Skip to content

MAINTAINING.md

Latest commit

 

History

History
49 lines (33 loc) · 2.13 KB

MAINTAINING.md

File metadata and controls

49 lines (33 loc) · 2.13 KB

Code maintenance

Recurring actions

  • Update the CI/CD if necessary to adapt to new features
  • Write tests for new/changed components
  • Ensure that the main branch is up-to-date to dev once the dev features are battle-tested.
  • Ensure secrets are secrets and backed up somewhere

Known potential issues

The CI/CD passed, but the application fail or isn't up-to-date.

This can happen when something went wrong in the CD script. There is currently no feedback if it worked or not. Logs are available in /tmp. Usually, running the script by hand (sh scripts/update(.dev).sh) solve/highlight the issue.

Linting pass locally but not in the CI/CD.

The CI/CD always run the latest version of black (python linter). The same version should be used locally to avoid conflicts.

Server maintenance

The server is accessible trough SSH using personal user accounts. To access it, it's necessary to be first connected to the VPN.

Recurring actions

  • Ensure that the container's log do not report suspicious errors (~biweekly; after merge, important changes on DB)
    • docker logs --details <container>
      • containers: docker ps (languagelab-frontend languagelab-backend(-dev))
    • or see docs docker
  • Clean up space on the disk (mainly old docker images) (every month)
    • Server limit = 40 GB
    • Mainly erase previous docker images
      • docker system prune (--all)
    • df -h
      • /!\ if higher than 90%
    • Any way to do this automatically after x days?
  • Ensure that the server is up-to-date (Email the first of each month for that, can be a good remember to do all those preventive tasks) (ideally every month)
    • sudo apt update
    • sudo apt upgrade

Known potential issues

The login page charge correctly, but it's impossible to log in.

This error usually happened when for some reason the backend container crashed. It's usually the symptom of a deeper problem, that usually require checking the container's log directly on the server.

The application return a 404 error on every page.

This usually means that the frontend container has crashed, and it also usually require digging in the container's log.