The Graphite time-series database software is broken.
The maintainers no longer have the will or the time to do what it takes to go forward. The project is stuck with feature creep and is being held back due to cruft.
With a few swipes, it could become nimble again. Not to say that it could be fully metrics 2.0 aware, but at least that it could continue to evolve.
- Fork Graphite-Web
- Strip out the web interface and only keep the API [Was done by brutasse, search for graphite-api]
- Strip out the event store as this is a client side problem [Was done by brutasse, search for graphite-api]
- Submit to the Graphite-Project as Graphite-API
- Open up Graphite development to more people [There is a last push underway to get a new version from master out the door, yea, kudos to the devs]
Does this sound crazy?
Not as crazy as the current backlog of fixes, pull requests and issues.
Graphite-web has 239 issues and 81 pull requests. The last 60 days has seen NO real commits other than ultra minor fixes (no bug fixes). That is not a healthy project. Especially for one with 400+ forks and 1000+ stars.
A single motivated dev could do 1 and 2, and it could even be created as graphite-api in the graphite project. Of course, the Cabal should open up development access to more people.
A concerned Graphite user.