Hi all! At Platmoshere I see kube-green (https://kube-green.dev/). Has anyone ever used it? It might be useful to turn off all the pods in a namespace at time x and turn back on only those used Slack Conversation

Haven’t used it and probably never will. The problem most of the time is not the pod - if it’s doing nothing it’s mainly consuming RAM and producing cloud cost. KEDA is a much more mature alternative if you want to work on your pod scaling. However, both strategies do not answer the real question: How do I scale my cluster nodes on demand in the right way? The impact of nods on CO2 and cost is much higher than the pods 🙂

My need is not so much to reduce costs or CO2, but to "release" resources in the "development" environment that are not being tested because the developments are on standby. This way I wouldn't have to continually increase the resources of the cluster due to applications that aren't being used. The idea was precisely to turn off all the pods at around 8pm and then automatically activate only those that were actually invoked

https://keda.sh/docs/2.14/scalers/cron/#:~:text=a%20recurring%20schedule.-,Scale%20to%200%20during%20off%20hours,0%20ouside%20that%20time%20window.

It's can be use the query vs prometheus and intercept the call to the service specific to scale up the pod.