I have a fork of Grafana where I have made some minor changes to the source. I wish to build & package my forked version of grafana, so that I end up with a .tar.gz artifact like I would find published on the grafana downloads page.
How are these artifacts build & packaged by the grafana team?
What is the correct series of scripts to run?
Some things I’ve tried so far:
- I’ve tried building the Dockerfile in the root of the project. I can get this to succeed with some minor tweaks (increase the memory limit to node). This produces a grafana docker image which is not exactly what I want, but could maybe serve my needs - but there are many issues that suggest this Dockerfile is not the canonical build processes used by the team and may not be reliable (issue #15083, issue #39108, issue #37225).
- I’ve tried to understand the build process by going through source, but there are a lot of moving parts. There seems to be circle ci, github actions, makefiles, docker images, bash scripts, go build scripts, go release_publisher scripts, fpm, and more. It’s hard to tell what is current.
- One example: my best guess was ./scripts/build/build.sh. I tried executing this script from inside the grafana/build-container docker image, but it passes the
-buildId=option to build.go, which seems to no longer be supported and suggests that this script is outdated. - Searching through issues & documentation it appears that the build process has changed over time, and this issue suggests that building & packaging from source is not supported?
Ideally I would leverage the existing build process used by the grafana team to guarantee stability & consistency.
So - how is grafana built & packaged by the grafana team? Can I use those same scripts to build & package my fork? If not, what is the recommendation?