A Fabric function for git tagging

Listed below is a Fabric function for determining the appropriate git reference to deploy during a deployment. It works well with projects run using the git-flow development model.


Assume there is a test environment where:

  • the QA team to assess release candidates
  • developers to run integration tests
  • developers can deploy ‘debug’ builds from a specific (untagged) commit

There will also be stage and production environments.

Fabric function

The following function can be used as part of Fabric build script. It’s purpose is to determine the git reference to deploy from.

def determine_refspec_to_deploy_from(is_test=False)
    local('git fetch --tags')

    if is_test:
        create_tag = prompt('Tag this release? [y/N] ')
        if create_tag.lower() == 'y':
            notify("Showing latest tags for reference")
            local('git tag | sort -V | tail -5')
            refspec = prompt('Tag name [in format x.x.x]? ')
            local('git tag %(ref)s -m "Tagging version %(ref)s in fabfile"' % {
                'ref': refspec})
            local('git push --tags')
            use_commit = prompt('Build from a specific commit? [y/N] ')
            if use_commit.lower() == 'y':
                refspec = prompt('Choose commit to build from: ')
                branch = local('git branch | grep "^*" | cut -d" " -f2', capture=True)
                refspec = local('git describe %s' % branch, capture=True).strip()
        # An existing tag must be specified
        local('git tag | sort -V | tail -5')
        refspec = prompt(red('Choose tag to build from: '))

        # Check this is valid
        local('git tag | grep "%s"' % refspec)

    return refspec

Building to test

When building to test, the script allows you to:

  1. Tag a release. This is for creating release candidates for the QA team.
  2. Build without tagging. In this case, we generate a build name using git describe. This is for developers who want to update the test build to run integration tests.
  3. Build from a specific commit. This is mainly used to dig yourself out of circular reference hell: when your test build emits spurious error messages that can’t be re-created locally. A simple bisection approach works well here, building from specific commits to find the commit that broke the build.

You can build to test from any branch which is often the case with git-flow, where your next release candidate could come from develop or a release branch releases/1.4.

Building to stage and production

Nothing fancy - builds to stage and production must use an existing tag to ensure they go through the QA process.

Interesting bits

Keeping everyone in sync

The script fetches tags at the start and, if a new one is created, pushes it back to the remote. This ensures that all users have access to the tagged releases.

What’s the next tag?

This snippet shows the latest 5 tags, making it easy to determine the next tag to use:

git tag | sort -V | tail -5

Constructing a build name

For builds to test that aren’t tagged, it’s still useful to give them a build number that indicates what the latest tagged release was. This can be done with git describe, which will output something like:


which indicates that the build came from the 149th commit after tag 0.1.3.


Something wrong? Suggest an improvement or add a comment (see article history)
Tagged with: git, fabric, python, deployment
Filed in: tips

Previous: Solving MySQL connection problems caused by a dead name server
Next: Prefer data migrations to initial data

Copyright © 2005-2023 David Winterbottom
Content licensed under CC BY-NC-SA 4.0.