inventory-sync

Inventory Sync — Olympus Cluster

You are helping maintain inventory consistency for the olympus Proxmox cluster. Hardware changes require updating multiple files — this skill ensures nothing is missed.

Inventory File Map

The olympus cluster inventory spans these files:

ansible/
├── inventory/
│   ├── hosts.yml                          # Group memberships, ansible_host IPs
│   └── host_vars/
│       ├── r420.infiquetra.com.yml        # Ceph bulk OSDs, ceph_nic
│       ├── r640-1.infiquetra.com.yml      # Ceph fast OSDs, ceph_nic
│       ├── r640-2.infiquetra.com.yml
│       ├── r640-3.infiquetra.com.yml
│       ├── r720xd.infiquetra.com.yml
│       └── r820.infiquetra.com.yml
├── inventory/group_vars/
│   └── all/
│       └── all.yml                        # Vault-encrypted shared vars
└── roles/*/defaults/main.yml              # Role-level defaults

Hardware Onboarding Checklist

When adding a new Proxmox node, update every file in this list:

1. ansible/inventory/hosts.yml

# Add to ALL applicable groups:
proxmox_hosts:
  hosts:
    <new-hostname>.infiquetra.com:
      ansible_host: 10.220.1.<x>

proxmox_nodes:        # If not the master
  hosts:
    <new-hostname>.infiquetra.com:

2. ansible/inventory/host_vars/<new-hostname>.infiquetra.com.yml

Create this file with the full disk/NIC definition. Required variables:

# Ceph cluster network NIC (the 10GbE interface)
ceph_nic: nic0          # e.g. nic0, nic4 — check 'ip link' on the physical host

# Ceph OSD disk definitions
ceph_disks:
  - name: nvme0n1       # kernel device name (no /dev/ prefix)
    device_class: ssd   # 'ssd' or 'hdd'
  - name: sdc
    device_class: ssd

# iDRAC management IP (for IPMI monitoring)
idrac_ip: 10.220.1.<management_ip>

For fast-tier nodes (R640s): use device_class: ssd or device_class: nvme For bulk-tier nodes (r420, r720xd, r820): use device_class: hdd

3. Monitoring scrape targets

The Prometheus config in roles/monitoring/files/prometheus/prometheus.yml has static scrape targets. Add the new node's IP to:

• node_exporter job — scrape <ip>:9100
• ipmi_exporter job — add iDRAC IP to the targets list
• (If Proxmox host) pve_exporter job — add to proxmox targets

4. CLAUDE.md hardware table

Update the hardware table in CLAUDE.md with the new node's specs.

5. Ansible Vault (if node has unique credentials)

If the new node uses a different iDRAC password, add to vault:

ansible-vault edit inventory/group_vars/all/all.yml --vault-password-file ~/.vault_pass.txt

---

Hardware Decommission Checklist

When removing a node, clean up these locations:

• Remove from inventory/hosts.yml (all groups)
• Delete inventory/host_vars/<hostname>.infiquetra.com.yml
• Remove scrape targets from monitoring Prometheus config
• Remove from Ceph cluster: ceph osd out osd.<id> then drain PGs
• Remove from corosync: pvecm delnode <hostname>
• Remove from CLAUDE.md hardware table
• Verify no role defaults/main.yml references the hostname directly

---

Variable Dependency Map

Key variables and where they flow:

Variable	Defined In	Used In
ansible_host	hosts.yml	All SSH connections
ceph_nic	host_vars/<host>.yml	proxmox_network/tasks/setup.yml — creates vmbr1 bridge
ceph_disks	host_vars/<host>.yml	proxmox_ceph/tasks/setup.yml — creates OSDs
idrac_ip	host_vars/<host>.yml	monitoring/files/prometheus.yml — IPMI scrape
proxmox_master	group_vars/all/all.yml	delegate_to: in cluster/ceph roles
proxmox_cluster_name	group_vars/all/all.yml	proxmox_cluster/tasks/setup.yml — pvecm create
vault_*	group_vars/all/all.yml (encrypted)	Roles that need credentials

---

Consistency Validation Commands

Check all hosts in a group have the same required variables:

# Verify all proxmox_hosts have ceph_nic defined
ansible proxmox_hosts -i inventory/hosts.yml -m debug \
  -a "msg={{ ceph_nic | default('MISSING') }}"

# Verify all proxmox_hosts have ceph_disks defined
ansible proxmox_hosts -i inventory/hosts.yml -m debug \
  -a "msg={{ ceph_disks | default('MISSING') }}"

# Check ansible connectivity to all hosts
ansible all -i inventory/hosts.yml -m ping

---

IP Address Reference

Current cluster IPs — verify consistency across hosts.yml, host_vars, and monitoring configs:

Host	Main IP	Ceph IP	iDRAC IP
r420	10.220.1.7	10.220.2.7	10.220.1.17
r640-1	10.220.1.8	10.220.2.8	10.220.1.18
r640-2	10.220.1.9	10.220.2.9	10.220.1.19
r720xd	10.220.1.10	10.220.2.10	10.220.1.20
r820	10.220.1.11	10.220.2.11	10.220.1.21
r640-3	10.220.1.12	10.220.2.12	10.220.1.22

VM IPs (agent VMs 100-107 → 10.220.1.50-57, service VMs 200-203 → 10.220.1.60-63)

When changing an IP, search for stale references:

grep -r "<old_ip>" ansible/  # Find all files referencing the old IP