Patroni Contributors Chart the Road Ahead in Munich
The day before pgconf.de 2024 in Munich, members of the Patroni project gathered to align on the framework's direction. Patroni, originally built at Zalando, is now steered by engineers from several companies. Ants Aasma and I joined the discussion on behalf of CYBERTEC. Because Patroni underpins many critical PostgreSQL deployments, coordinated development and clear agreements among contributors are essential.
Stability Through Versioning
A major shift over the past year was the adoption of semantic versioning. Starting with version 3.1.0, the versioning scheme is now strict: the last digit is for bugfix releases, the middle digit for minor updates, and the leading digit for major upgrades that may introduce breaking changes and require careful migration planning.
Development will continue on the master branch. Bugfixes get cherry-picked onto minor-release branches such as REL_3_1 and REL_3_2. There is no fixed release cadence for any of the three tracks; updates ship as needed. No LTS releases are planned at this time, though interested parties are welcome to take on that maintenance. In practice, existing clusters rarely need to pin a specific minor version—upgrades are usually straightforward. New deployments can typically adopt the latest release, which may be necessary for support of the newest PostgreSQL versions.
Raising the Bar on Code Quality
The project has also introduced type annotations across the entire codebase, alongside near-complete test coverage. These measures underscore the quality bar the maintainers set for Patroni's continued evolution.
Aligning Terminology with PostgreSQL
For the next major version, the plan is to replace every occurrence of the word "master" with "primary." This mirrors the terminology used by the PostgreSQL project to describe instance roles in binary streaming replication clusters.
Feature Development and Multi-Data-Center Work
Discussions also covered proposed feature additions, development goals, and the project's long-term purpose. CYBERTEC is actively researching and building approaches that enable replication and automatic failover with Patroni across multiple data centers. The aim is to let users remain in a single data center as long as possible, without forcing an availability trade-off by making switchover a manual process. Representatives from CYBERTEC are pleased to have contributed to the meeting and to the project's ongoing journey.
