Mount a staging platform

• Docker best practices
  • gitlab fast zips https://www.augmentedmind.de/2022/06/12/gitlab-vs-docker-caching-pipelines/
  • gitlab fast zips https://blog.nimbleways.com/let-s-make-faster-gitlab-ci-cd-pipelines/
  • use --mount=type=cache ?https://medium.com/@albertazzir/blazing-fast-python-docker-builds-with-poetry-a78a66f5aed0
  • https://github.com/containers/buildah/discussions/4536
• Define entrypoints for celery, celerybeat
  • celery -A <app> worker -l INFO (--autoscale 10 ?)
    • https://miguescri.com/post/2021-05-22-celery-hpa/
    • https://learnk8s.io/scaling-celery-rabbitmq-kubernetes
  • celery -A <app> beat -l INFO -s /file-to-store-last-run
  • build an image for waveqc-celerybeat, waveqc-celery, and waveqc-web
• We must make sure that celery and pyramid apps share the same mount point
• Use k8s jobs for alembic upgrade at the end of each deployment
  • Use a shell command as entrypoint ?
• Only monitor a subset of stations for staging
• Adapt celerybeat schedule to the number of celery workers ?
• How do we manage a Redis server in k8s ?
• Dump database on http://resif-waveqc-dev.u-ga.fr/ and restore it on staging?
• Make sure we have enough disk space on staging
• Stop celerybeat.service on http://resif-waveqc-dev.u-ga.fr/
• Destroy http://resif-waveqc-dev.u-ga.fr/