Training cluster page¶
Enterprise only
The On-prem training cluster page is available only on Enterprise organizations. See Plan permissions.
Connect your organization’s private training hardware to Arena. Managers on Enterprise orgs open On-prem training cluster from the profile menu to turn on private compute and manage resource classes.
The screen has two parts: the org-wide on-prem resource provider switch, then on-prem resource classes after you enable the provider.
Prerequisites¶
Enterprise plan on the current organization
Manager role
Arena CLI installed (
pip install agilerl) andarena loginif you plan to copy CLI install commands from Config
Steps¶
1. Enable the provider¶
If your org has no provider yet, the page says your organization does not have an on-prem training cluster yet. Click Enable on-prem.
After enable, you see Name, Description, and Status (Enabled or Disabled).
Action |
Effect |
|---|---|
Disable |
On-prem classes stop appearing on the experiment Resources step until you enable again. Classes are not deleted. |
Enable |
Turns the provider back on. |
Edit |
Modal to change Name and Description. |
Toast messages confirm enable, disable, and save.
2. Add resource classes¶
The On-prem resource classes table appears after the first Enable on-prem.
Header actions:
Refresh — reloads the class list
Add resource class — modal with fields described in Resource classes
3. Open deployment config¶
Click Config on a class row to expand the deployment panel inline.
Deployment configuration panel¶
Each cluster uses one install type — Docker Swarm or Helm on Kubernetes, not both. Panel order matches the UI:
Arena CLI (recommended) — copy
arena on-prem installand teardown commands for this class Name. Requires the CLI installed (pip install agilerl),arena login, and SSH to cluster hosts (Swarm only). The class already exists when you open Config from the table; the command installs workers for that class.Docker Swarm / Helm — toggle which template set and install command are shown.
Download setup (.tar) — same bundle as the CLI; extract on a bastion,
cd arena-train, run./setup.shor follow the README tab.File tabs below.
Tab (Swarm) |
Content |
|---|---|
README |
Swarm install instructions (rendered markdown in the UI) |
tun0.conf |
WireGuard peer config |
arena-stack.yaml |
Stack definition |
Tab (Helm) |
Content |
|---|---|
README |
Helm install instructions |
values.yaml |
Chart values |
head deployment |
Head pod template |
worker template |
Worker pod template |
Markdown tabs render in-page; YAML and conf files show as monospace text. Empty content shows as an em dash.
Install on hardware that matches the class CPU/GPU/memory and Number of nodes. Both Helm and Swarm bundles mount /opt/data on the Ray head and workers by default; multi-worker pools need shared NFS configured in values.yaml (Helm) or arena-stack.yaml (Swarm) — see Shared storage. Full install walkthrough: Install a cluster. Command reference: On-prem CLI.
Deleting a class¶
Delete opens a confirmation dialog naming the class. Deletion cannot be undone and removes the class from Arena; clean up the matching stack on your side separately.
Access¶
On-prem training cluster appears in the profile menu only for Managers on Enterprise orgs.
Members on Enterprise orgs do not see the menu item. They can still pick on-prem classes on Resources once a Manager has configured them.
The training cluster page requires a Manager on an Enterprise org.