postgres — In-cluster SQL store

The postgres service is a very small PostgreSQL installation to provide relational storage for applications and environments where data loss is acceptable. Two intended purposes for this service are:

• The JupyterHub user session database

• Backing store for Gafaelfawr’s authentication tokens

If either of those is destroyed, then all current user sessions and authentication tokens are invalidated, work up to the last checkpoint (five minutes in JupyterLab) may be lost. Users will have to log in, restart sessions, and recreate authentication tokens.

Important

Do not use this service for important data. Use a managed relational database, such as Google CloudSQL, instead.

Production instances of the Science Platform use CloudSQL for the Gafaelfawr token database instead of this service.

View on GitHub	/services/postgres Application template
Source	lsst-sqre/rsp-postgres
Type	Helm
Namespace	postgres
Environments	base values Argo CD ccin2p3 values Argo CD idfdev values Argo CD idfint values Argo CD idfprod values Argo CD minikube values roe values Argo CD summit values Argo CD

Guides

• Upgrading postgres
• Adding a new database
  • Decide on a database name
  • Add the database to the deployment
  • Add the database to Phalanx installer
  • Manually add the secret to Vault
  • Restart with new values
• Troubleshooting postgres
  • Recreating postgres PV/PVC
• postgres Helm values reference