Autoscaling doesn't work with workflow_job

[adcoutinho]

Greetings everyone, after a couple of months working with version 0.22.0 (kubernetes helm 0.17.0) and happy with GithubEvents scaling HRA:

scaleUpTriggers:
- githubEvent: {}
  amount: 1
  duration: "20m"

I decided to test version 0.23.0 (helm 0.18.0) and autoscaling just stopped, checking release notes I found #1379 wich leads me to try workflowJob: {}, but when I try a helm upgrade with HRA code like this:

scaleUpTriggers:
- githubEvent:
    workflowJob: {}
  duration: "30m"

I receive this error message:

Error: UPGRADE FAILED: error validating "": error validating data: ValidationError(HorizontalRunnerAutoscaler.spec.scaleUpTriggers[0].githubEvent): unknown field "workflowJob" in dev.summerwind.actions.v1alpha1.HorizontalRunnerAutoscaler.spec.scaleUpTriggers.githubEvent

I have one namespace, with one controller pod, one webhook pod and 3 different runners based in resource sizes with labels "small-runner", "medium-runner" and "large-runner" all runner in the same Default runner group.

Can anyone give me some help on this?

[toast-gear]

https://github.com/actions-runner-controller/actions-runner-controller/blob/master/charts/actions-runner-controller/docs/UPGRADING.md

[adcoutinho]

Didn't mention before but:

My ARC Helm chart was applied with Terraform Helm Release not helm upgrade command:

resource "helm_release" "actions_runner_controller" {
  for_each   = local.helm_must_be_deployed
  name       = "actions-runner-controller"
  repository = "https://actions-runner-controller.github.io/actions-runner-controller"
  chart      = "actions-runner-controller"
  version    = "0.18.0"
  namespace  = "actions-runner-system"

I split RunnerDeployment and HRA into another Helm Charts and repo for each runner size for example:

self-hosted-runner-large
|Charts.yaml
|values.yaml
| templates
|--deployment.yaml
|--hra.yaml

self-hosted-runner-small
|Charts.yaml
|values.yaml
| templates
|--deployment.yaml
|--hra.yaml

And the error appear on helm upgrade for command: helm upgrade runner-deployment-small --install --wait ./self-hosted-runner-small -f ./self-hosted-runner-small/values.yaml.

[toast-gear]

I don't know anything about the helm_release resource in terraform but I doubt it re-implements Helm (or at the very least doesn't diverge in logic) so the same limitations will apply. Helm will not touch CRDs outside of an initial install so you need an out-of-band step or add logic for handling them or do an uninstall of ARC and removal of the CRDs before doing an install again.

[adcoutinho]

Ok, helm_release has a force_update flag that will delete and recreate the resource.

https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release

I will make a test as soon as possible.

[adcoutinho]

Terraform helm_release force_update don't change anything but after I manually replaced CRDs with kubectl and recreate runners with Workflow_Job ScaleUP trigger everything works as expected.

Thanks @toast-gear