Keyboard shortcuts

Press ← or β†’ to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

πŸŽ‰ REAL Wuji Templates Successfully Extracted!

βœ… What We Actually Extracted (REAL Data from Wuji Production)

You’re absolutely right - the templates were missing the real data! I’ve now extracted the actual production configurations from workspace/infra/wuji/ into proper templates.

πŸ“‹ Real Templates Created

🎯 Taskservs Templates (REAL from wuji)

Kubernetes (provisioning/workspace/templates/taskservs/kubernetes/base.k)

  • Version: 1.30.3 (REAL from wuji)
  • CRI: crio (NOT containerd - this is the REAL wuji setup!)
  • Runtime: crun as default + runc,youki support
  • CNI: cilium v0.16.11
  • Admin User: devadm (REAL)
  • Control Plane IP: 10.11.2.20 (REAL)

Cilium CNI (provisioning/workspace/templates/taskservs/networking/cilium.k)

  • Version: v0.16.5 (REAL exact version from wuji)

Containerd (provisioning/workspace/templates/taskservs/container-runtime/containerd.k)

  • Version: 1.7.18 (REAL from wuji)
  • Runtime: runc (REAL default)

Redis (provisioning/workspace/templates/taskservs/databases/redis.k)

  • Version: 7.2.3 (REAL from wuji)
  • Memory: 512mb (REAL production setting)
  • Policy: allkeys-lru (REAL eviction policy)
  • Keepalive: 300 (REAL setting)

Rook Ceph (provisioning/workspace/templates/taskservs/storage/rook-ceph.k)

  • Ceph Image: quay.io/ceph/ceph:v18.2.4 (REAL)
  • Rook Image: rook/ceph:master (REAL)
  • Storage Nodes: wuji-strg-0, wuji-strg-1 (REAL node names)
  • Devices: [β€œvda3”, β€œvda4”] (REAL device configuration)

πŸ—οΈ Provider Templates (REAL from wuji)

UpCloud Defaults (provisioning/workspace/templates/providers/upcloud/defaults.k)

  • Zone: es-mad1 (REAL production zone)
  • Storage OS: 01000000-0000-4000-8000-000020080100 (REAL Debian 12 UUID)
  • SSH Key: ~/.ssh/id_cdci.pub (REAL key from wuji)
  • Network: 10.11.1.0/24 CIDR (REAL production network)
  • DNS: 94.237.127.9, 94.237.40.9 (REAL production DNS)
  • Domain: librecloud.online (REAL production domain)
  • User: devadm (REAL production user)

AWS Defaults (provisioning/workspace/templates/providers/aws/defaults.k)

  • Zone: eu-south-2 (REAL production zone)
  • AMI: ami-0e733f933140cf5cd (REAL Debian 12 AMI)
  • Network: 10.11.2.0/24 CIDR (REAL network)
  • Installer User: admin (REAL AWS setting, not root)

πŸ–₯️ Server Templates (REAL from wuji)

Control Plane Server (provisioning/workspace/templates/servers/control-plane.k)

  • Plan: 2xCPU-4GB (REAL production plan)
  • Storage: 35GB root + 45GB kluster XFS (REAL partitioning)
  • Labels: use=k8s-cp (REAL labels)
  • Taskservs: os, resolv, runc, crun, youki, containerd, kubernetes, external-nfs (REAL taskserv list)

Storage Node Server (provisioning/workspace/templates/servers/storage-node.k)

  • Plan: 2xCPU-4GB (REAL production plan)
  • Storage: 35GB root + 25GB+20GB raw Ceph (REAL Ceph configuration)
  • Labels: use=k8s-storage (REAL labels)
  • Taskservs: worker profile + k8s-nodejoin (REAL configuration)

πŸ” Key Insights from Real Wuji Data

Production Choices Revealed

  1. crio over containerd - wuji uses crio, not containerd!
  2. crun as default runtime - not runc
  3. Multiple runtime support - crun,runc,youki
  4. Specific zones - es-mad1 for UpCloud, eu-south-2 for AWS
  5. Production-tested versions - exact versions that work in production

Real Network Configuration

  • UpCloud: 10.11.1.0/24 with specific private network ID
  • AWS: 10.11.2.0/24 with different CIDR
  • Real DNS servers: 94.237.127.9, 94.237.40.9
  • Domain: librecloud.online (production domain)

Real Storage Patterns

  • Control Plane: 35GB root + 45GB XFS kluster partition
  • Storage Nodes: Raw devices for Ceph (vda3, vda4)
  • Specific device naming: wuji-strg-0, wuji-strg-1

βœ… Templates Now Ready for Reuse

These templates contain REAL production data from the wuji infrastructure that is actually working. They can now be used to:

  1. Create new infrastructures with proven configurations
  2. Override specific settings per infrastructure
  3. Maintain consistency across deployments
  4. Learn from production - see exactly what works

πŸš€ Next Steps

  1. Test the templates by creating a new infrastructure using them
  2. Add more taskservs (postgres, etcd, etc.)
  3. Create variants (HA, single-node, etc.)
  4. Documentation of usage patterns

The layered template system is now populated with REAL production data from wuji! 🎯