声明式管理 Proxmox API 对象与 Samba DNS 上游
yaml / yaml (pull_request) Failing after 25s
terraform / validate (pull_request) Successful in 51s
ansible / collection-test (pull_request) Successful in 1m27s
ansible / lint (pull_request) Successful in 3m45s

This commit is contained in:
2026-09-17 13:22:50 +00:00
parent 71eea7d8fc
commit 2e05b1a96a
15 changed files with 277 additions and 106 deletions
@@ -14,12 +14,26 @@ pve_ha_resources:
max_restart: 3
max_relocate: 2
# pve3 is the normal home for the critical router. Non-strict preserves pve1
# and pve2 as failover targets when pve3 is unavailable.
pve_ha_rules:
- name: prefer-vyos-pve3
type: node-affinity
resources: ["vm:100"]
nodes: ["pve3:100", "pve1:10", "pve2:10"]
strict: false
comment: "Prefer critical VyOS router on pve3, away from CI nodes"
pve_ha_portable_vms:
- vmid: 100
node: pve3
cpu: x86-64-v2-AES
# --- Watchdog ---------------------------------------------------------------
# ⚠ WHY THIS MATTERS: adding ANY HA resource arms fencing cluster-wide. PVE's
# default is `softdog`, a SOFTWARE watchdog — a kernel timer, which CANNOT fire if
# the kernel itself is frozen. That is precisely the failure this cluster has
# actually seen (pve2's Raven Ridge idle freeze). A hardware watchdog is
# independent silicon and fires regardless.
# the kernel itself is frozen. A hardware watchdog is independent silicon and
# fires regardless.
#
# Verified available 2026-07-26:
# pve1 (Intel i3-6100U) -> iTCO_wdt (timeout 30s)