Skip to main content
The Standalone Agent is in beta. Features and configuration may change.
This guide covers installing the Chamber Standalone Agent on a Linux GPU host (bare metal, cloud VM, or on-prem server).

​
Prerequisites

RequirementDetails
Chamber accountSign up at app.usechamber.io/signup and contact the Chamber team to enable standalone agent access. Then create an API token from Settings > API Tokens.
GPU hostLinux host with one or more NVIDIA GPUs (GCP, AWS, Azure, or bare metal)
NVIDIA driverPre-installed on most cloud GPU images. Verify with nvidia-smi.
Root accessThe installer requires sudo
Python 3.11+, Docker, and DCGM Exporter are installed automatically by the installer if not already present.

​
Step 1: Get Your Cluster Token

1

Log in to Chamber

Go to app.usechamber.io and navigate to Settings.
2

Create a token

Click New Token, give it a name, and copy the generated token immediately.
Store your token securely. It is only shown once and cannot be retrieved later.

​
Step 2: Install the Agent

SSH into your GPU host and run: 
curl -fsSL https://chamber-agent-releases.s3.amazonaws.com/install.sh \
  | sudo CHAMBER_CLUSTER_TOKEN=<your-token> bash
Replace <your-token> with the token from Step 1. The installer will:
  • Install Python 3.11+, Docker, and NVIDIA Container Toolkit (if not already present)
  • Set up DCGM Exporter for GPU profiling metrics
  • Install and configure the Chamber Standalone Agent
  • Start the agent as a systemd service
The full process takes 2–5 minutes depending on what is already installed.

​
Step 3: Verify

Your host should appear in the Chamber dashboard within 60 seconds. You can also verify locally: 
# Check service status
sudo systemctl status chamber-agent-standalone

# View recent logs
sudo journalctl -u chamber-agent-standalone --no-pager -n 20
You should see log lines indicating a successful WebSocket connection to the control plane.

​
Verify in Dashboard

  1. Go to app.usechamber.io/dashboard?tab=services
  2. Your host should appear with GPU count and metrics flowing
If the host doesn’t appear within 60 seconds, see Troubleshooting.

​
Common Operations

# View live logs
sudo journalctl -u chamber-agent-standalone -f

# Restart the agent
sudo systemctl restart chamber-agent-standalone

# Stop the agent
sudo systemctl stop chamber-agent-standalone

# Check agent version
/opt/chamber-standalone/venv/bin/chamber-standalone version

# View detected GPU and host info
/opt/chamber-standalone/venv/bin/chamber-standalone host-info

​
Network Requirements

The agent requires outbound HTTPS/WSS access only. No inbound ports need to be opened.
HostPortPurpose
controlplane-api.usechamber.io443Control plane communication (WebSocket)
If you are running behind a corporate firewall, ensure outbound WebSocket connections to port 443 are allowed.

​
Next Steps

Metrics Reference

Understand the metrics collected and what they mean

Configuration

Customize agent behavior with environment variables

OTLP Integration

Send your application metrics to Chamber

Troubleshooting

Common issues and solutions