GoCD Helm chart release

[sheroy]

Issue Type

• Feature proposal

Summary

This issue tracks the work remaining to release the GoCD Helm chart

• [Better UX] Update defaults in the Helm chart
  • Disable static agents
  • Enable ingress
  • Fix API versions
  • Enable persistence: Be default, all GoCD configuration should be stored on PVs, upgrading a helm chart shouldn't wipe out configuration.

---

• Merge Helm chart to stable
• Docker in Docker support:
  • [Better UX] Enable privileged mode for the GoCD agent image - DinD images shouldn't necessitate a pod yaml
  • Publish a DinD GoCD agent image based on docker:dind base image.
  • Accessing artifacts (such as test reports)
  • Test edge cases - SELinux based agents,
  • Documentation for using monster image if DinD doesn't work for you

---

• Elastic Agent - remaining work

  • Update EA plugins to conform to the new API

    • Kubernetes plugin
    • ECS plugin
    • Docker swarm plugin

  • Status report page usability improvements

    • "no assigned agent yet" message
    • Suffers from all the clickety clickety drilling down that anything associated with a job suffers from
    • Loads of information, how do you know if good things are happening?
    • High-level status and the rest is detail? Or maybe put the task status at the top.

  • 24 hour performance test. Discuss the outcomes of the previous performance run and the resolution of issues.

  • [Better UX] Error reporting on the status report page

---

• Documentation:
  • Getting Started guide for using the helm chart and the elastic agent. Should probably also talk about the DinD build workflow.
  • Sample pipeline that deploys a demo app to K8S
  • Recommendations for bundling build time dependencies with custom GoCD agent Docker images if the DinD workflow doesn't work for you.
  • Recommendations for size of K8s cluster to run GoCD.
  • Troubleshooting guide

---

• Marketing:
  • Landing page
  • Blog post

---

• Pre-configure elastic agent (2nd iteration):
  • [Better UX] Includes the GoCD agent Docker image and the K8S cluster url
  • Explore PersistentVolumeClaimResize

---

[bdpiprava]

Improvements for NOTES.txt

• helm install --name gocd-app --namespace gocd incubator/gocd
  1. Get the go-server ip using
     $ ip=$(kubectl get ingress gocd-app-gocd-server -o jsonpath="{.status.loadBalancer.ingress[0].ip}") (put --namespace)

• Update it with --namespace gocd

  2. Get a GoCD internal URL for plugin settings
     $ echo "https://$(kubectl --namespace gocd get service gocd-app-gocd-server -o jsonpath='{.spec.clusterIP}'):8154/go"

• Update a note.txt with above command how to get GoCD SSL url for agents

3. Get Cluster url kubectl cluster-info

• Put on note.txt

4. Get token kubectl --namespace=gocd get secret -o jsonpath='{.items[1].data.token} ' | base64 --decode

• kubectl get some-command always returns base64 encoded values for secretes and certs. Update command with base64 --decode to decode the values

5. Get a certificate kubectl --namespace=gocd get secret -o jsonpath='{.items[1].data.ca\.crt}' | base64 --decode

[varshavaradarajan]

• Make it easy to use GoCD and Postgres on K8s (2nd iteration). For Analysis: Check out whether the pg helm chart can be added as a dependency to bring up a pg instance to connect to, install pg-client and contrib packages in the gocd docker image.

[GaneshSPatil]

@bdpiparva @varshavaradarajan @sheroy -- any updates on this?

[ketan]

Closing most of the bigger points are done. Can open specific issues for additional features or improvements.