Log-Driven Development

https://www.infoworld.com/article/3017687/application-development/get-started-with-log-driven-development.html

Before pushing code to production you must have

unit tests
integration tests
descriptive log messages
comprehensive metrics
- numerical data
  - system error count
  - failure count
  - user download count
- log data
  - detailed and descriptive
  - types of potential system errors
  - types of potential exceptions
  - types of potential warnings
  - possible unpredictable user experiences ('never happen')
alerts on critical log messages
monitoring dashboards within the troubleshooting platform

Balance creativity with accountability

responsibility
- you develop and push a feature to production == you are accountable for the outcome of that feature
- whether it is successful or not, responsibility for the performance of the feature rests with the developer
recognition
- you successfully implement a feature == you own and are recognised for that

Writing logs

GDPR: no sensitive, personal identifiable information
Security: no business critical information
Performance: look out for potentially high CPU/memory/disk usage
variables in logs must be human-readable or what is the point

Potential Issues

humans reading logfiles is not efficient
you will run out of space if you log everything
knowing what and how to log is something best discussed with a devops expert and a support expert together