Checks
Controller Version
0.13.1
Deployment Method
ArgoCD
Checks
To Reproduce
Just run a GitHub job with any container.
Describe the bug
I'm trying to run some basic tests with containerMode.type=kubernetes-novolume. But the runner pod cannot seem to talk to workflow pod. The workflow pod comes up, and then the runner immediate shuts it down because it doesn't get response back from the workflow pod.
The controller seems fine. I actually have dind and kubernetes modes working already, but kubernetes-novolume is an attractive option for us, if we can get it working.
Describe the expected behavior
I'm just running a basic job that prints 'hello' in an ubuntu container. But the job fails immediately in the Intialize containers step. I see the workflow pod come up and get into "Running" state and then immediate get terminated.
Additional Context
controllerServiceAccount:
name: arc-controller-rdu1-gha-rs-controller
namespace: tenant-arc-systems
githubConfigUrl: https://github.com/infinity-microsoft
githubConfigSecret: github-app-secret
# gitHub runner labels for workflow targeting
runnerScaleSetName: "gpu-condor2-rss-k8s-nv"
runnerGroup: "condor2"
minRunners: 1
maxRunners: 12
template:
spec:
containers:
- name: runner
image: ghcr.io/actions/actions-runner:latest
command: ["/home/runner/run.sh"]
env:
- name: ACTIONS_RUNNER_CONTAINER_HOOKS
value: /home/runner/k8s-novolume/index.js
- name: ACTIONS_RUNNER_POD_NAME
valueFrom:
fieldRef:
fieldPath: metadata.name
- name: ACTIONS_RUNNER_IMAGE
value: ghcr.io/actions/actions-runner:latest # should match the runnerimage
- name: ACTIONS_RUNNER_REQUIRE_JOB_CONTAINER
value: "true"Controller Logs
https://gist.github.com/mtanida/de2e2868a1dbac3149c800305f14fa42
Runner Pod Logs
https://gist.github.com/mtanida/985bfda2da34d3f7f693de2bc4fc3f29
Checks
Controller Version
0.13.1
Deployment Method
ArgoCD
Checks
To Reproduce
Describe the bug
I'm trying to run some basic tests with
containerMode.type=kubernetes-novolume. But the runner pod cannot seem to talk to workflow pod. The workflow pod comes up, and then the runner immediate shuts it down because it doesn't get response back from the workflow pod.The controller seems fine. I actually have
dindandkubernetesmodes working already, butkubernetes-novolumeis an attractive option for us, if we can get it working.Describe the expected behavior
I'm just running a basic job that prints 'hello' in an ubuntu container. But the job fails immediately in the
Intialize containersstep. I see the workflow pod come up and get into "Running" state and then immediate get terminated.Additional Context
Controller Logs
Runner Pod Logs