Deploying a Docker Container

Documentation Index

Fetch the complete documentation index at: https://docs.shaktistudio.ai/llms.txt

Use this file to discover all available pages before exploring further.

This guide walks you through the complete process of deploying Docker containers to Shakti Studio platform, from credential setup to deployment.

​

Prerequisites

• Docker images uploaded to your Docker Hub repository.
• Docker Hub account with appropriate access permissions:

Important: Ensure your Docker Hub credentials have read access to your repositories.

​

Part 1: Setting Up Docker Hub Credentials

​

Creating a Secret:

1. Navigate to the Integrations section in your Shakti Studio dashboard
2. Select Secrets from the menu
3. Click Create Secret and configure:
   • Secret Name: Enter a name for your secret.
   • Secret Type: Select Docker Hub from the dropdown.
   • Credentials: Enter your Docker Hub username and access token.

​

Part 2: Adding Your Docker Image

​

Model Configuration

1. Go to the My Models page
2. Click Add Model and provide:
   • Model Name: Descriptive name for your application
   • Source: Select the model source as DockerHub from the dropdown
   • Registry Path: Your Docker Hub registry path
   • Image Tag: Specify the image version/ tag
   • GPU Per Container: Default is set to 1. You can allocate maximum 8 GPUs.
   • Credentials: Select the Docker Hub secret created earlier
3. The system will verify your image path and add it to your models list

​

Part 3: Deployment Configuration

​

Basic Setup

1. From the My Models section, select the previously added model and click on Deploy Model
2. Configure basic details:
   • Deployment Name: Enter a unique name for this deployment.
   • Cluster: Select Shakti Cloud as the cluster.
   • Node Group: Choose the required node group (H100/L40S).

​

Container Configuration

​

Service Configuration

Configure how external traffic reaches your container:

• HTTP Service (Required):
  • For example: 8000 (or your application’s port) (The port on which your server will be running).
  • This exposes your service within the cluster.
• Optional Services:
  • gRPC Service: Enable if your application uses gRPC. (This port can’t be the same as http service port)
  • Monitoring Service: Add for enhanced monitoring capabilities.

​

Health Check Settings

Essential for monitoring container health:

• Health Check Path: /health (endpoint your app responds to)
• Port: 8000 (same as your HTTP service port)
• Initial Delay: 30 seconds (startup time allowance)
• Check Interval: 10 seconds (frequency of health checks)
• Timeout: 5 seconds (maximum response wait time)

​

Environment Variables (Optional)

Set runtime environment variables as needed:

• Add key-value pairs for configuration
• Use the + button to add multiple variables

​

Command Override (Optional)

Override the default container startup command:

• Specify the command to run in the container to override the default command. Add each part of the command as a separate entry and then press enter or click on the add button.

​

Scaling Parameters

• Range: Set minimum and maximum instances (1-8)
• Scaling Metric: Choose CPU usage / memory usage as the scaling trigger
• Threshold: Define the percentage that triggers scaling
  (e.g CPU Utilisation 80%)

​

Node Distribution (Node Anti-Affinity)

Control how your containers are distributed across nodes:

• No Affinity: Containers deployed anywhere resources are available
• Preferred: System attempts to distribute across different nodes when possible
• Required: Strict enforcement of separate node placement

​

Advanced Options

• Rapid Auto-scaling: Enable for faster scaling response when needed

​

Part 4: Deployment

1. Review all configuration settings
2. Click Add Deployment to initiate the deployment process
3. Monitor the deployment status

​

Part 5: Monitoring and Access

​

Health Status

Once deployed successfully, you’ll see:

• Health Status: Healthy indicator on the deployment page
• Deployment URL: Direct access link to your application

​

Troubleshooting

• Verify the health check endpoint is properly implemented in your application
• Ensure Docker image is accessible with the provided credentials
• Confirm port configuration matches your application