Why does my KEDA ScalingModifiers formula fail when referencing a trigger name?

Below is an explanation of why your formula fails and some alternative approaches you might consider.

---

Why the Formula Fails

KEDA’s scalingModifiers expects any identifiers in the formula to evaluate to numeric metric values. In your formula:

YAMLCopied
formula: "one < 75 ? 1 : 0"

the expression “one” is interpreted as a literal value (i.e. a variable name) that should be convertible to a float. Since there isn’t any built‐in numeric variable named “one” (trigger names aren’t automatically cast to their metric value), the engine attempts to convert the literal string "one" to a number and fails with the error:

error validating formula in ScalingModifiers invalid argument for float(one)

---

What You Can Do

At this time, KEDA does not support using trigger names (or arbitrary user-defined variables) inside the scalingModifiers formula to dynamically pull the metric value for comparison. The formula syntax expects direct numeric expressions instead of references to external trigger metrics.

---

Alternative Approaches

1. Leverage HPA with Multiple Metrics:
   One approach is to use the Horizontal Pod Autoscaler (HPA) with multiple metric blocks. When multiple metrics are configured (for example, one for CPU utilization and one from an external source), HPA calculates a proposed replica count for each and then selects the highest count. This way, you can let HPA decide the scaling behavior if the CPU utilization metric is below a threshold and combine that with scheduled metrics.

   For example, you could set up HPA like the following:

   YAMLCopied
   apiVersion: autoscaling/v2
   kind: HorizontalPodAutoscaler
   metadata:
     name: your-application
   spec:
     scaleTargetRef:
       apiVersion: apps/v1
       kind: Deployment
       name: your-deployment
     minReplicas: 1
     maxReplicas: 10
     metrics:
     - type: Resource
       resource:
         name: cpu
         target:
           type: Utilization
           averageUtilization: 75
     - type: External
       external:
         metric:
           name: scheduled-replicas
         target:
           type: Value
           value: "5"

   In this example, when the external metric (which you can drive with a scheduled job or a custom metric adapter) pushes the desired value, HPA will choose the higher of the counts. Then you can adjust your external metric logic or even use a custom controller that evaluates your CPU and then “overrides” the external metric value when needed (for example, setting it to 3 if CPU is lower than 75).

2. Implement a Custom Metric Adapter:
   If you need more complex logic (e.g., “if CPU is less than 75 then change the desired replicas from 5 to 3 during a schedule window”), you could consider shipping an external metric that performs that logic. A custom metric adapter or intermediary controller might calculate and expose a metric (say, scheduled-desired-replicas) based on both the schedule and the CPU usage. Then, use that metric when scaling.

---

To summarize, the error is due to the inability to reference trigger names in your scalingModifiers formula. You will need to use a different approach (such as combining metrics in HPA or leveraging an external metric that embeds your custom logic) to achieve the desired behavior.