're right [~tiemsn]. The problem seems to be that the actual slot allocation
has a different lifetime than a concrete slot request. I think we have to
introduce a slot request id which can be used to fail a specific slot request.
That way we decouple the requests from the actual slots a {{SlotPool}} has
available.
> There may be allocatedSlots leak in SlotPool
> --------------------------------------------
>
> Key: FLINK-6434
> URL: https://issues.apache.org/jira/browse/FLINK-6434
> Project: Flink
> Issue Type: Bug
> Components: Cluster Management
> Reporter: shuai.xu
> Assignee: shuai.xu
> Labels: flip-6
>
> If the call allocateSlot() from Execution to Slotpool timeout, the job will
> begin to failover, but the pending request are still in SlotPool, if then a
> new slot register to SlotPool, it may be fulfill the outdated pending request
> and be added to allocatedSlots, but it will never be used and will never be
> recycled.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
