Skip to content

DEPLOYMENT_PROCEDURE.md

Latest commit

 

History

History
48 lines (28 loc) · 2.98 KB

DEPLOYMENT_PROCEDURE.md

File metadata and controls

48 lines (28 loc) · 2.98 KB
Raw

Deployment procedure

This document is only intended for Pathoplexus sysadmins.

To develop new versions of Pathoplexus

Work in a development branch. You can choose what version of Loculus to target by adjusting the loculusVersion in pathoplexus_app/values.yaml (use a Loculus commit SHA). Adding the preview label will create a preview.

To simply bump the version of Loculus used in Pathoplexus main

We have a GitHub workflow for this. Go to the "Actions" tab -> Open the "Create PR to bump main pathoplexus.." action -> Open the "Run workflow" dropdown -> Click on the "Run workflow" button. This will create a PR, which someone else will need to approve. The PR will contain a changelist for Loculus, which you should carefully review for breaking changes.

gh workflow run bump-main-loculus-version-to-latest.yaml -R pathoplexus/pathoplexus

Rolling out a new release to production (via staging)

Full docs are in the loculus_deployments README.

Staging the latest main

Once you have a new release candidate on main, you can point the staging server at it by changing the head_sha value at https://github.com/pathoplexus/loculus_deployments/blob/main/deploy/staging/config.json.

The easiest way to do this is with the GitHub action "Create PR to bump staging to latest" in the loculus_deployments repository: https://github.com/pathoplexus/loculus_deployments/actions

A little after the PR is merged, changes will be reflected at https://staging.pathoplexus.org. You should also check the health of the application in ArgoCD.

Promoting a new version from staging to production

Once you have a new release candidate you can point the production server at it by changing the head_sha value in the production folder at https://github.com/pathoplexus/loculus_deployments/blob/main/deploy/production/config.json.

The appropriate way to do this is with the GitHub action "Create PR to Promote Staging to Production" in the loculus_deployments repository which will copy the SHA from staging to production, creating a PR: https://github.com/pathoplexus/loculus_deployments/actions

gh workflow run promote-staging-to-production.yml -R pathoplexus/loculus_deployments

After merging, the changes will be reflected at https://pathoplexus.org. You should check ArgoCD to monitor the health of the application.

What determines the configuration of an instance?

All configuration is in this repository, and all is versioned such that only the version pointed to in the loculus_deployments repo has an effect.

  • Customisations to the website image, made in the monorepo folder, will be automatically used in deployments.
  • The Loculus version to target is set in pathoplexus_app/values.yaml
  • The core configuration of Pathoplexus is set in loculus_values/values.yaml.
  • Additional configuration that will only be used on special deployments is set in loculus_values/environment_specific_values/.