Vm Provisioning Patterns

VM Provisioning Patterns

VM Existence Check Requirements

1. Each provision role MUST check for existing VM/container before creating. Guard ALL creation tasks with when: not vm_exists | bool (VMs) or when: not lxc_exists | bool (containers via proxmox_lxc).

2. Standard VM existence check pattern:

   - name: Check if VM already exists
     ansible.builtin.command:
       cmd: qm status {{ vm_id }}
     register: vm_status
     failed_when: false
     changed_when: false
   
   - name: Set VM existence flag
     ansible.builtin.set_fact:
       vm_exists: "{{ vm_status.rc == 0 }}"
   

VM Startup Configuration

3. Every VM MUST configure --onboot 1 --startup order=N via qm set. This task runs unconditionally to self-heal. Define <type>_vm_startup_order in role defaults.

4. Auto-start configuration pattern:

   - name: Configure VM to start on boot
     ansible.builtin.command:
       cmd: >-
         qm set {{ vm_id }}
         --onboot 1
         --startup order={{ vm_startup_order }}
   

Step-by-Step: Adding a New VM Type

5. Complete process for creating a new VM service (using homeassistant as example):

Create provision role:

roles/homeassistant_vm/
├── defaults/main.yml
├── meta/main.yml
└── tasks/main.yml

Create configure role:

roles/homeassistant_configure/
├── defaults/main.yml
├── meta/main.yml
└── tasks/main.yml

Add VMID to group_vars:

# inventory/group_vars/all.yml
homeassistant_vm_id: 200
homeassistant_vm_name: homeassistant
homeassistant_vm_memory: 2048
homeassistant_vm_cores: 2
homeassistant_vm_disk_size: 32G
homeassistant_image_path: images/haos.qcow2

Add to inventory and site.yml, update Molecule, create VM-specific skill.

Add Host Pattern

6. Dynamic inventory pattern:

   - name: Add VM to dynamic inventory
     ansible.builtin.add_host:
       name: "{{ vm_name }}"
       groups: dynamic_group
       ansible_host: "<bootstrap_ip>"
   

Design Principles

7. Bake, don't configure at runtime: Custom images are REQUIRED. Provision roles verify the image exists and hard-fail if missing. Configure roles NEVER install packages.

8. One path, no fallbacks: NEVER add stock/generic image fallback logic. One tested code path per feature. Missing prerequisites fail with an actionable error message.

9. Follow community standards: Check upstream tooling before writing custom workarounds.

Documented Exceptions

10. Three documented exceptions to bake principle (each MUST be explicitly documented):

    • Docker pull of pinned image tag: deterministic and versioned
    • Desktop LXC via build-images.sh: rootfs tarball with all packages baked in
    • Windows VMs via ISO + autounattend.xml: install-from-ISO IS the bake approach for Windows

11. Any OTHER runtime package installation is rejected. If you need a new package, add it to the image build script.

Per-Host VMID Collision Avoidance

12. When multiple hosts share a Proxmox node (test environments), a static VMID causes conflicts. Compute per-host VMID as base_id + groups['flavor_group'].index(inventory_hostname). Apply the SAME computation in provision, verify, and cleanup.

13. Previous bug: VMID 600 was used for both home and mesh1 on the same physical Proxmox host. The second qm create failed with "Configuration file already exists."

Upload Path Selection

14. NEVER upload large images (>5 GB) to /tmp/ on Proxmox — it's typically tmpfs with limited size (~7.8 GB). Use /var/tmp/ which lives on the root filesystem. This applies to both image uploads and qemu-img convert output.

15. Previous bug: qemu-img convert of an 18 GB Windows qcow2 to /tmp/ failed with "No space left on device".