https://github.com/manzil-infinity180/pod-cleanup-controller
A controller that watches for pods in "Evicted" or "Failed" status and deletes them automatically after X minutes with Slack Notification
https://github.com/manzil-infinity180/pod-cleanup-controller
client-go controller golang
Last synced: 7 months ago
JSON representation
A controller that watches for pods in "Evicted" or "Failed" status and deletes them automatically after X minutes with Slack Notification
- Host: GitHub
- URL: https://github.com/manzil-infinity180/pod-cleanup-controller
- Owner: manzil-infinity180
- Created: 2025-07-08T16:37:13.000Z (11 months ago)
- Default Branch: main
- Last Pushed: 2025-07-09T14:05:57.000Z (11 months ago)
- Last Synced: 2025-08-12T00:52:19.797Z (10 months ago)
- Topics: client-go, controller, golang
- Language: Go
- Homepage:
- Size: 29.3 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🧹 Pod Cleanup Controller
A controller that watches for pods in "Evicted" or "Failed" status and deletes them automatically after X minutes.
A Kubernetes controller written in Go that automatically deletes pods that are:
- In `Failed` phase
- In `Evicted` state
- In `CrashLoopBackOff` with restart count ≥ 5
This project is built using `client-go` and uses shared informers for real-time event processing.
## DEMO
```bash
kubectl apply -f examples/failed-terminated-pod.yml
kubectl apply -f examples/failed.yml
// and after that put your cred into .env and run
SLACK_AUTH_TOKEN="---"
SLACK_CHANNEL_ID="XXXXXXXX" // https://app.slack.com/client/T08Q5RCFWGM/C09XXXXEG (CXVVXVXVVXX <- channel id)
go run main.go
```
https://github.com/user-attachments/assets/40987d28-b83e-475f-aa99-310f488e0894
---
## ✨ Features (IN Progress)
- Watches all pods across the cluster
- Identifies problematic pods
- Waits for a configurable delay (default 5 minutes)
- Deletes the pod after grace period
- Logs structured pod metadata in JSON
- Lightweight and stateless
---
## 📦 Use Cases (In progress)
- Automated cleanup for crashlooping test/dev pods
- Reclaiming cluster resources
- Simplifying observability during CI/CD tests
- Educational use for learning Kubernetes controllers and shared informers
---
## 🚀 Getting Started
### Prerequisites
- Go 1.20+
- Access to a Kubernetes cluster (e.g., via `kubectl`)
- [`kubectl` context](https://kubernetes.io/docs/concepts/configuration/organize-cluster-access-kubeconfig/) configured to a running cluster
---
## 🛠️ Running Locally
Clone the repo and build:
```bash
git clone https://github.com/manzil-infinity180/pod-cleanup-controller.git
cd pod-cleanup-controller
go mod tidy
go run main.go
```
---
## 🔬 How It Works
The controller uses a shared informer to watch pod changes. It checks:
1. pod.Status.Phase == Failed
2. pod.Status.Reason == "Evicted"
3. Any container with:
- Waiting.Reason == CrashLoopBackOff
- RestartCount >= 5
* If a pod matches, it will:
- Sleep for 5 minutes (configurable)
- Delete the pod
- Log the deletion in JSON
## Example
```yaml
# crashloop.yaml
apiVersion: v1
kind: Pod
metadata:
name: failed-pod
spec:
containers:
- name: busybox
image: busybox
command: ["false"]
restartPolicy: Always
```
```bash
kubectl apply -f crashloop.yaml
```
# 🙌 Acknowledgments
Built by [@manzil-infinity180](https://github.com/manzil-infinity180) for learning and exploration of Kubernetes controllers and Go internals.