autoscaler | Autoscaling components for Kubernetes

I have a Github repo with 2 branches on it, develop and main. The first is the "test" environment and the other is the "production" environment. I am working with Google Kubernetes Engine and I have automated deployment from the push on Github to the deploy on GKE. So our workflow is :

1. Pull develop
2. Write code and test locally
3. When everything is fine locally, push on develop (it will automatically deploy on GKE workload app_name_develop)
4. QA tests on app_name_develop
5. If QA tests passed, we create a pull request to put develop into main
6. Automatically deploy on GKE workload app_name_production (from the main branch)

The deployment of the container is defined in Dockerfile and the Kubernetes deployment is defined in kubernetes/app.yaml. Those two files are tracked with Git inside the repo.

The problem here is when we create a pull request to put develop into main, it also take the two files app.yaml and Dockerfile from develop to main. We end up with the settings from develop in main, and it messes the whole thing.

I can't define env variables in those files because it could end up in the wrong branch. My question is : How can I exclude those files from the pull request ? Or is there any way to manage multiples environment without having to manually modify the files after each pull request ?

I don't know if it can hlphere is my Dockerfile :

I am trying to change the sync period as mentioned in the following k8s document. I found the file named kube-controller-manager.yaml in /etc/kubernetes/manifests. I changed the timeoutSeconds: value from 15 secs(default) to 60 secs. Now I have 2 questions based on above info:

• 1. Is this the right way to change the sync period ? Cause I have read in the document that there is some flag named --horizontal-pod-autoscaler-sync-period, but adding that is also not working.
• 1. And any changes made to the file kube-controller-manager.yaml are getting restored to default whenever I restart the minikube? what should I do? Please let me know any solution or any view in this.

I am trying to auto-scale my pods based on CloudSQL instance response time. We are using cloudsql-proxy for secure connection. Deployed the Custom Metrics Adapter.

https://raw.githubusercontent.com/GoogleCloudPlatform/k8s-stackdriver/master/custom-metrics-stackdriver-adapter/deploy/production/adapter_new_resource_model.yaml

I came across this page regarding the kube auto-scaler: https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/FAQ.md#what-are-the-parameters-to-ca

From this, I can tell that part of the reason why some of my nodes aren't being scaled down, is because they have local-data set on them...

By default, the auto-scaler ignores nodes that have local data. I want to change this to false. However, I can't find instructions anywhere about how to change it.

I can't find any obvious things to edit. There's nothing in kube-system that implies it is about the autoscaler configuration - just the autoscaler status ConfigMap.

Could somebody help please? Thank you!

I have a GCP managed instance group that I want to scale out and in between 0 and 1 instances using a cron schedule. GCP has a limitation that says:

Scaling schedules can only be used for MIGs that have at least one other type of autoscaling signal, such as a signal for scaling based on average CPU utilization, load balancing serving capacity, or Cloud Monitoring metrics.

So I must specify an additional autoscaling signal. The documentation goes on to suggest a workaround:

to scale based only on a schedule, you can set your CPU utilization target to 100%.

So I did. But then the managed group does not scale in to 0, it just stays at 1. I've not used the Scale-in controls, so the only thing AFAICT that can prevent scale in is the 10 minute Stabilization period, which I have accounted for.

My autoscaler configuration:

I want to add some flags to change sync periods. can I do it with minikube and kubectl? Or will I have to install and use kubeadm for any such kind of initialization? I refered the this link.

I created and ran the yaml file but there was an error stating that

error: unable to recognize "sync.yaml": no matches for kind "ClusterConfiguration" in version "kubeadm.k8s.io/v1beta2"

sync.yaml that I have used to change the flag (with minikube):

I see there are 2 separate metrics ApproximateNumberOfMessagesVisible and ApproximateNumberOfMessagesNotVisible.

Using number of messages visible causes processing pods to get triggered for termination immediately after they pick up the message from queue, as they're no longer visible. If I use number of messages not visible, it will not scale up.

I'm trying to scale a kubernetes service using horizontal pod autoscaler and external metric from SQS. Here is template external metric:

I am running a Kubernetes cluster including metrics server add-on and Prometheus monitoring. I would like to know which Kubernetes components or activities use/can use monitoring data from the cluster.

What do I mean with "Kubernetes components or activities"?

Obviously, one of the main use cases for monitoring data are all autoscaling mechanisms, including Horizontal Pod Autoscaler, Vertical Pod Autoscaler and Cluster Autoscaler. I am searching for further components or activities, which use live monitoring data from a Kubernetes cluster, and potentially a short explanation why they use it (if it is not obvious). Also it would be interesting to know, which of those components or activities must work with monitoring data and which can work with monitoring data, i.e. can be configured to work with monitoring data.

What do I mean with "monitoring data"?

Monitoring data include, but are not limited to: Node metrics, Pod metrics, Container metrics, network metrics and custom/application-specific metrics (e.g. captured and exposed by third-party tools like Prometheus).

I am thankful for every answer or comment in advance!

This is very strange today, I used AWS EKS cluster, and it works fine for my HPA yesterday and today morning. Starting afternoon, nothing change, my HPA suddenly not work!!

This is my HPA: