mjmcconnell / go_gke_pipeline

Go K8s Tracking Pipeline

This repo is setup to play about with running Golang base services within a kubernetes environment.

One of the main objects, is to be able to effectively track a request as it goes through different services within the k8s cluster.

Tracking is useful for keeping processing requests in a non blocking format, and allow to see the latency of the request at each step of its journey, and the overall request time.

Infrastructure

• Terraform -> Kubernetes
• Helm -> Kubernets resources
• Vault -> Secrets

CI/CD

• Github CI

Pipeline:

• Lambda style applications that enrich data as it flows through the system
• Each app records heart beat of request as it passes through

Monitoring:

• Prometheus
• Grafana
• Jaeger
  • https://github.com/jaegertracing/jaeger-operator

Request flow:

• Client request
• API gateway
  • Validates request
    • Failing requests return an error to the client
  • Determine which service should handle query
  • Message is placed sent to a given PubSub topic
• Secondary service
  • Picks up message from PubSub subscription
  • Processes request
  • Publishes response back onto the PubSub, for the API gateway to consume
• API gateway
  • Watches for responses
  • Gets response
    • Either directly from the PubSub message or Redis
  • Response is sent back to the client

Services

API Gateway

• Create tracking span, and add it to the client request
• Create datastore record
• Pass client request to the enrichment API
• Wait for record to be finalised
  • Redirect loopg
• Return response to client

Setup

This project is setup with Docker (docker-compose), and Makefiles, for ease of use.

Simply make sure you have docker-compose installed locally.

Minikube (Optional)

This repo not only contains the Golang source, but also comes with helm chart for installing the application on kubernetes.

To run application locally within a kubernetes environment, you can install minikube.

Local development

To get the application running locally, simply run:

make docker.run

Packages

[Logrus](golang logrus vs zerolog) - Logging Open Telemetry - Metrics and tracing