Install the Torque Agent
An agent is the component that communicates with your Torque tenant and creates the required environment resources. It executes, manages, and monitors the workloads that Torque deploys on your behalf.
This quick-start guide walks you through the fastest path to get Torque up and running: connecting a single agent. Unlike blueprint-driven onboarding, an agent does not require a space or a repository — you can install it on its own and associate it with spaces and repositories later.
In this guide we will cover:
- Where an agent can be installed (cloud or on-premises)
- Generating the deployment for your agent
- Installing the agent and verifying that it is connected
- (Optional) Associating the agent with a space and launching blueprints from a repository
The agent is a lightweight orchestration component deployed in your own environment. Once connected, it appears in Resources → Agents, where you can monitor its health (Connected / Disconnected), type, version, and IP. Torque supports several agent types, including virtual machine images (OVA), managed Kubernetes (for example EKS), Docker, and self-managed Kubernetes.
Step #1: Open the Agents page
- Navigate to Resources → Agents.
- Review the list of existing agents. The page shows an at-a-glance Health and Type breakdown, plus a table with each agent's Name, Status, Type, Version, and IP.
- Click New Agent in the top-right corner to open the Connect an Agent wizard.
Step #2: Choose where to install (Setup)
In the Setup step, select where you would like to install your agent:
| Option | Description |
|---|---|
| Cloud | Install into a public or GPU cloud. After choosing this option, select Public Cloud or GPU Cloud. |
| On-Premises | Install as a virtual machine (VM) in your own datacenter. After choosing this option, select the VM type and then the hypervisor: VMWare, Redhat KVM, or Nutanix. |
Make your selection and click Next.
The remaining steps depend on your choice. Installing on VMware adds an optional Add Credentials step for your vCenter (covered below). Cloud installations skip that step and go straight to the installation instructions.
This guide follows the On-Premises → VM → VMWare path as an example.
Step #3: Generate the agent (Generate Agent)
In the Generate Agent step, fill in the details Torque needs to build the deployment for your agent.
| Field | Required | Description | Example |
|---|---|---|---|
| Agent Name | Yes | A unique, human-readable name for the agent. | mgmt-vcenter-97 |
| VM Hostname | Yes | The hostname to assign to the deployed virtual machine. | mgmt-vcenter-97 |
| IP Address | Yes | The static IP address the agent will use. | 192.168.1.10 |
| Gateway | Yes | The default gateway IP address for the VM's network. | 192.168.1.1 |
| Subnet Mask Bits (CIDR Notation) | Yes | The subnet mask expressed as CIDR bits. | 24 |
| NTP Servers | Yes | One or more NTP servers (comma-separated) used to keep the VM's clock in sync. | pool.ntp.org,pool2.ntp.org |
| DNS Server | No | The DNS server IP address the VM should use for name resolution. | 192.168.1.53 |
| Location | No | The geographic location of the agent, selected from the city search. | Austin, United States |
| Proxy | No | Toggle on to route the agent's outbound traffic through an HTTP/HTTPS proxy, then provide the proxy details. | — |
When all required fields are complete, click Next.
Hover over the ⓘ icon next to any field for inline guidance. Use a static IP that is reserved for the agent so its address does not change after deployment.
Step #4: Add vCenter credentials (VMware only, optional)
For the VMware path, the wizard adds an Add vCenter Credentials (Optional) step (shown as Add Credentials in the step list). It lets you store the vCenter credentials that Torque will later use when this vCenter is added as a target Provider.
This step is optional: you can skip it now and add the target credentials later from Resources → Inventory. Adding them now can save time if the agent is deployed in a network with access to your internal vCenter, because you complete the setup in one flow instead of returning later to create the same credentials.
Credential connectivity can be tested only after the agent is deployed and connected to Torque. Choose an account scoped to the target datacenter or cluster rather than a full administrator account where possible.
Enter the credentials (or leave the step empty) and click Next.
Step #5: Install the agent (Installation Instructions)
The final Installation Instructions step prepares the agent appliance for your chosen target:
- SSH private key - Torque generates an SSH key for authenticating to the appliance. Copy and save it now; it is not accessible after you close the wizard.
- Download Agent File - Torque builds the agent OVA file (this takes approximately two minutes). Click Download to save it. Finish stays disabled until the OVA file has been downloaded.
- Follow the on-screen Instructions to deploy the appliance:
- Go to your ESX server.
- Click Create or Register a VM and follow the steps.
- Select the downloaded
.ovafile and make sure the network adapter that allows external connectivity is selected before you click Finish.
For cloud targets, this step instead shows a command (for example a kubectl-based command for Kubernetes) that you run against your cluster, followed by a Connections Status panel.
It may take up to a couple of minutes for a freshly deployed agent to start up and register. Wait for the Connected indication before continuing.
Verify the connection
Return to Resources → Agents. Your new server should appear in the list with a green Connected status. If it still shows Disconnected after a few minutes, confirm that the VM is powered on and that it has outbound network access (and correct proxy settings, if configured) to reach your Torque tenant.
🎉 That's it — your agent is connected and ready to deploy blueprints from the Self-Service Catalog.
What's next (optional)
A connected agent is fully functional on its own, but to start deploying from blueprints you'll typically pair it with a space and a repository.
Associate the agent with a space
Spaces are logically separate areas that group agents, repositories, and permitted users. To let a space deploy through your new agent:
- From the agent's actions menu (or the post-connection dialog), choose Associate to space.
- Select the target space.
- Set the default namespace and service account that Torque will use to deploy workloads through this server. These defaults can be changed later in Resources → Agents.
Launch blueprints from a repository
Once an agent is associated with a space, you can connect a repository to that space, discover its automation assets, and generate blueprints from them. Publishing a blueprint to the space catalog lets end-users launch environments — Torque will execute those environments through the associated agent.
For the full workflow, see the guides on creating a space, connecting a repository, and launching an environment.