Built-in utility services - Brightnode Documentation

Every Brightnode Bnode comes with pre-configured utility services that run automatically in your container. These services are accessible via HTTPS URLs that are created when your Bnode deploys.

How to access services

All services are accessed through unique HTTPS URLs in this format:

https://{service}-{bnode-name}.{region}.brightnode.cloud

For example, if your Bnode is named workload-abc123 in the singapore region:

JupyterLab: https://jupyter-workload-abc123.singapore.brightnode.cloud
VS Code: https://vscode-workload-abc123.singapore.brightnode.cloud
System Monitor: https://monitor-workload-abc123.singapore.brightnode.cloud

You can find your Bnode’s service URLs in the Bnode management interface after your Bnode is running.

Available services

The services available depend on which template you’re using.

Base services (all templates)

These services are included in every Bnode regardless of template.

JupyterLab

Web-based interactive development environment with Python notebooks, web terminal, and file management. URL format: https://jupyter-{bnode-name}.{region}.brightnode.cloud Authentication: Password-protected (password is your user ID) Access from UI: Click the Jupyter button (BookOpen icon) on your Bnode card Features:

Python notebooks with GPU support
Integrated web terminal
File browser and editor
Extensions for widgets and archiving

Example usage:

# In a Jupyter notebook
import torch
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"GPUs: {torch.cuda.device_count()}")

VS Code Server

Full-featured Visual Studio Code IDE running in your browser. URL format: https://vscode-{bnode-name}.{region}.brightnode.cloud Authentication: Password-protected (set via environment variable or auto-generated) Features:

Complete VS Code experience
Extension marketplace
Integrated terminal
Git integration
IntelliSense and debugging

Set custom password: When deploying your Bnode, add this environment variable:

CODE_SERVER_PASSWORD=your-secure-password

System Monitor (Glances)

Real-time system monitoring dashboard showing CPU, GPU, memory, and network stats. URL format: https://monitor-{bnode-name}.{region}.brightnode.cloud Authentication: None (read-only) Features:

GPU utilization and temperature
Per-core CPU usage
Memory and swap monitoring
Disk I/O statistics
Network traffic
Process list with resource usage

File Browser

Web-based file manager for uploading and downloading files. URL format: https://files-{bnode-name}.{region}.brightnode.cloud Authentication: None (isolated per user) Features:

Drag-and-drop file uploads
Download files and folders
Create/delete directories
File preview
Archive support

Your persistent data is in /workspace directory.

SSH Access

Terminal access via SSH gateway for command-line operations. Connection format: ssh {bnode-name}@{bnode-name}.{region}.brightnode.cloud Authentication: SSH key (add your public key when deploying) Access from UI: Click the SSH button on your Bnode card to get connection details Example:

ssh workload-abc123@workload-abc123.singapore.brightnode.cloud -i ~/.ssh/id_ed25519

PyTorch template services

Additional services when using PyTorch templates.

TensorBoard

Visualization toolkit for machine learning experiments. URL format: https://tensorboard-{bnode-name}.{region}.brightnode.cloud Authentication: None Features:

Training metrics visualization
Model graph visualization
Embedding projector
Profiler for performance analysis

Usage example:

from torch.utils.tensorboard import SummaryWriter

writer = SummaryWriter('/workspace/logs')

for epoch in range(num_epochs):
    loss = train_one_epoch()
    writer.add_scalar('Loss/train', loss, epoch)

writer.close()

Then visit the TensorBoard URL to see your metrics in real-time.

Gradio Demo

Pre-configured Gradio app for creating ML demos. URL format: https://gradio-{bnode-name}.{region}.brightnode.cloud Authentication: None Demo location: /workspace/gradio_demo.py Customize:

import gradio as gr
import torch

def my_model(image):
    # Your model code here
    return processed_image

demo = gr.Interface(
    fn=my_model,
    inputs=gr.Image(type="pil"),
    outputs=gr.Image(type="pil")
)

demo.launch(server_name="0.0.0.0", server_port=7860)

Streamlit Demo

Pre-configured Streamlit app for data dashboards. URL format: https://streamlit-{bnode-name}.{region}.brightnode.cloud Authentication: None Demo location: /workspace/streamlit_demo.py Customize:

import streamlit as st
import torch

st.title("My ML Dashboard")

if st.button("Run Inference"):
    result = model.predict(input_data)
    st.write(result)

TensorFlow template services

TensorFlow templates include the same ML tools as PyTorch (TensorBoard, Gradio, Streamlit) but with TensorFlow-specific examples.

vLLM template services

Specialized services for large language model inference.

vLLM API Server

OpenAI-compatible API server for high-throughput LLM inference. URL format: https://api-{bnode-name}.{region}.brightnode.cloud Authentication: API key recommended Features:

OpenAI API compatibility
PagedAttention for efficient memory use
Continuous batching
Quantization support (4-bit, 8-bit, GPTQ, AWQ)

Usage example:

from openai import OpenAI

client = OpenAI(
    base_url="https://api-workload-abc123.singapore.brightnode.cloud/v1",
    api_key="your-api-key"  # Optional
)

response = client.chat.completions.create(
    model="meta-llama/Llama-3.2-8B-Instruct",
    messages=[{"role": "user", "content": "Hello!"}]
)

print(response.choices[0].message.content)

Open WebUI

ChatGPT-style interface for interacting with your LLMs. URL format: https://chat-{bnode-name}.{region}.brightnode.cloud Authentication: Built-in user management Features:

ChatGPT-style chat interface
Conversation history
Markdown and code rendering
File uploads (for vision models)

Setup:

Visit the URL and create an admin account
Go to Settings → Connections → OpenAI API
Set Base URL to http://localhost:8000/v1
Start chatting with your model

ComfyUI template services

Services for Stable Diffusion image generation.

ComfyUI

Node-based interface for Stable Diffusion workflows. URL format: https://comfyui-{bnode-name}.{region}.brightnode.cloud Authentication: None Features:

Visual workflow builder
ComfyUI Manager pre-installed
Custom nodes support
Civitai integration

Automatic1111 Web UI

Classic Stable Diffusion web interface. URL format: https://webui-{bnode-name}.{region}.brightnode.cloud Authentication: None Features:

Simple image generation
LoRA support
ControlNet support
Shared models with ComfyUI

Image Gallery

Web gallery for viewing generated images. URL format: https://gallery-{bnode-name}.{region}.brightnode.cloud Authentication: None Automatically displays images from /workspace/ComfyUI/output/.

Quick reference

Service	Port	URL Prefix	Authentication	Available In
JupyterLab	8888	`jupyter-`	Password	All templates
VS Code	8080	`vscode-`	Password	All templates
System Monitor	61208	`monitor-`	None	All templates
File Browser	4040	`files-`	None	All templates
SSH	22	N/A (special)	SSH key	All templates
TensorBoard	6006	`tensorboard-`	None	PyTorch, TensorFlow
Gradio	7860	`gradio-`	None	PyTorch, TensorFlow
Streamlit	8501	`streamlit-`	None	PyTorch, TensorFlow
vLLM API	8000	`api-`	Optional	vLLM
Open WebUI	3000	`chat-`	Built-in	vLLM
ComfyUI	8188	`comfyui-`	None	ComfyUI
Automatic1111	7860	`webui-`	None	ComfyUI
Image Gallery	8765	`gallery-`	None	ComfyUI

Setting passwords

Some services require authentication. Set passwords using environment variables when deploying your Bnode.

During deployment

In the deployment form, add environment variables:

Variable	Service	Example
`JUPYTER_PASSWORD`	JupyterLab	`my-secure-password`
`CODE_SERVER_PASSWORD`	VS Code	`another-password`
`PUBLIC_KEY`	SSH	`ssh-rsa AAAAB3NzaC1...`

Via API

curl -X POST https://api.brightnode.cloud/v1/instances/provision \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "template_id": "pytorch",
    "gpu_option": "A100-1x",
    "environment": {
      "JUPYTER_PASSWORD": "secure-password",
      "CODE_SERVER_PASSWORD": "another-password",
      "PUBLIC_KEY": "ssh-rsa AAAAB3NzaC1..."
    }
  }'

If you don’t set passwords for JupyterLab or VS Code, they will auto-generate passwords. Check your Bnode logs to find them.

Common workflows

ML Training with TensorBoard

# In JupyterLab notebook
from torch.utils.tensorboard import SummaryWriter
import torch

writer = SummaryWriter('/workspace/logs')

for epoch in range(100):
    loss = train_one_epoch()
    writer.add_scalar('Loss/train', loss, epoch)

writer.close()

Then open your TensorBoard URL (https://tensorboard-{bnode}.{region}.brightnode.cloud) to watch metrics in real-time.

LLM Inference with vLLM

Your vLLM Bnode automatically starts the API server
Access the API at https://api-{bnode}.{region}.brightnode.cloud
Use OpenAI client library to interact
Or use Open WebUI at https://chat-{bnode}.{region}.brightnode.cloud for a chat interface

Image Generation with ComfyUI

Open ComfyUI at https://comfyui-{bnode}.{region}.brightnode.cloud
Load a workflow and generate images
View outputs at https://gallery-{bnode}.{region}.brightnode.cloud
Or use File Browser to download specific images

Security

Encrypted connections

All services use HTTPS with automatic TLS certificates via Let’s Encrypt.

Service isolation

Each Bnode’s services are isolated - only you can access your Bnode using your unique deployment name in the URL.

SSH keys

For SSH access, always use key-based authentication instead of passwords:

# Generate SSH key
ssh-keygen -t ed25519 -f ~/.ssh/id_ed25519_brightnode

# Copy public key
cat ~/.ssh/id_ed25519_brightnode.pub

# Add to PUBLIC_KEY environment variable when deploying

Never share your Bnode service URLs publicly if they contain sensitive data. Each URL includes your unique Bnode name and provides direct access.

Troubleshooting

Service not accessible

If you can’t access a service:

Verify your Bnode is in Running state
Wait 1-2 minutes after deployment for all services to initialize
Check that the URL format is correct: https://{service}-{bnode-name}.{region}.brightnode.cloud
Try accessing from an incognito/private browser window

Finding your service URLs

Your service URLs are automatically generated when your Bnode deploys. The format is always:

https://{service-prefix}-{your-bnode-name}.{your-region}.brightnode.cloud

Where:

{service-prefix} = jupyter, vscode, monitor, files, etc.
{your-bnode-name} = The name shown in your Bnode list
{your-region} = The region where you deployed

Service shows blank page

Wait another minute - services may still be starting
Hard refresh your browser (Ctrl+F5 or Cmd+Shift+R)
Check your browser console for errors
SSH into your Bnode and check if the service is running:

# Check if service is running
ps aux | grep jupyter  # or code-server, glances, etc.

# Check service logs
tail -f /jupyter.log
tail -f /code-server.log

Forgot password

JupyterLab and VS Code passwords are auto-generated if not set. To find them:

SSH into your Bnode
Check the logs:

grep -i password /jupyter.log
grep -i password /code-server.log

Next steps

Learn how to configure your Bnode
Set up persistent storage for your data
Explore templates to find the right one for your workflow
Connect via SSH for command-line access

Get started

Bnodes

Storage

Reference

​How to access services

​Available services

​Base services (all templates)

​JupyterLab

​VS Code Server

​System Monitor (Glances)

​File Browser

​SSH Access

​PyTorch template services

​TensorBoard

​Gradio Demo

​Streamlit Demo

​TensorFlow template services

​vLLM template services

​vLLM API Server

​Open WebUI

​ComfyUI template services

​ComfyUI

​Automatic1111 Web UI

​Image Gallery

​Quick reference

​Setting passwords

​During deployment

​Via API

​Common workflows

​ML Training with TensorBoard

​LLM Inference with vLLM

​Image Generation with ComfyUI

​Security

​Encrypted connections

​Service isolation

​SSH keys

​Troubleshooting

​Service not accessible

​Finding your service URLs

​Service shows blank page

​Forgot password

​Next steps

How to access services

Available services

Base services (all templates)

JupyterLab

VS Code Server

System Monitor (Glances)

File Browser

SSH Access

PyTorch template services

TensorBoard

Gradio Demo

Streamlit Demo

TensorFlow template services

vLLM template services

vLLM API Server

Open WebUI

ComfyUI template services

ComfyUI

Automatic1111 Web UI

Image Gallery

Quick reference

Setting passwords

During deployment

Via API

Common workflows

ML Training with TensorBoard

LLM Inference with vLLM

Image Generation with ComfyUI

Security

Encrypted connections

Service isolation

SSH keys

Troubleshooting

Service not accessible

Finding your service URLs

Service shows blank page

Forgot password

Next steps