Build code from a Git repository

Northflank offers different methods for building code from a Git repository:

• You can manually or automatically build commits on specified branches or pull requests from a linked Git repository using a build service
• You can build and deploy commits on a branch from a linked Git repository with continuous integration and delivery in a combined service
• You can build commits on a branch from a linked Git repository manually or on a regular schedule using a job

Enabling CI will allow you to automatically build code when a new commit is pushed to a linked branch from your repository, or specific branches and/or pull requests. Built code can then be continuously deployed and added to a pipeline to create complex workflows.

You'll need an account on a Git provider that is linked to Northflank to get started.

To build and run code from one branch on a repository you can use:

• a combined service
• a build service linked to a deployment service
• a cron or manual job

Create a new service or job, select the repository you want to build from, then select the branch.

Once created your service will build an image for every new commit to the branches or pull requests it is monitoring, as long as CI is enabled.

You can manually trigger a build by navigating to the branch or pull request to build from and selecting a commit to build.

You can find the list of branches and pull requests in the build service menu, or by clicking the build button in the build service header.

Northflank will begin building your commit, which may take several minutes. You can monitor the progress from the builds page and click through to view the logs for the build.

You can use a build service to build and run code from specific branches and/or pull requests by specifying build rules when creating a build service. You can edit the build rules of an existing build service on the build options page.

A build service will not build any branches or pull requests without specified build rules. By default, a build service will build from all pull requests (*) and the master branch.

Example build rules:

If you have a single repository with multiple services, or your repository is structured so that your build context is not the root, you can specify the build context when creating or editing your services.

You can specify the set of files to be used in the build process by referring to a specific subdirectory relative to the root of your repository. To build all files from the root, use / as the build context or refer to a subdirectory such as /directory or /directory/subdirectory.

If you're building using a Dockerfile, its location is always assumed to be root, so you must change this as well if you want to use a Dockerfile in another directory.

You can combine the build context with path rules to create services that only build specific services or jobs in a monorepo.

You can use path rules to either ignore or only monitor specified files or directories in a repository. This allows you to trigger builds with CI only when specific files within a repository are changed, or ignore changes to files that don't necessitate a new build. You can combine this with the build context to create services or jobs that only update when the relevant code is changed.

Path rules can be useful, for example, to stop changes to documentation files such as README.md from triggering a new build, or to only trigger a build when the required microservice or job code from a monorepo is modified.

You can add path rules to a service or job that builds from a repository. You can add or modify path rules in the build options section of a service or job, under advanced build settings.

Ignore or allow

You can toggle the path rules to either ignore or allow the specified paths.

You can use an ignore list to stop commits that modify certain files or directories from triggering a new build.

You can use an allow list to trigger a build only when the specified files or directories are changed.

Path rules syntax

Path rules are written the same way as a .gitignore file , for example:

README.md               # all README.md files
node_modules/           # all files and subdirectories in the node_modules directory
*.foo                   # all files ending in .foo
**/bar                  # all directories in the repository named 'bar'
!important/file.foo     # not important/file.foo

You can specify commit message ignore flags that Northflank will detect in commit messages that will stop CI from being triggered, even if it matches other build triggers.

• If you push a commit with a message that contains one of the ignore flags you have added, a build will not be triggered for that commit
• If you push multiple commits at once, a build will not be triggered if any of the commits contain an ignore flag in their commit message
• If your service is configured to build from a pull request, that PR will not be built if the HEAD commit contains an ignore flag in its commit message

You can enable commit message ignore flags from the build options section of a service or job, under advanced build settings.

Ignore flags can be added as strings, each between square brackets and on separate lines.

Example ignore flags

[skip ci]
[ci skip]
[no ci]
[skip nf]
[nf skip]
[northflank skip]
[skip northflank]

You can enable layer caching with Dockerfile and buildpack builds to cache and reuse build images to speed up subsequent builds.