Skip to content

Latest commit

 

History

History
140 lines (104 loc) · 5.63 KB

RELEASE.md

File metadata and controls

140 lines (104 loc) · 5.63 KB
Raw

AppScope Release Process

Our process is pretty straightforward. It focuses primarily on developing the next release while keeping the effort devoted to maintaining older releases to a minimum.

Branching & Tagging

Branching

  • The default branch is master.

  • We create branches from the default branch for defects bug/123-name and feature requests feature/123-name. The number prefix is an issue number and the -name usually aligns with the issue title. Occasionally, we skip creating an issue and omit the number.

  • We create PRs when work on issue branches is complete and ready to be reviewed and merged back into the default branch.

  • We name releases using the typical major.minor.maintenance pattern for Semantic Versioning. We start with 0.0.0 and increment...

    • ... major when breaking changes (BCs) are made.
    • ... minor when we add features without BCs.
    • ... maintenance for bug fixes without BCs.

    We append -rc# suffixes for release candidates.

  • We create release candidate tags v*.*.0-rc* on the default branch as we approach the next major or minor release. When it's ready to go, we create the release tag v*.*.0, also on the default branch.

  • We create a release branch release/0.x, with the major number and x instead of the minor number. These are created from the default branch at the point where the corresponding v*.0.0 tag was created. This provides a starting point for future maintenance releases.

  • Work toward a maintenance release is done on hotfix issue branches hotfix/1234-name created from the release branch, not the default branch. Like merges into the default branch, PRs are created to merge hotfixes back into the release branch when they're ready. Tags for the maintenance release candidates v*.*.1-rc1 and the release itself v*.*.1 are created on the release branch. Periodically, the release branch is merged back into the default branch.

  • Minor releases after the *.0.0 release are tagged on the default branch as usual — but after they're released, we merge them back out to the existing release branch. This blocks further maintenance releases for the prior minor release and provides a starting point for maintenance releases to the new minor release.

  • For website content, the staging website always reflects what is on the master branch. The documentation team provides PRs that target the master branch. When they are merged, the website workflow runs, causing the staging website content to be updated.

Workflows

We use GitHub Workflows and Actions to automate CI/CD tasks for the project when changes are made in the repository.

The build workflow builds the code and runs the unit tests with every push, on every branch. When the tests pass, some additional steps are taken depending on the trigger.

  • The scope binary and other artifacts of the build are pushed to our CDN for release tags and pushes to any branch.

  • We build container images and push them to Docker Hub for release tags.

  • We run our integration tests for pull requests to the default and release branches. We build and push the container images these tests use up to Docker Hub on pushes to the default branch.

The website workflow handles building and deploying the website/ content to https://staging.appscope.dev/ and https://appscope.dev/. The staging website is intended to always reflect the master branch. The production website is updated only when a "web" tag has been applied and pushed. See the build script in that folder for details.

The update_latest workflow updates the value returned by https://cdn.cribl.io/dl/scope/latest, and updates the latest tag at https://hub.docker.com/r/cribl/scope/tags. This workflow is run manually, and does not have any automatic triggers.

The update_latest_release workflow updates the value returned by https://cdn.cribl.io/dl/scope/latest-release. This workflow is run manually, and does not have any automatic triggers.

CDN

You can download the AppScope binary from the Cribl downloads page and follow the instructions there.

Or, you can use these CLI commands to directly download the binary and make it executable:

LATEST=$(curl -Ls https://cdn.cribl.io/dl/scope/latest)
curl -Lo scope https://cdn.cribl.io/dl/scope/$LATEST/linux/$(uname -m)/scope
curl -Ls https://cdn.cribl.io/dl/scope/$LATEST/linux/$(uname -m)/scope.md5 | md5sum -c 
chmod +x scope

Container Images

We build and push container images to the cribl/scope and cribl/scope-demo repositories at Docker Hub. See docker/ for details on how those images are built.

We currently build these for release v* tags and tag the images to match with the leading v stripped off.

docker run --rm -it cribl/scope:latest

or

docker run --rm -it cribl/scope:1.1.3

Tag Usage

next

  • Users can pick up the latest master build
  • Created automatically from master
  • Available on docker

latest

  • Users can pick up the latest release after announcement
  • Created manually on Release Day
  • Available on docker and the CDN

latest-release

  • Cribl Edge/Stream or Users can pick up the latest release before announcement
  • Created manually on GA Ready Day
  • Available on the CDN