Integrations

Proxmox VE

Create a limited API token and monitor a node, VM, or LXC.

NasDash calls the Proxmox HTTPS API on port 8006. It constructs the authorization header from the Token ID and Token Secret entered separately.

Create the token

A dedicated read-only user and its own token, with Privilege Separation kept on. The token then holds only the permissions given to it, never more than its user's. In Proxmox VE, under Datacenter → Permissions:

  1. Users → Add: a user such as nasdash in the pve realm (no password needed for API use).
  2. Add → User Permission: path /, that user, role PVEAuditor, Propagate on.
  3. API Tokens → Add: that user, Token ID monitoring, Privilege Separation checked. Copy the secret at once: it is shown only once.
  4. Add → API Token Permission: path /, the token nasdash@pve!monitoring, role PVEAuditor, Propagate on.

PVEAuditor is read-only and includes Sys.Audit, VM.Audit and Datastore.Audit. To limit the scope, give both permissions on /nodes/<node> and the /vms/<vmid> to watch instead of /. The same from a node's shell:

pveum user add nasdash@pve
pveum acl modify / --users nasdash@pve --roles PVEAuditor
pveum user token add nasdash@pve monitoring --privsep 1
pveum acl modify / --tokens 'nasdash@pve!monitoring' --roles PVEAuditor

Step 4 is the one most often missed: a token with Privilege Separation and no permission of its own reads nothing, even when its user may (the API then answers 403). Unticking Privilege Separation also works, but the token then carries every right of its user; prefer the token permission. Tokens are described in the Proxmox VE administration guide.

NasDash fields

In Settings → Integrations, enter the host, for example pve.example.com, without scheme, port or API path. Enter 8006 separately in Port, the full Token ID such as nasdash@pve!monitoring in Token ID, and its secret in Token Secret. NasDash adds HTTPS and the API path. Then add a machine and select the connection, exact node name and, if needed, VM/LXC ID and qemu or lxc type.

Node storage is read when the token permits it. 401/403 indicate credentials or permissions; timeouts/refusals are retryable. For a local self-signed certificate, explicitly enable Accept self-signed certificates on this connection. Otherwise normal TLS validation applies. Do not disable verification globally.

When discovery works but the device is offline

Test and list machines checks discovery, not access to each machine's metrics. Proxmox can return a node as online from /nodes while refusing /nodes/<node>/status with 403. In the device editor, choose the machine and use Test metrics to read the same metrics as the widget.

Check the effective permissions in Proxmox under Datacenter → Permissions:

  • Node CPU, RAM and uptime: Sys.Audit on /nodes/<node>.
  • VM or LXC metrics: VM.Audit on /vms/<vmid>.
  • Optional storage statistics: Datastore.Audit on each required /storage/<storage-id>. A storage refusal does not prevent CPU, RAM or uptime from loading.

With Privilege Separation enabled, grant the required read-only permissions to both the user (or their group) and the API token on the appropriate paths. The token's effective access is the intersection of these permissions. Keep Privilege Separation enabled; administrator or write permissions are unnecessary. Creating another token does not grant missing permissions.

401 indicates an authentication problem; 403 means the API denied access to the requested resource and may occur without a failed-login entry. Recheck the selected node or guest after changing its permissions. NasDash retries monitoring automatically. A 403 alone does not establish a Proxmox version incompatibility.

The standard Proxmox status API supplies CPU, RAM, storage and uptime where available. It does not supply CPU/GPU temperatures or GPU usage to this integration; use a supported sensor source such as Glances for those readings.