High availability

Exovo's container-first design makes the entire phone system — telephony core, web console, database — one VM running one compose stack. That's the high-availability story: instead of a proprietary clustering layer, the PBX is a perfect citizen of the hypervisor HA you already operate, and FreeSWITCH's runtime configuration regenerates from the database on startup, so a recovered VM comes back complete.

The supported reference architecture

Run the Exovo VM under your hypervisor's HA with the same address and disk, and a host failure recovers without any Exovo-specific steps:

• Proxmox VE — add the VM to an HA group. Shared storage (Ceph, NFS, iSCSI) gives zero data loss; ZFS storage replication gives near-zero at lower cost (loss bounded by the replication interval). On host failure the VM restarts on a surviving node.
• VMware vSphere HA — the VM restarts on another host in the cluster from shared storage. vSphere FT is unnecessary — a VM restart is well within telephony tolerances.
• Hyper-V Replica / failover clustering — a clustered VM on Cluster Shared Volumes restarts on failure; Hyper-V Replica alternatively maintains an async copy to promote (data loss bounded by the replication frequency).

What recovery looks like: the VM boots, containers start, FreeSWITCH regenerates its configuration from the database, and phones re-register — the same address is simply back. Expect a recovery time of a few minutes (VM boot + stack start) and a recovery point of zero on shared storage or one replication interval on async replicas. Calls in progress at the moment of failure drop; new calls flow as soon as registrations return.

Phone failover while the PBX is down

Desk phones ride out a short VM restart on their own (they retry registration). For anything longer, give phones a backup SIP server under Admin → Connectivity → Network: Yealink desk phones, Yealink DECT bases and Grandstream phones are then provisioned with a secondary registrar they fail over to when the primary stops answering — and fail back automatically. Phones connected through an SBC or registering to the FQDN don't need this; their failover is a DNS repoint.

Backups are the floor

Hypervisor HA protects against host failure, not against data corruption, ransomware, or fat-fingered deletion. Keep scheduled encrypted backups to remote storage regardless — a backup restores a complete system onto any fresh VM, which is also your recovery path when there's no cluster to fail over inside.

What "supported" means

Running Exovo under hypervisor HA per this page is a supported configuration — it needs no special license, no second key, and support treats a post-failover system as a normal system. Exovo doesn't currently ship its own replication layer; a built-in warm-standby mode (a second box continuously staging the latest backup, promoted on failure) is on the roadmap.