[
https://issues.apache.org/jira/browse/FLINK-23892?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17402098#comment-17402098
]
Chesnay Schepler commented on FLINK-23892:
------------------------------------------
I have attached the JM log. The scenario is pretty simple; 1 TM with 1 CPU, one
SSG requiring 0.1 CPU, submit with parallelism=11.
> FineGrainedSlotManager is not sending signal for lacking resources
> ------------------------------------------------------------------
>
> Key: FLINK-23892
> URL: https://issues.apache.org/jira/browse/FLINK-23892
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Reporter: Chesnay Schepler
> Priority: Blocker
> Fix For: 1.14.0
>
> Attachments: flink-zento-standalonesession-10-DESKTOP-FMAKMJ9 -
> Copy.log
>
>
> When the FineGrainedSlotManager is used and resource requirements have been
> explicitly configured, then if the requirements exceed the amount of
> available resources the slot manager will try to allocate as many slots as
> possible but never inform the JM about lack of resources.
> This causes the job to stall indefinitely.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
