An empty chart at charts/inbuxa (helpers, install notes, a values schema that refuses keys it doesn't know) and tools/check.sh, which is all of CI: helm lint --strict under Helm 4 and Helm 3, a check that the schema refuses an unknown key, kubeconform against Kubernetes 1.35, 1.36 and 1.37, kube-linter with every built-in check, and helm-unittest. tools/install-tools.sh fetches those tools at pinned versions, each checked against a pinned SHA-256. CONTRIBUTING.md (Sending a change, the checks), issue forms for a bug and a release-candidate report, a PR template and CHANGELOG.md. Closes #1
3.3 KiB
Contributing to inbuxa-charts
This repository holds the Helm chart that runs inbuxa on Kubernetes: the mail server, the webmail and the console. Bug reports, release-candidate reports and changes are all welcome.
Before you start
- The chart runs the inbuxa images; it doesn't build them. A bug in the server, the webmail or the console belongs in inbuxa-server, inbuxa-webmail or inbuxa-admin.
- The data layer (PostgreSQL, NATS, S3) is yours to bring. The chart connects to it and doesn't install it.
- No operator and no custom resources: a chart and plain manifests only.
- The chart is licensed under Apache-2.0, and a change you send is distributed under it. The images it runs stay under their own licences.
Reporting a problem
Search the issues first. A new issue offers two forms: Bug for something that's wrong, and Release candidate report for how an RC went on your cluster, good or bad. Both ask for the chart version, your Kubernetes version and distribution, and what you set in your values. Leave out passwords, keys and hostnames you'd rather not publish.
Sending a change
New accounts on git.coffeylabs.org can file issues but can't push or fork. Until you have repository access, open an issue for the change, or say hello on community.coffeylabs.org, and attach the change to the issue as a patch:
git format-patch origin/main --stdout > my-change.patch
Repository access goes to volunteers we have worked with; a few good issues
or patches is the usual way in. After that you push a branch here and open a
pull request against main. The GitHub copy is a read-only mirror, so pull
requests opened there are not merged.
A pull request:
- Does one thing, on a branch named for it (
fix/probe-timeout,feat/webmail-hpa). - Passes
tools/check.shbefore it's pushed (see below). - Adds or updates a test in
charts/inbuxa/tests/for what it changes. - Describes any new value in
values.yamland invalues.schema.json, with a sensible default. The schema refuses keys it doesn't know, so a value missing from it can't be set at all. - Adds a line to
CHANGELOG.mdunder Unreleased.
main is protected: a change reaches it through a pull request whose
chart check has passed.
The checks
tools/check.sh is everything CI runs:
helm lint --strictunder Helm 4 and Helm 3, with each values file incharts/inbuxa/ci/- that the values schema refuses an unknown key
- kubeconform against the three newest Kubernetes minors
- kube-linter, with every built-in check on
- helm-unittest
tools/install-tools.sh downloads the exact versions CI uses, checked
against pinned SHA-256 hashes, into .tools/bin (linux/amd64):
tools/install-tools.sh && tools/check.sh
To check a new set of values, add charts/inbuxa/ci/<name>-values.yaml;
every file there is linted, rendered and checked.
Installing on a real cluster, the mail tests and the upgrade tests live in inbuxa-k8s-lab.