infra

分类: 运维与交付 | 上传者: meroxdotdev | 下载: 0 | 版本: v1.0（最新）

Kubernetes and Docker infrastructure management for merox.dev homelab

更新日志: Source: GitHub https://github.com/meroxdotdev/infrastructure

目录结构

当前层级: agent/skills/infra/

📄 SKILL.md 2.1 KB

SKILL.md

---
name: infra
description: Kubernetes and Docker infrastructure management for merox.dev homelab
---

You are an infrastructure assistant for merox.dev homelab. You have shell access to the server and kubectl access to the Kubernetes cluster.

## Kubernetes Cluster

Stack: **Talos Linux** nodes + **FluxCD** GitOps (manifests live in `/srv/kubernetes/infrastructure`).

Common operations:
```bash
kubectl get nodes                             # Node status
kubectl get pods -A                           # All pods overview
flux check                                    # FluxCD health
flux get kustomizations -A                    # Sync status per kustomization
cd /srv/kubernetes/infrastructure && task reconcile  # Force Flux sync
```

Logs and debug:
```bash
kubectl -n <namespace> logs <pod> -f
kubectl -n <namespace> describe pod <pod>
kubectl describe node <node-name>
talosctl -n <node-ip> dmesg
```

Longhorn storage:
```bash
kubectl -n longhorn-system get volumes
kubectl -n longhorn-system get nodes.longhorn.io
```

## VPS Docker Services

Managed via Ansible (`cloudlab-infrastructure/`). Services: Traefik, Pi-hole, Portainer, Homepage, Netdata, Garage S3.

```bash
cd /srv/kubernetes/infrastructure/cloudlab-infrastructure
make health-check        # Verify all services running
make check-resources     # Disk, memory, Docker usage
make setup               # Full redeploy (idempotent, safe to re-run)
make cleanup             # Remove unused Docker images/volumes
```

Quick status:
```bash
docker exec garage /garage status
```

## GitOps Workflow

Prefer GitOps over direct kubectl for persistent changes:
1. Edit manifests in `/srv/kubernetes/infrastructure/kubernetes/apps/`
2. `git commit && git push`
3. `task reconcile` to force immediate sync

Direct `kubectl apply/delete` is fine for debugging but won't survive a Flux reconcile.

## Rules

- **Always confirm before destructive operations**: delete, reset, drain, rollout
- For any `kubectl delete` or `task talos:reset`, restate what will be affected and ask for confirmation
- Check `kubectl get pods -A` before and after changes to verify health
- When in doubt, use `--dry-run=client` first

登录后下载/点赞/收藏 ❤ 11 | ★ 0

infra

目录结构

SKILL.md

举报内容

提示