Composed (Logical/Priority) Triggers

[theauheral]

Proposal

I would like to suggest a new feature: composed (Logical/Priority) triggers.

Use-Case

I would like to be able to compose triggers or define a logic/priority between triggers.

Example:
Scaling from 0 to 1<=n<=maxReplicas replicas on a fixed CRON schedule (On/Off) with n managed by the metric (kafka) trigger during the runtime window.

The current design only enables "OR" logic between triggers. I would like to specify an "AND" logic between triggers or a priority of evaluation between triggers. Check trigger 1, if true then check trigger 2.

Anything else?

Specification

• AND logic between triggers
• Alternatively, define priority triggers
• Probably most rigorous, creating a composed trigger class enabling the above two

Possible inspiration: Log4J2 rules and triggers for rotating files based on time and file size

[JorTurFer]

hey @theauheral
Could this be duplicated? #2614

[theauheral]

Hello @JorTurFer,

Yes maybe. Different use cases but same technical implementation I suspect. I would say related but not identical.

I see "Combination of triggers can be used, either with an OR or AND fashion." which is similar yes.
but I also suggest priorities and composability. I guess #2614 has two feature requests with the second one being similar.
The example of Microsoft Azure autoscale has multiple rules but not composed rules.

You could definitely build the two feature requests in one.

[JorTurFer]

Yes, you are right.
I link this too because this is more related (#2440).
Just to clarify this a bit, KEDA delegates the scaling to the HPA Controller so applying an AND or prioritize the metrics it's not possible. The only doable option is the creation of composite scaler IMO

[zroubalik]

Yeah, what we can do is to provide just one metric to HPA in this case and calculate the formula consisting of multiple triggers on KEDA side.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.

[tomkerkhove]

Keep-Alive

[miqm]

@tomkerkhove any update on this?

[ghantasunil]

@tomkerkhove, any update on this?

[tomkerkhove]

Nope, are you willing to contribute it?

[zroubalik]

This is already covered by Scaling Modifiers https://keda.sh/docs/2.14/concepts/scaling-deployments/#scaling-modifiers-experimental

I belive we can close this issue already.

[ghantasunil]

Hi @zroubalik,

I have a use case where I need to scale my application in and out based on the Azure Service Bus queue length and Azure SQL CPU. Below is the condition I am planning to use for scaling. How can I implement this using Scaling Modifiers? I plan to use the Azure Service Bus scalar to get the queue length and the Azure Monitor scalar to monitor the Azure SQL Managed Instance CPU. Appreciate your help. Thanks.

Scaling Out Condition: Scale out when both conditions are met:

• Azure Service Bus queue length > 500 messages
• Azure SQL Managed Instance CPU < 80%

Scaling In Condition: Scale in when any of the following conditions are true:

• Azure Service Bus queue length <= 500 messages
• Azure SQL Managed Instance CPU >= 80%