Our recipe for Daily Releases
It was upon a time a team that struggled to release…
It was upon a time a team that struggled to release…
Cet article est également disponible en français.
I have written a set of articles telling the story of a team that successfully refined its releasing process and practices, up to the point where releasing several times on the same day is a common, un-shocking view.
Here comes all the articles. The order of writing is intentional: you’ll find in the articles links to the previous ones.
Happy reading!
From 1 release per quarter to 2+ releases per week
How we used visual management and gamification for our releasing processjp-lambert.me
Boosting the release pace
Obviously with even more visual management!jp-lambert.me
Daily Stand-Up: getting in sync about releases
The moment when all this visual management come to lifejp-lambert.me
Continuous Isolation: freely choose the next feature to push to production
Another instrumental element in our adventure to Continuous Deploymentjp-lambert.me
Our release process in detail
Made to mitigate known problemsjp-lambert.me
Risk mitigation: formalization and communication
If you lose information along the way, the whole process will jamjp-lambert.me
Wrap-up — Our recipe for daily releases
If I were to sum up all these articles as some kind of big recipe, I would list the following ingredients. I have linked the articles in case you’d want to dive more specifically into some items:
We used a lot of visual management
… To ensure the process is followed
… To sync several teams together
… To speed-up the rate of release, treating push to production as more important than new developments
The visual management artefacts integrated nicely with the existing meetings: daily stand-up, sprint planning, sprint review
We moved away from GitFlow to another branching model that improved our flexibility
We put a particular emphasis on communication and synchronization with dependencies outside the team
We also made sure to mitigate any known risks in the process as we wouldn’t be able to fix them immediately, like staging environment issues or making sure we can rollback in case anything goes wrong
We stopped making full regression testing to favor targeted, risk-based regression testing
… Keeping all the risk analysis data at hand to help having a proper communication with the rest of the world
That’s it! 💥
What about you?
Have you managed a similar feat? How have you done? Would you mind sharing in the comments? ✍️
Thanks in advance! 👋
Did you like these articles?
Follow me to be notified of my new articles!
And don’t forget to clap 👏 and share the articles! Don’t forget that it is your love 💓 that makes me putting my heart and soul into writing.
Thank you 😄