Install Proxmox VE with Ceph — hyperconverged cluster

Proxmox VE with Ceph gives you a hyperconverged cluster: compute and replicated storage on the same nodes, no SAN and no single point of failure. A practical guide from scratch, with commands.

Why Proxmox + Ceph

Ceph spreads data across the disks of all nodes, with 3x replication. Lose a disk or a whole node — the VMs keep running from the copies on the other nodes and can migrate automatically (HA). No external array, no licenses.

Requirements

At least 3 nodes (Ceph quorum + real HA).
A dedicated network for Ceph, ideally 10GbE — replication generates traffic.
Raw disks for OSDs (SSD/NVMe for performance) — Ceph manages them directly.
A separate network for corosync (cluster) so it doesn't compete with storage.

Ceph wants RAW disks, not hardware RAID. Put the controller in HBA/IT (pass-through) mode. RAID under Ceph breaks replication and performance.

Architecture

1. Repo + update on each node

# pe FIECARE nod: repo no-subscription (fara licenta) + update
echo 'deb http://download.proxmox.com/debian/pve bookworm pve-no-subscription' \
  > /etc/apt/sources.list.d/pve-no-sub.list
apt update && apt -y full-upgrade
reboot

2. Create the Proxmox cluster

# pe nod1: creeaza clusterul
pvecm create CYMMUNITY-CLUSTER

# pe nod2 si nod3: alatura-te (IP-ul nodului 1)
pvecm add 10.10.20.11

# verifica (de pe orice nod)
pvecm status

3. Install + init Ceph

# pe FIECARE nod: instaleaza Ceph (reef, repo no-subscription)
pveceph install --repository no-subscription

# initializeaza Ceph pe nod1 — retea dedicata pt replicare
pveceph init --network 10.10.30.0/24

4. Monitors + Managers

# monitor + manager pe fiecare nod (quorum = 3)
pveceph mon create
pveceph mgr create
ceph -s   # HEALTH_OK cand cei 3 MON sunt in quorum

5. Add OSDs (disks)

# adauga discurile brute ca OSD (repeta per disc, per nod)
pveceph osd create /dev/nvme0n1
pveceph osd create /dev/nvme1n1

ceph osd tree   # arata OSD-urile per nod

6. RBD pool for VMs

# pool RBD pt discurile VM-urilor (replica x3, min 2)
pveceph pool create vm-storage --size 3 --min_size 2 --application rbd

# se adauga automat ca storage 'vm-storage' in Datacenter -> Storage

7. High Availability

# activeaza HA pentru o masina virtuala (ex. VMID 100)
ha-manager add vm:100 --state started
ha-manager status

8. Verify

ceph -s                 # HEALTH_OK, PG-uri active+clean
ceph osd df tree        # utilizare per OSD/nod
pvecm status            # quorum cluster Proxmox
ceph df                 # spatiu pool-uri

Conclusion

You now have a Proxmox + Ceph cluster resilient to disk and node failures, with HA and replicated storage — suitable from 3 nodes up. We design, size and monitor it for you.