[Feature Request] - Support for topologySpreadConstraints in Helm Chart v2 #66594
Description
Topic
topologySpreadConstraints support for k8s pod scheduling
Relevant information
Issue Encountered:
Some of our Enterprise Self-Hosted customers are running into an issue with hot AWS AZs, where the pod scheduler is putting pods in a given AZ/node that's causing issues with IP exhaustion. AWS offers guidance here: https://docs.aws.amazon.com/prescriptive-guidance/latest/ha-resiliency-amazon-eks-apps/spread-workloads.html
However this requires the use of topologySpreadConstraints to set the scheduling policies which isn't currently supported by the latest helm chart versions.
This is having an impact on Self-Hosted Enterprise customers upgrading chart version and this is causing strain on their end, especially if we need that cluster to run to help us with our issues raised via Airbyte Support.
Alternatives Considered
Using affinities/taints to alleviate the issue, but doesn't totally rectify the problem.
Requested Solution:
A way to define the topologySpreadConstraints for all components of the deployment, but primarily jobs being scheduled as those are of the most concern among affected customers.