xiechenling created FLINK-40105:
-----------------------------------
Summary: Support configuring Kubernetes memory request factor for
JobManager and TaskManager
Key: FLINK-40105
URL: https://issues.apache.org/jira/browse/FLINK-40105
Project: Flink
Issue Type: Improvement
Components: Deployment / Kubernetes
Reporter: xiechenling
Flink Kubernetes deployments currently support configuring memory limit factors
for JobManager and TaskManager resources, but memory requests are always set to
the full configured process memory.
This makes it hard to run clusters where Kubernetes memory requests should be
lower than memory limits while preserving the existing limit behavior.
This improvement adds two Kubernetes configuration options:
- kubernetes.jobmanager.memory.request-factor
- kubernetes.taskmanager.memory.request-factor
Both options default to 1.0 to preserve the current behavior. Valid values are
greater than 0 and less than or equal to 1.
The generated Kubernetes resource request memory is calculated as configured
memory * memory request factor. Memory limits continue to use the existing
memory limit factor.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
