ARC runner scale-set unable to start pod for Enterprise runner groups

[poonam2309]

Select Topic Area

Bug

Body

Hi all,

We are trying to execute the github runners agent in GKE cluster therefore we have created the runner-scale-set on enterprise level. When we register the agent using enterprise url then we are able to set the runner-group with name test but when the pipeline starts then the pipeline looks for the agent and stays in queue; moreover; scale-set no pod started in GKE namespace.
We followed this url to configure the agent.

https://docs.github.com/en/actions/hosting-your-own-runners/managing-self-hosted-runners-with-actions-runner-controller/quickstart-for-actions-runner-controller

The below is my workflow file format to calling the enterprise runner-group.

on:
  workflow_dispatch:

jobs:
  Explore-GitHub-Actions:
    runs-on: 
        group: "enterprise/my-group"
        labels: "arc-runner-set"
    steps:
    - run: echo "🎉 This job uses runner scale set runners!"

Are we doing any wrong while calling the runner-group in workflow.

[harmeetsingh11]

Hi,

It seems like you're on the right track with the configuration, but there are a couple of things to double-check.

1. Runner Registration: Ensure that the self-hosted runners are properly registered in the enterprise level, and the runner agent is healthy in your GKE cluster. Sometimes, the registration might not complete successfully, causing it to not show up as available for running the jobs.

2. Correct Runner Group & Labels:

• The runner group in the runs-on field should match the name of the runner group you've created at the enterprise level. Based on your configuration, you're using "enterprise/my-enterprise", so ensure that my-enterprise is the correct name for your runner group in GitHub.
• The labels: "arc-runner-set" should match the label that you've defined for your runner in the runner-scale-set.

3. Workflow Syntax: The syntax for specifying a group in the runs-on field should be correct. You are mixing two parameters here: group and labels. You should make sure that your runner group and labels are being properly referenced.

Try adjusting your workflow file like this:

on:
  workflow_dispatch:

jobs:
  Explore-GitHub-Actions:
    runs-on:
      - group: "enterprise/my-enterprise"
      - labels: ["arc-runner-set"]  # It's an array format
    steps:
    - run: echo "🎉 This job uses runner

Also, check the following:

• Verify that your GKE cluster is properly configured and the agent has the correct permissions to scale and create pods.
• Ensure that the Actions Runner Controller in your GKE cluster is properly set up and has access to the runner scale set.
• Review logs for any errors related to the runner startup and deployment in the GKE namespace.

If after this, the issue still persists, reviewing the logs from both GitHub Actions and the Kubernetes namespace can help identify where the issue lies.

[poonam2309]

hi @harmeetsingh11 ,

All the configuration we have validated, moreover the scale-set agent in enterprise group is showing online. With the below format it is giving error of wrong format
runs-on:
- group: "enterprise/my-enterprise"
- labels: ["arc-runner-set"] # It's an array format

By mistake we mentioned the wrong format in community, this is right format as per the github doc, and I am using the same as below but we are still getting same the error(waiting for a runner to pickup this job), the jobs are in queue only, no agent pod started in kubernetes.
on:
workflow_dispatch:
jobs:
Explore-GitHub-Actions:
runs-on:
group: "enterprise/my-group"
labels: "arc-runner-set"
steps:
- run: echo "🎉 This job uses runner scale set runners!"