[FLINK-33526] Improve default autoscaler configs

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: kubernetes-operator-1.7.0
• Component/s: Autoscaler, Kubernetes Operator
• Labels: pull-request-available
• Assignee: Gyula Fora
• Reporter: Gyula Fora
• Created: 12/Nov/23 11:52
• Updated: 13/Nov/23 03:15
• Resolved: 13/Nov/23 03:15

Description

There are a few config defaults that should be improved based on prod usage:

• Metric window : 10 -> 15m
• Catch up duration: 15 -> 30m
• Restart time: 3 -> 5m
• Utilisation boundary: 0.4 -> 0.3

These configs help make the default autoscaler behaviour smoother and less aggressive.

Issue Links

• links to: GitHub Pull Request #708