Your Cart
Loading

Cluster Without Fear: Proxmox VE Clusters, Quorum, Shared Storage, Ceph, Live Migration and High Availability for a Homelab That Survives a Dead Node

On Sale
$19.00
$19.00
Added to cart

Cluster Without Fear is a practical guide to turning two or three Proxmox VE hosts into a cluster that keeps running when one of them dies. It starts with what a cluster actually gives you (one web interface, live migration, high availability) and what it does not (a cluster is not a backup), and sets out the three honest designs as one decision table: two nodes plus a QDevice with ZFS replication, three nodes with ZFS replication, or three or more nodes with Ceph. Then it builds the cluster one decision at a time. Planning covers a dedicated corosync link with a redundant fallback, the latency the docs ask for, fixed hostnames and addresses, time sync, matching versions, and why a joining node must be empty. Creating the cluster covers pvecm create and add, what /etc/pve really is, editing corosync.conf the safe way and removing a node without leaving a mess. The quorum chapter explains votes and expected votes, why two nodes lose quorum, how a QDevice on a Raspberry Pi fixes it, and why pvecm expected 1 is an emergency lever only. Storage covers local versus shared, ZFS replication schedules and the data you lose on failover, NFS for ISOs and backups, and iSCSI. Ceph gets the hardware the documentation asks for, pveceph step by step, size 3 / min_size 2 and OSD replacement. Then live migration and CPU types, HA with the node and resource affinity rules that replaced HA groups in Proxmox VE 9, fencing, shutdown policies and maintenance mode, the day-2 routine of rolling updates, backups and alerts, and a calm recovery playbook for lost quorum, split networks and dead nodes. Every version, default and flag was checked against the official Proxmox VE 9.2 and Ceph documentation in October 2026. No prices, uptime or speed claims.

What you get

  • What a cluster gives you and what it does not, and three honest designs in one decision table
  • Planning: a dedicated corosync link plus a fallback, the latency the docs ask for, fixed names and addresses, chrony, matching versions, CPU vendors, and the empty-joining-node rule
  • Creating the cluster: pvecm create and pvecm add, the join information in the web interface, what /etc/pve (pmxcfs) is, editing corosync.conf the safe way, removing and re-adding a node
  • Quorum and the QDevice: votes and expected votes, read-only /etc/pve without quorum, corosync-qnetd on a Raspberry Pi, pvecm qdevice setup, failure tests, and pvecm expected 1 as an emergency lever only
  • Storage for a cluster: local vs shared, ZFS replication with pvesr (schedules, rate limits, the data-loss window), NFS and SMB for ISOs and backups, iSCSI with shared LVM, and the nodes and shared properties in storage.cfg
  • Ceph on Proxmox VE: the documented hardware requirements, pveceph install with the repository choice, monitors, managers and OSDs, size 3 / min_size 2 pools, RBD and CephFS, the PG autoscaler, health checks and replacing an OSD
  • Migration: live migration requirements, --with-local-disks, choosing a CPU type every node supports, container restart migration, a migration network, and emptying a node with pvenode migrateall
  • High availability: pve-ha-crm and pve-ha-lrm, the watchdog and fencing, the Proxmox VE 9 node affinity and resource affinity rules, a node failure step by step, shutdown policies, maintenance mode and the 9.2 disarm-ha switch
  • Day 2 and recovery: rolling updates one node at a time, cluster-wide backups to Proxmox Backup Server, the health signals worth an alert, the datacenter firewall, ACME certificates, and recovery steps for lost quorum, split networks, failed OSDs and dead nodes
  • Two complete reference builds as commands (three nodes with Ceph, two nodes with a QDevice and replication), a cluster checklist and a cheat sheet

74 pages, 24 diagrams plus a chapter-opener graphic for every chapter, a cluster checklist and a cheat sheet. PDF and EPUB. Homelab the Right Way, Book 13.

Not sure yet? Read the free sample first: the cover, the full table of contents and all of Chapter 3, "Quorum and the QDevice" (votes, expected votes and quorum, what a node does without quorum, why two nodes need a third vote, how the QDevice works, setting up corosync-qnetd on a Raspberry Pi with pvecm qdevice setup, verifying it, failure tests, living with a QDevice, and pvecm expected 1 as an emergency-only lever), at https://payhip.com/b/Fkx2c


Also in a bundle: Everything Bundle: every ebook in the store, $89 ยท Homelab Bundle, $59.

You will get the following files:
  • PDF (3MB)
  • EPUB (3MB)