Skip to content

Latest commit

 

History

History
80 lines (54 loc) · 6.85 KB

README.md

File metadata and controls

80 lines (54 loc) · 6.85 KB

Krishna's OpenFAAS function (akin to Lambda functions)

Note

For development and deployment instructions, see the OpenFaaS documentation. This repository is for tracking the functions I create.

Overview

This repository contains OpenFaaS functions. OpenFaaS is a framework for building serverless functions on top of containers that makes it easy to deploy event-driven functions and microservices to Kubernetes without repetitive boilerplate coding.

Why did I choose OpenFaaS?

  • Portable across cloud and on-premises environments
  • Build and deploy using familiar containers
  • Automatic scaling based on demand
  • Metrics and logs for observability
  • Write functions in any programming language and framework using templates

Deployment

The stack.yml file contains the function definitions and deployment configuration. To deploy the functions, run:

# Copy the stack.sample.yml to stack.yml
# Update the stack.yml with your environment variables
cp stack.sample.yml stack.yml

OpenFaaS supports various languages through its own templates concept. We use their python3-http template, which is based on Alpine Linux and has a small footprint.

# This is an official template maintained by OpenFaaS
faas-cli template store pull python3-http
faas-cli update -f stack.yml

Serverless Functions

Within the intranet, the serverless orchestration is done using OpenFaaS. It is available at https://functions.server.local.

Function Catalog

llm-global-spend

  • Endpoint: /function/llm-global-spend
  • Content-Type: application/json

Calculates spend from Lite-LLM API logs instead of using the periodic spend API endpoint, which is now paywalled. Tested to work with 1.53.7.

Sample Request

Warning

Requires a hosted Lite-LLM instance. You need to add LITELLM_API_KEY and LITELLM_API_BASE_URL environment variables in stack.yaml for the function to work.

Description Features Method Request Body Sample Responses
Retrieves spending logs for a specified user from an API endpoint and calculates the total spend within a given date range. Queries spend logs API with user ID and date range parameters GET None { "global_spend": 622.5803327750004, "current_month_spend": 1.10207825, "today_spend": 0.0 }

spot-start-service

  • Endpoint: /function/spot-start-service
  • Content-Type: application/json

Warning

Requires a hosted Portainer instance. You need to add PORTAINER_API_URL, PORTAINER_USERNAME and PORTAINER_PASSWORD environment variables in stack.yaml for the function to work.

Description Features Method Request Body Sample Responses
Starts or restarts Docker service stacks managed by Portainer. Accepts service name as input parameter POST {"service": "stopped_service"} { "success": true, "message": "Service nextcloud successfully started", "details": { "service": "nextcloud", "endpoint": "prod-vm-1", "stack_id": "80" } }
It verifies if the specified service is running and starts it if it's stopped or in an unhealthy state. Authenticates with Portainer API POST {"service": "unhealthy/partial_service"} { "success": true, "message": "Service mediaservice successfully restarted", "details": { "service": "mediaservice", "endpoint": "prod-vm-1", "stack_id": "119" } }
Locates the correct endpoint and stack ID POST {"service": "healthy_service"} { "success": true, "message": "Service n8n is already running and healthy", "details": { "service": "n8n", "endpoint": "prod-vm-1", "stack_id": "181", "status": "healthy" } }
Checks container status POST {"referral_url": "https://example.server.local"} {'statusCode': 200, 'body': '{"success": true, "message": "Service 'example-stack' is already running", "action": "none", "endpoint_id": "10, "stack_id": "200"}', 'headers': {'Content-Type': 'application/json'}}
Starts or restarts service stack as needed
Returns JSON response with operation status