[E2E Test] KEDA Autoscaler Pause/Unpause Behavior.

[ramantehlan]

Expected Behaviour

This test verifies that Elasti correctly manages the paused state of a KEDA ScaledObject using annotations, in response to scaling decisions.

We want to confirm that the Elasti controller:

• Automatically pauses the ScaledObject (autoscaling.keda.sh/paused=true) during scale-down.
• Unpauses it (paused=false) when load is detected and scale-up is required.
• Accurately sets the autoscaling.keda.sh/paused-replicas annotation to reflect the last known replica count.

Scenario

You don’t have to follow this rigidly—it’s just a starting point for planning:

1. Deploy an ElastiService

   • Use autoscaler.type = keda
   • Ensure a valid, pre-created KEDA ScaledObject exists for the workload.

2. Trigger Scale-Down

   • Wait for the Elasti controller to patch:

     autoscaling.keda.sh/paused: "true"

     on the ScaledObject.

3. Trigger Scale-Up

   • Simulate traffic or metrics that would cause scaling.

   • Wait for the controller to update:

     autoscaling.keda.sh/paused: "false"

4. Assertion

   • Confirm that the ScaledObject has an annotation:

     autoscaling.keda.sh/paused-replicas: "<expected-replica-count>"