Page MenuHomePhabricator

Create production code deployment management process
Open, MediumPublic


As the Wikimedia Foundation, we have a responsibility to ensure reliability and stability of the production environment. To that end, we need to look for ways to continuously improve how we manage what code is deployed to production.

Although there are existing process is place that we already follow, there are likely improvements that could be made.

This is a parent task to track activities that are related to managing how and what things are deployed into production.

Event Timeline

Quiddity mentioned this in Unknown Object (Task).Oct 25 2018, 7:04 AM

Although the goal is to have all code that deploys to production go through this process, it may be necessary to have two branches in this process. One that is for "permanent" deployment to production, and one that is "conditional". For example, team needs to get new capabilities into the production environment in order to obtain timely user feedback, but the code is not fully compliant to Code Health, security, performance, etc... criteria. They may have a time-based condition to address the gaps or be removed from production.

Not sure that's the only/best approach, but I think an exception process needs to be addressed.

Another thought to capture... This process should probably result in a workflow that is as automated as possible. For example - initiated by the creation of a phab task that then auto creates other tasks from template with appropriate folks subscribed/assigned. In the end, minimizing the room for error in a manual set of steps on a wiki.

greg triaged this task as Medium priority.Feb 27 2019, 6:36 PM