====== klynge001 ======

klynge001 is the Proxmox VE cluster used for Hackeriet service VMs. This page is an operational map and emergency orientation page, not the inventory source of truth. Keep canonical device, cable, IP, rack, and VM inventory in NetBox.

===== Maintenance =====

  * [[infra:operations:proxmox-maintenance|Proxmox maintenance runbook]]
  * [[infra:operations:proxmox-backups|Proxmox backup notes]]

===== Known nodes =====

Verified while documenting:

  * [[infra:hosts:host006|host006]] - Proxmox VE host, management IP 10.10.50.26/24
  * [[infra:hosts:host007|host007]] - Proxmox VE host, management IP 10.10.50.27/24

Mentioned in older handover notes but not yet verified in this documentation pass:

  * host008

===== Access =====

  * Proxmox UI: use the host-specific UI, for example https://host006.hackeriet.no:8006/ or https://host007.hackeriet.no:8006/.
  * SSH: use the host-specific root account, for example root@host006.hackeriet.no or root@host007.hackeriet.no.
  * Connectivity: use the Hackeriet internal network or VPN before assuming a host is down.
  * Credentials: use the Hackeriet password store. Do not copy secrets into the wiki.

===== Cluster state observed =====

Observed from both host006 and host007 during documentation:

  * Cluster name: klynge001
  * Quorate: yes
  * Expected votes: 3
  * Total votes: 2
  * Members observed: 10.10.50.26 and 10.10.50.27

The expected-votes/total-votes mismatch should be understood before planned maintenance. Do not reboot or isolate a node casually if quorum is already marginal.

===== Useful host-level checks =====

Run on a Proxmox host when investigating cluster problems:

  * hostname -f
  * pveversion
  * pvecm status
  * systemctl --failed
  * ip -br addr
  * df -h
  * pvesm status
  * qm list

===== Guest handling =====

Do not touch guest VMs unless the incident requires it. Use this page to find the host, then use the host page and NetBox to orient before making changes.

Service VM pages should document service-specific emergency access. Host pages should document how to reach the hypervisor. NetBox should remain the canonical inventory.