Installing a Hypervisor Node¶
The MFCloud Hypervisor ISO is an air-gapped installer based on Rocky Linux 9. It produces a fully self-contained KVM/QEMU host with libvirt, OVS/OVN, ZFS, Ceph, and LINSTOR satellite components pre-staged.
What you need¶
- A blank server with VT-x/AMD-V virtualization extensions enabled in BIOS.
- At least 64 GB RAM, 2× SSD/NVMe, 2× NICs (recommended).
- The
mfcloud-v27-enterprise.isobuild artifact. - An IP address plan: either a DHCP lease for the management interface, or a static IP / netmask / gateway / DNS.
Step 1 — Boot the ISO¶
Burn the ISO to USB or attach via iDRAC/IPMI virtual media, then boot. The Anaconda installer launches automatically.
Step 2 — Network configuration menu¶
The installer pauses on tty6 with this prompt:
==================================================================
MFCloud V27 Hypervisor — Management Network Setup
==================================================================
[1] DHCP (auto-configure from the local network)
[2] Static IP (manual address, gateway, DNS, hostname)
- DHCP — accept the default. Optionally type a hostname.
- Static — enter the IP, netmask, gateway, DNS server, and hostname when prompted.
After 4 seconds the screen flips back to the install progress.
Step 3 — Wait for first boot¶
When install completes the host reboots. On first boot, mfcloud-firstboot.service runs once. It:
- Loads the LINSTOR satellite container image.
- Generates a per-node ed25519 keypair and self-signed TLS cert.
- Generates a random password for the Emergency Host UI and prints it once on the install console:
================================================================
MFCloud Emergency Host UI — first-boot credential
----------------------------------------------------------------
URL: https://<node-ip>:8443/
User: root
Password: <generated-password>
RECORD THIS NOW. It will not be shown again.
================================================================
Capture the password now
The cleartext is printed exactly once. If you missed it, see Reset the Emergency UI Password.
Step 4 — Verify the Emergency UI¶
Browse to https://<node-ip>:8443/ and log in as root with the captured password. You should see CPU, memory, disk, and VM stats. Accept the self-signed certificate warning.
Step 5 — Onboard the node into the cluster¶
In the master console, Inventory → Add Host, then enter the node's IP. The master will SSH in (or use the Agent Trust API if configured) and register it as a compute node.