Clusters
Connect Proxmox clusters, watch hosts and storage, fail over between API endpoints, and onboard existing guests.
Connect a cluster
- 1On Clusters, click Connect cluster (cluster.connect).
- 2Enter a name and the endpoint URL, for example https://pve01.example.com:8006. VMware vSphere is shown but not available yet.
- 3Authenticate with an API token or with an admin username and password. Credentials are stored encrypted.
- 4Keep Skip TLS verification on for internal clusters with self-signed certificates.
- 5Click Test & Connect. The connection is verified first and the number of discovered hosts is reported.
Cluster list and status
- Each row shows status, type, endpoint, auth method, hosts, VMs, version and CPU, memory and storage usage.
- Connected means the API is reachable. Degraded means partly reachable, for example while a fallback endpoint is in use. Disconnected means unreachable.
Inside a cluster
| Tab | Contents |
|---|---|
| Overview | Hosts online, VMs, storage volumes, utilization, details and endpoints |
| Hosts | Per node: status, capacity, live CPU and memory, version, and whether it is enabled for this project |
| Storage | Live volumes with type, shared or local, node, content and usage |
| Virtual Machines | Every guest on the cluster, with a Portal column (vm.view) |
| Activity | Recent audit events for this cluster (audit.view) |
Rescan and onboard
- Rescan (cluster.edit) re-reads hosts, version and usage figures. VM and storage counts are always live.
- The Portal column on the Virtual Machines tab shows whether a guest is already in a project. Guests that aren't get an Onboard button: pick an owner and the guest is adopted into the current project with its details read from Proxmox.
Endpoints, disconnect and delete
- Manage endpoints (cluster.edit) keeps an ordered list of API endpoints. The first is primary; the others are tried in order when it stops answering. Each can be tested before saving.
- Disconnect pauses the connection without deleting anything.
- Delete (cluster.delete, press-and-hold) removes the cluster and its tracked data from the portal only. Nothing changes on the hypervisor.
Features on this page
- Cluster connections
- Endpoint failover
- Cluster rescan and onboarding
More in PlatformAll topics
TemplatesAllow golden images per project, cap their resources and set per-template provisioning rules.High availabilityRead the cluster's HA state live, manage node-affinity rules, and place new VMs into HA automatically.Fleet and agentsThe Verlocx agents inside your guests: enrollment, health, self-update and offboarding.