Solutions · Proxmox VE

Proxmox monitoring
that sees every guest.

One agent on the node and the whole hypervisor becomes observable: every VM and container with its own metrics, storage pools, cluster quorum, HA state, and the backup job that failed at 3 a.m. Read-only, configless, and guests are never billed.

0 tokens
The agent reads the node locally over a read-only allow-list. No API tokens, no cluster credentials uploaded.
1 host
A node is one billable host. Every guest it runs is monitored through it for free.
60 s
From install command to the first metrics: discovery finds Proxmox by itself.
Read-only
Nothing is written, nothing is restarted. The agent observes; you operate.

Every guest, without an agent in the guest

The node's agent enumerates all VMs and containers and reports per-guest CPU, memory, disk usage and uptime, tagged with the guest's id and name. A guest under memory pressure or a container filling its rootfs is visible from the node — no agent inside the guest, no exporter, no per-VM setup. Storage pools report utilization the same way, so local-lvm at 82% shows up before it becomes an outage.

The cluster, as each node sees it

On clustered nodes the agent reports quorum, member count and per-node liveness from that node's point of view — which means a split brain is visible as two nodes disagreeing, not hidden behind one API answer. HA resource errors are tracked, corosync is monitored as its own service, and cluster rules simply do not exist on single-node installs, so a homelab box never pages you about a quorum it does not have.

Failed backups become incidents, not folklore

A failed vzdump or migration opens an incident attributed to the node that actually ran the task, carrying the reason Proxmox itself recorded. The next successful run of the same job resolves it automatically. Backup freshness is checked too: a backup that silently stopped running is an alert, not a surprise during a restore.

DRBD and LINSTOR, if you run them

Replicated storage is monitored natively: resource and connection states, out-of-sync bytes that linger on a healthy link, quorum loss, satellites and storage pools. The transient snapshot resources that backup jobs create are recognized and excluded, so a nightly backup never mints a false out-of-sync alert.

Alert rules included

Quorum lost, HA resource errors, storage pool filling, failed tasks, host down in about a minute — calibrated on live fleets, tunable per team and per host.

The node is a server too

CPU, memory, disks, network, pending updates, SSH brute-force counters, certificates — the full host catalog applies to the node itself.

Alerts that reach a human

Telegram, Slack, Teams, email, escalation chains, on-call schedules — and a phone call that speaks, for the incidents that cannot wait.

Frequently asked

Do I need a Proxmox API token to monitor it?
No. The agent runs on the node and reads Proxmox locally through read-only commands from a fixed allow-list. No API token is created, no cluster credential leaves the node, and nothing is ever written or restarted.
Are my VMs and containers billed as hosts?
No. A Proxmox node with an agent is one host. Every VM and container it runs is monitored through that agent for free. A guest only becomes a billable host if you install its own agent inside it.
Does Wakora understand Proxmox clusters?
Yes. Each node reports quorum and membership as it sees them, so a split brain is visible by comparing views. HA resource errors, corosync health and failed cluster tasks are tracked, and a failed task is attributed to the node that actually ran it.

Three nodes fit in the free plan.

Install the agent on a node, watch discovery do the tour. No card, no trial clock.

Start free See pricing