https://github.com/zkfmapf123/k8s-github-action

github-action use k8s
https://github.com/zkfmapf123/k8s-github-action

Last synced: 5 months ago
JSON representation

github-action use k8s

• Host: GitHub
• URL: https://github.com/zkfmapf123/k8s-github-action
• Owner: zkfmapf123
• Created: 2025-06-14T10:48:45.000Z (about 1 year ago)
• Default Branch: master
• Last Pushed: 2025-06-15T14:42:11.000Z (about 1 year ago)
• Last Synced: 2025-09-05T20:03:26.989Z (10 months ago)
• Size: 350 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# K8S Github Action Runners

## ...

- [x] Github Action use K8S

- [ ] 요청 시 Action Pod / Node Scale-out

- [ ] Github Action 병렬처리

    - [ ] Karpenter 깊게 이해하기

    - [ ] Github Action 요청이 많을때 자체적으로 Action 이 늘어날순없을까? ( Action : Pod = 1: 1 )

    - [ ] Action Deployment를 Replica를 늘리면 Karpenter가 자체적으로 늘어나게 할 수 있을까?

## Environments

- EKS

- Node Spec (t3.small)

    - 2 CPU

    - 2 MEM

- Karpenter

## Namespace / Secret 생성

```sh

kubectl create namespace action

kubectl create secret generic github-runner-secret \

  --namespace action \

  --from-literal=token=YOUR_GITHUB_TOKEN

```

## Karpetner 설정

- ttlSecondsAfterEmpty: 30 ## 인스턴스가 비워지면 30초후 삭제

## 설정방법 (Action Runner)

![1](./public/1.png)

![2](./public/2.png)

![3](./public/3.png)

- Secret에 토큰 저장하기

![4](./public/4.png)

![5](./public/5.png)

## Trouble Shooting...

### EC2 Node Issue

- Node Status가 NotReady

- EC2 Instance Profile이 없어서 -> Kubelet이 EKS Control Plane에 접근 불가

- InstanceProfile을 직접 지정

![ts-1](./public/ts-1.png)

![ts-2](./public/ts-2.png)

```sh

## karpenter log 분석

kubectl -n karpenter logs deployment/karpenter

## karpenter.yaml

provider:

    instanceProfile: eks-karpenter-action-instance-profile

```

### EC2 NodePool 없음 (진행 중)

- 6개의 노드가 실패 후 -> karpenter가 nodeclaim을 기준으로 노드를 생성하려고 시도

```sh

  Warning  FailedScheduling  14s   default-scheduler  0/6 nodes are available: 6 Insufficient cpu. preemption: 0/6 nodes are available: 6 No preemption victims found for incoming pod.

  Normal   Nominated         13s   karpenter          Pod should schedule on: nodeclaim/default-tn8fg

```

- k describe nodeclaims.karpenter.sh 

![ts-4](./public/ts-4.png)

- nodeClaim 에 구성된 노드가 EKS 클러스터에 join 되지 않았음 (Node not registerd with cluster)

- nodeClaim으로 구성된 Ec2에 들어가서 보면 아래와 같은 에러가 발생하고 있음

```sh

 1498 csi_plugin.go:887] Failed to contact API server when waiting for CSINode publishing: Unauthorized

```