-
Notifications
You must be signed in to change notification settings - Fork 0
Features
Ferrum connects to any number of Proxmox VE clusters, standalone hosts, and Proxmox Backup Server (PBS) instances at once, and rolls them up into one fleet-wide view. Everything below is scoped per-connection where that makes sense, and fleet-wide everywhere it doesn't.
The landing page is a fleet-wide rollup: server/node/guest counts, weighted CPU/memory/storage utilization, a sortable cluster-comparison table, a fleet health heatmap, "hottest guests" by live CPU, and active alerts — auto-refreshing every 15s. Custom Dashboard is a separate, drag-and-drop-arranged dashboard built from 20+ widgets (capacity forecasts, fleet health score, storage treemap, uptime leaderboard, utilization heatmap/histogram, node/cluster comparisons, running tasks, backup activity, and more), each independently scoped to one connection or "all".
Inventory lists every node, VM, and LXC across every connection, with search/filter by status, type, and resource pool, and live console/shell access (noVNC for VNC/SPICE-style consoles, xterm.js for a shell) straight from the guest row. Topology renders the same estate as an interactive force-directed graph — connections → nodes → guests — with draggable, position-remembered cards and SVG export.
Local and shared/external storage capacity per connection, a Ceph health card (cluster status, pools, OSDs) where Ceph is present, and a "fullest pools" ranked view across the whole fleet.
Schedule vzdump backup jobs (or run one immediately) across any connection, see every configured job's schedule/storage/guest scope, and view storage replication (pvesr) job status per node.
- High Availability — HA resources, groups (PVE 8) or node-affinity rules (PVE 9's replacement for groups), and live HA manager/CRM/LRM status, across every connection.
- Firewall — cluster-wide and per-node rules, aliases, and IP sets; a guest's own firewall rules are managed from its detail dialog in Inventory.
- Cluster & SDN — cluster membership (create a cluster, join an existing one, view/remove members) and SDN zones/VNets/subnets/controllers/IPAMs.
- Alerts — threshold-based rules on node/guest CPU, memory, disk, and storage pool usage, plus built-in system alerts (certificate expiry, connection unreachable, orphaned storage volumes) that don't need a rule configured.
- Config drift detection — flags nodes/guests whose live configuration has drifted from a saved baseline.
- Guest lifecycle policies — automated actions on guests matching a policy (e.g. stale template cleanup).
- Capacity forecasting & fleet health score — trend-based projections of when a resource runs out, and a single rolled-up health number for the fleet.
- Scheduled fleet health digests — a periodic email summary of fleet health, sent via the configured SMTP settings.
- Terraform/Ansible export — generate infrastructure-as-code inventory from the live fleet.
Migrate a guest between two different Proxmox clusters (not just between nodes of the same cluster) directly from Ferrum, and manage Proxmox Backup Server remotes alongside native PVE connections — both show up in the same Inventory/Topology/Storage views.
Task Center is a fleet-wide view of running and recent PVE tasks (backups, migrations, reboots, ...) across every connection. Audit Log (admin-only) records who did what and when for every mutating action — through the UI, the REST API, or MCP.
A chat assistant with tool-calling access to your fleet (inventory lookups, guest actions, and more), backed by any OpenAI-compatible provider you configure, including a built-in local model with no API key or network required. See AI Assistant and MCP.
Run one action (start/stop/migrate/tag/...) across a filtered selection of guests spanning multiple connections at once, instead of one-at-a-time from Inventory.
Outbound webhooks deliver real-time events (guest state changes, alerts firing, task completion, ...) to any HTTP endpoint you configure, with a delivery log and a "send test event" button. Separately, any user can generate scoped API keys (Profile > API Keys) for the general REST API or the MCP endpoint — both are off by default and enabled per-instance from Settings > API & MCP.
Per-user roles, optional OIDC single sign-on (any standard provider — Keycloak, Authentik, Entra ID, Okta, ...) alongside local username/password login, and certificate/session monitoring across connections.
A dozen look-and-feel presets (Enterprise, Proxmox-native, Terminal, Glass Flight Deck, Midnight, Paper, Glassmorphism, Neumorphism, Brutalist, Solarized, High Contrast, Aurora), each with independent light/dark theming, a choice of 8 accent colors, and a density toggle (comfortable/compact) — all saved to the account and followed across devices.