Skip to content

Latest commit

 

History

History
221 lines (163 loc) · 9.03 KB

README.md

File metadata and controls

221 lines (163 loc) · 9.03 KB
Raw
comments

GitLab Runner release process

To handle the growth of this project, in v1.6 we've introduced a release process correlated with GitLab's CE/EE release process.

Release roadmap

Starting with v1.4, GitLab Runner is released on the 22nd day of each month - together with GitLab CE and GitLab EE projects.

Stable release timeline

  • 12th day of a month:

    • tag first RC version on master branch, e.g., v1.6.0-rc1
    • deploy the RC version to docker-ci-X.gitlap.com (each next RC version until the next date should be deployed to those hosts)

  • 17th day of a month:

    • tag next RC version on master branch, e.g., v1.6.0-rc2
    • deploy the current RC version to shared-runners-manager-X.gitlab.com (each next RC version until the next date should be deployed to those hosts)

  • 20th day of a month:

    • close the new features window.

      From now on we're merging only features that are ready and are eventually waiting for documentation or last fixes. After the new features window is closed, we will not merge any feature that wasn't discussed even if it contains all needed changes (code, new tests, documentation) and for which all tests are passing.

      There is nothing bad in moving a feature to the next release at this stage, if it's still not production ready!

  • 21th day of a month:

    • tag last RC version, e.g., v1.6.0-rc5

  • 22nd day of a month:

    • update the CHANGELOG file with entries for current release
    • tag a stable version on master branch
    • create the X-Y-stable on the current master
    • increase version number in VERSION file
    • deploy stable version to docker-ci-X.gitlap.com and shared-runners-manager-X.gitlab.com
    • announce the new release in GitLab's release blog post
    • open the new features window for the next release
    • all MRs that are meant to go into the upcoming release should have the correct milestone assigned and the Pick into X.Y label where X.Y is equal to the milestone, so that release managers can find and pick them. Merge requests without this label will not be merged into the stable branch.

Supported releases

Due to a fast development and release cycle - we release a new version each 22nd day of a month! - we need to prepare a strict policy of releases' supporting.

With this release process description, we're starting the last three releases support policy. That means when we'll release a v1.6 version we will still support v1.5.x and v1.4.x versions. But only them.

After releasing v1.7 we'll support v1.5.x and v1.6.x and so on.

Each support requests for previous versions will be closed with a ~wontfix label.

What is supported?

By the release support we understand:

  • fixes for security bugs
  • fixes for other bugs
  • requests for documentation
  • questions of type "How can I ...?" related to a supported version

Proposals for new features or improvements are welcome, but will be not prepared for supported releases. Instead - if we decide to implement them - they will be planned for one of the upcoming releases.

Release planning

For release planning we use the milestones feature.

If a merge request is to be picked into more than one release it will need one Pick into X.Y label per release where the merge request should be back-ported to.

After releasing a version, the vX.Y milestone will be still used to assign issues and merge requests related to support process (bugs, security fixes, etc.).

We will plan only one version ahead. However, we still want to have a way to mark issues or merge requests that we decided to include in upcoming releases even if we don't know when we'll have time for it. For this purpose we've prepared the %Backlog milestone.

Workflow, merging and tagging strategy

Stable release

For a particular change:

  • start work from master branch and open an MR into the master branch
  • assign the MR to a milestone related to the currently upcoming release
  • choose a good, descriptive title for the MR since it will be automatically inserted in the CHANGELOG before doing the release
  • if the feature is planned for the current release (confirmed by the assigned milestone), merge the feature branch into master

For a whole release please follow the Stable release timeline.

Patch releases

For a particular change:

  • if bug exists in the currently upcoming version:
    • start working from the master branch and open an MR against master
    • assign the MR to a milestone related to the oldest version in which the bug exists
    • choose a good, descriptive title for the MR since it will be automatically inserted in the CHANGELOG before doing the release
    • if a merge request is to be picked into more than one release it will need one Pick into X.Y label per release where the merge request should be back-ported to.
    • merge the feature branch into master
    • after the branch is merged into master, cherry-pick the merge commit to each X-Y-stable branch starting from the branch related to the assigned milestone up to the latest release
  • if bug doesn't exist in the currently upcoming version:
    • start work from X-Y-stable of the most recent version where the bug exists and open an MR against this branch
    • assign the MR to a milestone related to the oldest version in which the bug exists
    • choose a good, descriptive title for the MR since it will be automatically inserted in the CHANGELOG before doing the release
    • if a merge request is to be picked into more than one release it will need one Pick into X.Y label per release where the merge request should be back-ported to.
    • merge the feature branch into the assigned X-Y-stable branch
    • after the branch is merged into the assigned X-Y-stable branch, cherry-pick the merge and commit to each X-Y-stable branch corresponding to the Pick into X.Y labels assigned. Remove each label once picked into their respective stable branches.

For each X-Y-stable branch - if the release should be published:

  • update the CHANGELOG file with entries for the current release
  • create the vX.Y.patch tag on the X-Y-stable branch
  • add the created CHANGELOG entries into the CHANGELOG file on the master branch

Branch naming

While we don't enforce any strict branch naming strategy, we recommend following these guidelines:

  1. Choose descriptive names for branches.

    For example don't name the branch patch-1 or test1 (which tells nothing about its content nor its purpose) when it could be named remove-unused-method-from-docker-executor.

  2. Use name prefixes:

    • if the branch adds/updates documentation, start its name with docs/,
    • if the branch adds a new feature, start its name with feature/,
    • if the branch adds a new improvement, start its name with improvement/,
    • if the branch fixes a bug, start its name with fix/ or bugfix/.

  3. Including issues number in branch name is neither recommended nor discouraged. However, if you want to link the changes with an issue, it's better to create the MR from this branch as soon as possible (you can use the WIP: prefix in the title to prevent any unexpected merges) and link all relevant issues in its description.

    Use Fixes #123, #456 or Closes #123, #345 or mix of them if it's reasonable. Thanks to this issue closing pattern, the issue will be closed along with merging the MR.

Documentation

Some documentation tips:

  1. Create documentation as early as possible and before the change is added into the master/X-Y-stable branch (before the MR is merged).

  2. Properly mark features that need modifications in both Runner and GitLab CE/EE.

    When we introduce new features we mostly mark them in the documentation with the following:

    > Introduced in GitLab Runner v1.6.0

    Most of the times that's enough, but sometimes we introduce a change that needs to be released in both GitLab Runner and GitLab CE/EE to work, e.g., support for artifacts expiration.

    In that case, it should be properly marked in documentation, so it's clear to all what exactly is required for the feature to work.

    We can mark it like:

    > Introduced in GitLab Runner v1.6.0 but it needs at least GitLab 8.12 to work.

    On GitLab CE/EE side (e.g., in API documentation) we would then mark it like:

    > This endpoint was introduced in GitLab 8.12 but it needs at least GitLab
  Runner v1.6.0 to work.

    If the changes are not released at the same time, it would be good to mark which version is not released yet:

    This endpoint was introduced in GitLab 8.12 but it will need at least GitLab Runner v1.7.0 (not released yet) to work.