[
https://issues.apache.org/jira/browse/FLINK-14058?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhu Zhu updated FLINK-14058:
----------------------------
Release Note:
The memory configs for table operators are no longer absolute memory
requirements but weight hints now. This means that the actual managed memory
allocated can be smaller or larger than the config values, which depends on the
actual slot managed memory capacity.
The affected configs are table.exec.resource.external-buffer-memory,
table.exec.resource.hash-agg.memory, table.exec.resource.hash-join.memory, and
table.exec.resource.sort.memory.
This change ensures that operators do not over allocate memory or leave
available memory unallocated. It helps the task to be more stable if the slot
managed memory is limited, or more efficient if the slot has adequate managed
memory.
was:
The memory configs for table operators are weight hints now rather than
absolute memory requirements. This means that the actual managed memory
allocated may be smaller or larger than the config values, which depends on the
actual slot managed memory capacity.
The affected configs are table.exec.resource.external-buffer-memory,
table.exec.resource.hash-agg.memory, table.exec.resource.hash-join.memory, and
table.exec.resource.sort.memory.
This ensures that operators do not over allocate memory or leave available
memory unallocated. It helps the task to run more stable if the slot managed
memory is limited, or more efficient if their are adequate slot managed memory.
> FLIP-53 Fine Grained Operator Resource Management
> -------------------------------------------------
>
> Key: FLINK-14058
> URL: https://issues.apache.org/jira/browse/FLINK-14058
> Project: Flink
> Issue Type: New Feature
> Components: Runtime / Coordination
> Affects Versions: 1.9.0
> Reporter: Xintong Song
> Assignee: Zhu Zhu
> Priority: Major
> Labels: Umbrella
> Fix For: 1.10.0
>
>
> This is the umbrella issue of 'FLIP-53: Fine Grained Operator Resource
> Management'.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)