dev_docs/contributing/how_we_use_github.mdx
We follow the GitHub forking model for collaborating on Kibana code. This model assumes that you have a remote called upstream which points to the official Kibana repo, which we’ll refer to in later code snippets.
At Elastic, all products in the stack, including Kibana, are released at the same time with the same version number. Most of these projects have the following branching strategy:
main points to the next minor version.<major>.<minor> is the previously released minor version, including patch releases.As an example, let’s assume that the main branch is currently a not-yet-released 8.1.0. Once 8.1.0 has reached feature freeze, it will be branched to 8.1 and main will be updated to reflect 8.2.0. The release of 8.1.0 and subsequent patch releases will be cut from the 8.1 branch. At any time, you can verify the current version of a branch by inspecting the version attribute in the package.json file within the Kibana source.
The active branches and their roles are defined in versions.json at the repository root, categorized by branchType:
"development" -- Branches like main for new features and breaking changes"release" -- Maintained release branches for backporting fixes, docs, tests, security, and dependency updatesPull requests are made into the main branch and only backported when it is safe and appropriate.
main if there has been at least an 18 month deprecation period and the breaking change has been approved. Telemetry showing current usage is crucial for gaining approval.<major>.<minor> branch.<major>.<minor> branch if the changes are safe and appropriate. Safety is a judgment call you make based on factors like the bug’s severity, test coverage, confidence in the changes, etc. Your reasoning should be included in the pull request description.yarn.lock across branchesWe want to keep updates to dependencies (both transitive dependencies and direct dependencies) in sync across both main and <previous major>.<last minor> (ex: 7.17) as much as possible.
A good rule of thumb is that most package upgrades should be backported to the <previous major>.<last minor> branch, though as always, exceptions may apply – if an upgrade requires significant code changes, then it might make sense to skip a backport for it.
@elastic.co email addressIn order to assist with developer tooling we ask that all Elastic engineers use their @elastic.co email address when committing to the Kibana repo. We have implemented a CI check that validates any PR opened by a member of the @elastic organization has at least one commit that is attributed to an @elastic.co email address. If you have a PR that is failing because of this check you can fix your PR by following these steps:
Ensure that you don't have any staged changes
Checkout the branch for your PR
Update the git config for your current repository to commit with your @elastic.co email:
git config user.email [email protected]
Create a commit using the new email address
git commit -m 'commit using @elastic.co' --allow-empty
Push the new commit to your PR and the status should now be green
Note: If doing this prevents your commits from being attributed to your Github account then make sure to add your @elastic.co address at https://github.com/settings/emails.
Rebasing can be tricky, and fixing merge conflicts can be even trickier because it involves force pushing. This is all compounded by the fact that attempting to push a rebased branch remotely will be rejected by git, and you’ll be prompted to do a pull, which is not at all what you should do (this will really mess up your branch’s history).
Here’s how you should rebase main onto your branch, and how to fix merge conflicts when they arise.
First, make sure main is up-to-date.
git checkout main
git fetch upstream
git rebase upstream/main
Then, check out your branch and rebase main on top of it, which will apply all of the new commits on main to your branch, and then apply all of your branch’s new commits after that.
git checkout name-of-your-branch
git rebase main
You want to make sure there are no merge conflicts. If there are merge conflicts, git will pause the rebase and allow you to fix the conflicts before continuing.
You can use git status to see which files contain conflicts. They’ll be the ones that aren’t staged for commit. Open those files, and look for where git has marked the conflicts. Resolve the conflicts so that the changes you want to make to the code have been incorporated in a way that doesn’t destroy work that’s been done in main. Refer to main commit history on GitHub if you need to gain a better understanding of how code is conflicting and how best to resolve it.
Once you’ve resolved all of the merge conflicts, use git add -A to stage them to be committed, and then use git rebase --continue to tell git to continue the rebase.
When the rebase has completed, you will need to force push your branch because the history is now completely different than what’s on the remote. This is potentially dangerous because it will completely overwrite what you have on the remote, so you need to be sure that you haven’t lost any work when resolving merge conflicts. (If there weren’t any merge conflicts, then you can force push without having to worry about this.)
git push origin name-of-your-branch --force
This will overwrite the remote branch with what you have locally. You’re done!
Note that you should not run git pull, for example in response to a push rejection like this:
! [rejected] name-of-your-branch -> name-of-your-branch (non-fast-forward)
error: failed to push some refs to 'https://github.com/YourGitHubHandle/kibana.git'
hint: Updates were rejected because the tip of your current branch is behind
hint: its remote counterpart. Integrate the remote changes (e.g.
hint: 'git pull ...') before pushing again.
hint: See the 'Note about fast-forwards' in 'git push --help' for details.
Assuming you’ve successfully rebased and you’re happy with the code, you should force push instead.
See Submitting a pull request for the next steps on getting your code changes merged into Kibana.
The following information notes how we use GitHub labels in Kibana. Note that only internal Elasticians are able to create and assign labels to issues, but for searching purposes, the information is likely useful for external developers as well.
Many of our labels follow the pattern of {key}:{value}.
Teams can create labels at their own discretion, but we have over 600 labels at the time of this writing. Consider using an existing convention before creating a new one. If you think a new label or convention would be useful to all teams, talk to your team or tech lead about getting it added here.
</DocCallOut>Examples: Team:Security, Team:Operations, Team:Docs.
These labels map the issue to the team that owns the particular area. Part of the responsibilities of (todo) is to ensure every issue has at least a team or a project label.
<DocCallOut title="Internal only"> View our org chart [here](https://wiki.elastic.co/display/DEV/Kibana+Team+Organization) to view all our teams and appropriate contacts. </DocCallOut>Examples: Feature:Lens, Feature:TSVB, Feature:Vega.
Feature labels break down architectural domains that are owned by a given team.
Examples: Project:RuntimeFields, Project:MakeItSlow.
Sometimes issues span multiple teams, that is often when Project labels are more appropriate. To avoid too much noise, these should be used for high visibility projects. Try not to use project labels for small, single team projects, where a team and feature label would be applicable. Use your best judgement when determining whether to add a new project label.
NeededFor:{Team})Examples: NeededFor:APM, NeededFor:AppServices.
We use these labels to help us organize internal dependencies. An issue with the labels
NeededFor:APM and Team:AppServices means APM has a dependency on the App services team. The owning team
can filter on these labels during roadmap prioritization, and the dependent team can use these labels to
search and view the status of its dependencies. To avoid noise, use these labels for high priority requests that
need to be taken into account in roadmap planning. A low priority item that can be prioritized along with
other community requests does not need this label, as part of its usefulness is helping teams wade through the noise
of external feature requests.
Examples: v7.9.2, v8.0
We use version labels on PRs to indicate which versions a PR will be merged into. For issues, teams use these labels inconsistently. On a bug, it might mean the version the bug was found in, or it might mean the version the team is tentatively planning to merge a fix.
Consult the owning team if you have a question about how a version label is meant to be used on an issue.
These labels categorize the type of work. For example:
blocked: Indicates the issue is currently blockedblocker: Indicates that we should not release the product at the next
proposed version without the issue being resolvedbug: Indicates an unexpected problem or unintended behaviordiscuss: Indicates that an issue is a discussion topicdocs: Indicates improvements or additions to documentationenhancement: Indicates new feature or enhancement requestsmeta: Indicates that the issue tracks tasks related to a projecttriage-needed: Indicates that someone from the area team needs to investigate.needs_team: Indicates that issue is missing area team label. All issues should be assigned to one or more area teams for follow up. This label is assigned automatically and removed automatically once a team label is added.These labels affect whether your PR appears in the release notes (that is to say, it's notable and affects our users) and which section it appears in. For example:
release_note:breaking: Specifies a breaking change and adds the PR to the Breaking changes section in the release notesrelease_note:deprecation: Specifies a deprecated feature and adds the PR to the Deprecations section in the release notesrelease_note:enhancement: Specifies a feature enhancement and adds the PR to the Enhancements section in the release notesrelease_note:feature: Specifies a new feature and adds the PR to the Features section in the release notesrelease_note:fix: Specifies a bug fix and adds the PR to the Bug fixes section in the release notesrelease_node:plugin_api_changes: Specifies a changes to the plugin API and adds the PR to the Plugin API changes page in the Developer Guiderelease_note:skip: Omits the PR from release notesThe following labels are related to backporting PRs:
backport:version: Automatically backport this PR (to the branches related to
version labels) after it's merged. Requires adding desired target versions labels.backport:all-open: Automatically backport to all generally available versions. This functionally is equivalent to backport:prev-major at the time of writing.backport:skip: This PR does not require backporting.backport: This PR was backported (added by CI).If automated backporting fails (e.g. due to merge conflicts), you can backport manually:
# Interactive backport (select commits and branches)
node scripts/backport
# Backport a specific commit to specific branches
node scripts/backport --sha <commit-sha> --branch 9.2 --branch 9.1