I have an application (SonarQube) that is used relatively infrequently. Basically it just does a few code scans a day and which point a developer may visit the application for a few minutes and look at the results before heading back to the code to get back to work or implement fixes. Basically, what I’d like to do is upon the first request the application will scale up to one container (which can take up to two minutes) and for the container to stay scaled to one until there haven’t been any requests for the last 30 minutes. That way if a developer is flipping back and forth between code and the scan results it the application will be prompt but when they haven’t interacted with the application for the last 30 minutes the application will shut down.

I use terraform for deploying my infrastructure and I noticed that under the custom_scale_rule section it lists all the available custom_rule_types and one of those is http which made me think that there is the possibility to customize the http scaling rule instead of only being able to provide the concurrent requests like you can with the http_scale_rule.

I’m not sure what metadata I can provide to the http scale rule since the azure docs point to keda.sh which has a KEDA HTTP scaler but it is a custom resource and doesn’t specifically have a metadata section in the custom resource so I’m a little at a loss for how to configure anything.

I’m not specifically tied to this architecture but the ability for container apps to scale to 0 and scale back up to 1 as soon as they receive an http request is extremely appealing to me.

If anyone could point me in the right direction on how to configure the http custom scaler or suggest and alternative infrastructure that accomplishes something similar I’d be very appreciative.

Thanks!

Also posted to https://learn.microsoft.com/en-us/answers/questions/1846777/is-it-possible-to-have-a-little-more-control-of-co