Skip to content

Integrate with Awanio Cockpit ​

With a Cockpit target, Condensa migrates VMs onto the fleet of KVM hosts managed by the Cockpit control plane. You choose the target host, storage pool, and network per migration, and migration transfer jobs appear in Cockpit's Tasks panel automatically.

Prerequisites ​

  • A running Cockpit control plane, reachable from the Condensa server over HTTPS.
  • The target hosts are registered in Cockpit (added through Cockpit's normal host-provisioning flow).
  • A Cockpit service account (access key + secret) for Condensa.
  • For direct-pull transfers and warm migration: VDDK installed on the target hosts and network access from those hosts to ESXi — see About VDDK.
  • For the staged fallback (used automatically when a host has no VDDK; cold migrations only): the hosts must be able to reach CONDENSA_PUBLIC_URL.

1. Create a service account in Cockpit ​

In Cockpit, create a service account for Condensa and note its access key and secret.

2. Register the provider in Condensa ​

  1. Go to Providers → Add Provider.
  2. Choose vendor Awanio Cockpit.
  3. Fill in:
    • Label — a name you'll recognize in the wizard (e.g. cockpit-prod)
    • API URL — the Cockpit API base, e.g. https://cockpit.example.com/api/v1 (the origin alone also works — Condensa appends /api/v1)
    • Access key / secret — the service account credentials
  4. Save, then click Test on the provider card — it must report success before migrations can run.

3. Run a migration ​

In the migration wizard select your VMware source, pick the VMs, then:

  • Target provider — the Cockpit provider you registered
  • Target host — which KVM host receives the VM
  • Storage pool — where the converted qcow2 disks land
  • Network — the switch the VM's NIC attaches to (default None, to avoid IP conflicts with a still-running source)

Warm migration is offered for powered-on VMware VMs with CBT available; the wizard flags each VM's readiness.

How the data moves ​

EnginePathUsed when
VDDK direct pullTarget host pulls from ESXi, converts to qcow2 on the flyHost has VDDK (preferred; required for warm)
Staged transferCondensa exports via vSphere, converts, host downloadsAutomatic fallback, cold only

The engine that actually ran is shown on the migration detail page as the Method badge.

Firmware (UEFI/BIOS) and Secure Boot ​

Condensa detects the source VM's boot firmware and creates the target to match — a UEFI guest gets a UEFI (OVMF) target, a legacy guest gets BIOS. This is automatic; a UEFI source is noted in the migration log. No action is needed for the VM to boot with the correct firmware.

Secure Boot is deliberately left off on the migrated VM. Secure Boot state lives in the firmware's NVRAM, not on the disk, so it does not travel with the migrated disk. Turning it on against a fresh target NVRAM — whose enrolled keys and OVMF build may not match what the guest was signed against — commonly prevents boot. A guest that was installed under Secure Boot still boots normally with it off: Secure Boot only verifies boot signatures, it is not required to boot.

If a specific guest genuinely requires Secure Boot (for example a compliance policy), enable it on the VM in Cockpit, after migration:

  1. Power off the migrated VM in Cockpit.
  2. Open the VM and edit Firmware & Options.
  3. Enable Secure Boot — and TPM if the guest needs it (e.g. Windows 11). Secure Boot requires the q35 machine type on x86_64.
  4. Make sure the target host has a Secure Boot-capable OVMF firmware installed.
  5. Start the VM.

See Cockpit → Virtual Machines for the Firmware & Options controls.

Windows + BitLocker

If the guest uses BitLocker tied to TPM/Secure Boot, any firmware change can trigger a one-time BitLocker recovery-key prompt on next boot. Have the recovery key on hand before changing firmware settings.

CPU model ​

Condensa creates every migrated VM with CPU mode host-model: the guest sees the target host's CPU model and features instead of the generic qemu64 model. RHEL 9 and the distributions built from it (CentOS Stream 9, Rocky Linux 9, AlmaLinux 9) need this to boot. host-model keeps the VM live-migratable between hosts with the same CPU generation. The VM starts with the source VM's vCPU count; its CPU topology follows the vCPU maximum Cockpit sets for the VM.

Condensa sets the CPU mode through the VM's settings right after it creates the VM, before the VM is first started. This needs Cockpit v2.0.3 or later, and a service account that may change VM settings (VirtualMachine.Config.Settings). If Cockpit refuses the change, the migration still completes and its log shows a warning; set CPU Mode to host-model in the VM's Edit Settings → CPU Configuration before you start it.

Troubleshooting ​

  • Provider test fails — check the API URL, service account credentials, and that Cockpit's TLS certificate is accepted.
  • Import jobs fail with a thumbprint/VDDK error — run Test Connection on the VMware provider first: since 1.9.2 it reports a stored thumbprint the host no longer matches (an ESXi reinstall or upgrade replaces the certificate). Then verify VDDK on the target host (Vapor UI → System → Third-party Libraries) and that the host can reach ESXi on port 443/902.
  • Staged transfer used when you expected direct pull — the target host has no VDDK or cannot reach ESXi; the migration still completes cold via staging.
  • Migrated VM shows "no bootable device" / an empty disk — the target firmware must match the source. Condensa sets this automatically, but migrations created before firmware carry-over was added default to BIOS. Re-create the migration so the source firmware is detected, or switch the VM's firmware to UEFI in Cockpit (Firmware & Options).
  • A migrated RHEL 9, CentOS Stream 9, Rocky Linux 9 or AlmaLinux 9 VM stops after Probing EDD with its CPU at 100% — the VM runs on the generic qemu64 CPU model, which these guests cannot boot on. VMs migrated by Condensa 1.10.0 or earlier have that model. Power the VM off, open Edit Settings → CPU Configuration, set CPU Mode to host-model, and start it again.