CV Demo Setup

Create a kubernetes cluster

• One node is enough, with minimum 2 CPU and 6 GB memory
• You will need more capacity if you plan to put a delegate in the cluster

Add kubernetes cloud provider in Harness (docs)

• In Harness, add kubernetes cloud provider for the cluster, called cv-demo
• This could use service account credentials and master endpoint if you want to use a delegate from outside the cluster, or
• Otherwise, install a delegate into the cluster first, then create the cloud provider

Add Dockerhub artifact repo connector (docs)

• Add Dockerhub artifact repo connector called public-docker
• Use URL https://index.docker.io/v2/
• The images are public but docker credentials are needed for the connector
  • Note: if you don’t name the connector the same the artifact source won’t sync.
    If that happens you can add the artifact source yourself.
    cv-demo service: harness/cv-demo
    cv-demo-ui service: harness/cv-demo-ui

Clone app from git (docs)

• Create a new app called cv-demo
• Setup bi-directional git sync while the app is still empty
  • Add git connector and create webhook for bidirectional
  • In git, add the webhook and change content type to application/json
• Clone our cv-demo repo:
  git clone https://github.com/wings-software/cv-demo.git
• Copy cv-demo configuration into your own sync repo:
  cp -R cv-demo/Setup/Applications/cv-demo/* my-cv-demo/Setup/Applications/cv-demo/
• cd my-cv-demo
• Remove Service Guard config for now (we’ll add it back later):
  rm -rf Setup/Applications/cv-demo/Environments/cv-demo/Service\ Verification/
• git add -A
• git commit -a
• git push
• Check that your app was synced in the Harness UI
  • Check for errors in Configuration as Code

Reserve static IPs

• Reserve static IPs (see instructions for AWS, GCP, or Azure) for:
  • Ingress controller
  • Prometheus
  • Elastic Search
• Optionally reserve DNS name for the ingress controller static IP

Update Harness environment cv-demo

• Edit values YAML override for nginx-ingress-controller service and override:
  loadBalancerIP: <ingress-static-ip>
• Edit values YAML override for cv-demo service and override:
  

  env:
    config:
      ALLOWED_ORIGINS: http://<ingress-static-ip or dns>

• Edit service variable override for cv-demo-ui service and override:
  baseUrl: http://<ingress-static-ip or dns>
• Edit values YAML override for prometheus service and override:
  loadBalancerIP: <prometheus-static-ip>
• Edit values YAML override for elastic-search service and override:
  loadBalancerIP: <elk-static-ip>

Execute Setup pipeline

• Execute the pipeline CV Demo - Cluster Setup
• Select stable artifact for cv-demo service and latest for cv-demo-ui
• This deploys the Nginx ingress controller, prometheus, and elastic search, as well as a baseline cv-demo backend and the cv-demo-ui for controlling it

Create verification connectors (docs)

• Create ELK-cv-demo with URL http://<elastic-search-ip>
• Create Prometheus-cv-demo with URL http://<prometheus-ip>:8080/

Edit cv-demo-canary workflow

• In Prometheus step, select your Prometheus connector
• In ELK step, select your ELK connector

Execute workflow

• Execute workflow cv-demo-canary
• Select verify_canary = yes
• You can either select the unstable image tag, or:
  • In browser, visit http://<ingress-static-ip or dns>
  • Adjust the canary log and metric error rates and values via the UI
• Observe Verification steps detecting the anomalies

Setup 24/7 Service Guard

24/7 docs
prometheus docs
Elasticsearch docs

• Copy Service Guard configuration into your own sync repo:
  cp -R cv-demo/Setup/Applications/cv-demo/Environments/cv-demo/Service\ Verification my-cv-demo/Setup/Applications/cv-demo/Environments/cv-demo
• git add -A
• git commit -a
• git push
• Check that 24/7 Service Guard is now configured in Continuous Verification
  • Check for errors in Configuration as Code
• Open the ELK configuration
  • Select baseline as last 30 minutes
  • Submit
• In browser, visit http://<ingress-static-ip or dns>
• Adjust the primary log and metric error rates and values via the UI
• Wait some time and observe 24/7 Service Guard detecting anomalies