Proxmox Cleanup Safety

Proxmox Cleanup & Maintenance Rules

Cleanup Completeness Requirement

1. When ANY role deploys a file to the Proxmox host, ALWAYS add it to the removal list in BOTH cleanup playbooks (molecule/default/cleanup.yml AND playbooks/cleanup.yml).

Current Ansible-Managed Files

2. Current ansible-managed files that must be cleaned:
   • /etc/network/interfaces.d/ansible-bridges.conf (bridge config, may be modified to inet dhcp)
   • /etc/network/interfaces.d/ansible-proxmox-lan.conf (legacy LAN management IP, superseded)
   • /etc/network/interfaces.d/ansible-temp-lan.conf (test workaround, cleaned up)
   • /etc/modprobe.d/blacklist-wifi.conf (WiFi driver blacklist)
   • /etc/modprobe.d/vfio-pci.conf (PCI passthrough config)
   • /etc/ansible/facts.d/vm_builds.fact (deploy stamp tracking)
   • /etc/apt/sources.list.d/pve-no-subscription.sources (added by proxmox_igpu)
   • /tmp/openwrt-router-*.img* (left behind if build fails mid-upload)
   • /var/lib/vz/template/cache/debian-*.tar.zst (LXC templates)
   • Enterprise repos: restore pve-enterprise.sources.disabled → .sources and ceph.sources.disabled → .sources

Local State Files Cleanup

3. Local state files that must be cleaned (via delegate_to: localhost):

   • .state/addresses.json (cached host IPs)

4. Previous bug: ansible-proxmox-lan.conf was deployed but not cleaned up, leaving stale LAN management IPs across test runs.

Test Machine Protocol

5. Before running destructive operations (cleanup, VM destroy):
   1. Confirm the target is the test machine (check PROXMOX_HOST env var)
   2. Verify a backup exists (check for manifest.json in backup dir)
   3. Use the cleanup.sh wrapper which enforces env file sourcing

PCI Device Cleanup Requirement

6. Devices bound to vfio-pci do NOT auto-revert when the VM is destroyed. Without cleanup, the next run can't detect WiFi hardware.

7. Required PCI cleanup sequence:

   # 1. Unbind all vfio-pci devices
   for dev in /sys/bus/pci/drivers/vfio-pci/0000:*/; do
     addr=$(basename "$dev")
     echo "$addr" > /sys/bus/pci/drivers/vfio-pci/unbind
   done
   
   # 2. Remove blacklist and vfio config files
   rm -f /etc/modprobe.d/blacklist-wifi.conf /etc/modprobe.d/vfio-pci.conf
   
   # 3. Rebind WiFi via sysfs (NEVER modprobe -r)
   # Use tasks/sysfs_wifi_rebind.yml or inline:
   for phy in /sys/class/ieee80211/phy*; do
     PCI=$(basename $(readlink -f "$phy/device"))
     DRV=$(basename $(readlink -f "$phy/device/driver"))
     echo "$PCI" > /sys/bus/pci/drivers/$DRV/unbind
     echo "" > /sys/bus/pci/devices/$PCI/driver_override
   done
   
   # 4. Rescan PCI bus + explicit bind
   echo 1 > /sys/bus/pci/rescan
   sleep 1
   echo "$PCI" > /sys/bus/pci/drivers/$DRV/bind
   

8. All steps are required. The explicit bind (step 4) is critical -- echo 1 > /sys/bus/pci/rescan alone is insufficient because the kernel won't auto-bind drivers that were explicitly unbound.

9. WiFi module unload (modprobe -r iwlwifi/iwlmvm) is BANNED on ALL hardware. On AMD APUs (Raven Ridge), the NBIO handles ALL PCIe, USB, and SATA on a shared die — a WiFi module unload triggers a PCIe reset that kills USB ethernet hours later. Use sysfs unbind + PCI rescan + explicit bind instead (tasks/sysfs_wifi_rebind.yml). This pattern is universally safe on Intel and AMD, with no hardware-specific branching.

GPU Driver Cleanup — Separate from WiFi

10. GPU driver cleanup (i915/amdgpu) is DIFFERENT from WiFi cleanup. GPU modprobe -r amdgpu on a single-GPU AMD host causes a kernel panic (sole framebuffer removal).

11. NEVER run modprobe -r amdgpu or modprobe -r i915 in broad-scope plays (hosts: proxmox*). For E2E cleanup, PCI bus rescan after vfio-pci unbind is sufficient — skip GPU driver unload.

12. ONLY run GPU driver unload in per-feature cleanup (e.g., sunshine-vm, gaming-rollback) gated on lspci | grep -c 'VGA compatible controller' >= 2.

13. tests/test_host_safety.py is a static linter that catches modprobe -r amdgpu/i915 in broad-scope plays without VGA guards. Run pytest tests/ to catch this at dev time.

14. Previous bug: E2E cleanup ran modprobe -r amdgpu on ALL hosts including ai (single AMD GPU, USB ethernet). Kernel panicked, host crashed. Required physical power-on 3000 miles away.

Host Recoverability

14. Every host MUST declare wol_capable (true/false) in inventory/host_vars/. Cleanup MUST NOT run operations that could crash hosts with wol_capable: false.

15. Non-WoL hosts (USB ethernet — e.g., ai) cannot be recovered remotely. A kernel panic or shutdown is a production-breaking incident requiring physical access.

16. tests/test_wol.py enforces WoL exclusion. tests/test_host_safety.py enforces GPU safety. Both run as part of pytest tests/.