Creating a new builder

This doc describes how to set up a new builder on LUCI. It's focused on chromium builders, but parts may be applicable to other projects.

TL;DR

For a typical chromium builder using the chromium recipe, you'll need to acquire a host and then land three CLs:

  1. in infradata/config, modifying chromium.star.
  2. in chromium/tools/build, modifying the chromium_tests configuration.
  3. in chromium/src, modifying all of the following:
    1. LUCI service configurations in //infra/config
    2. Compile configuration in //tools/mb
    3. Test configuration in //testing/buildbot

Pick a name and a master

Your new builder's name should follow the chromium builder naming scheme.

We still use master names to group builders in a variety of places (even though buildbot itself is largely deprecated). FYI builders should use chromium.fyi, while other builders should mostly use chromium.$OS.

Note: If you're creating a try builder, its name should match the name of the CI builder it mirrors.

Obtain a host

When you‘re setting up a new builder, you’ll need a host to run it.

For CQ try bots, you'll likely need a large number of hosts to handle the load in parallel. To prepare for this, please file a capacity bug (internal) before proceeding.

For CI / waterfall builders or manually triggered try builders, you'll typically only need a single host.

If your bot needs to run on a physical host or needs specialized hardware (e.g. mac hardware, attached mobile devices, a specific GPU, etc.), please file a labs bug (internal). They'll help you get the host you need.

If your bot can run just fine on the Google Compute Engine virtual machines that supply the majority of our current Linux and Windows fleets, skip to the next section below.

Register hardware with swarming

Once you‘ve obtained hardware, you’ll need to associate it with your new builder in swarming. You can do so by modifying the relevant swarming instance's configuration.

This configuration is written in Starlark, and then used to generate Protobuf files which are also checked in to the repo. Chromium's configuration is in chromium.star (internal only).

If you're simply using a generic GCE bot, find the stanza corresponding to the OS and size that you want, and increment the number of bots allocated for that configuration. For example:

    # os:Ubuntu-16.04, cpu:x86-64
    chrome.gce_xenial(
        prefix = 'luci-chromium-ci-xenial-8',
        zone = 'us-central1-b',
        disk_gb = 400,
        lifetime = time.week,
-       amount = 20,
+       amount = 21,
    )

If you've been given a specific hostname, instead add an entry for your bot name to be mapped to that hostname. For example:

+   # os:Ubuntu-16.04, cpu:x86-64
+   'Linux Tests': 'swarm1234-c4',

Recipe configuration

Recipes tell your builder what to do. Many require some degree of per-builder configuration outside of the chromium repo, though the specifics vary. The recipe you use depends on what you want your builder to do.

For typical chromium compile and/or test builders, the chromium and chromium_trybot recipes should be sufficient.

To configure a chromium CI builder, you‘ll want to add a config block to the file in recipe_modules/chromium_tests corresponding to your new builder’s master name. The format is somewhat in flux and is not very consistent among the different masters, but something like this should suffice:

'your-new-builder': {
  'chromium_config': 'chromium',
  'gclient_config': 'chromium',
  'chromium_apply_config': ['mb', 'ninja_confirm_noop'],
  'chromium_config_kwargs': {
    'BUILD_CONFIG': 'Release', # or 'Debug', as appropriate
    'TARGET_BITS': 64, # or 32, for some mobile builders
  },
  'testing': {
    'platform': '$PLATFORM', # one of 'mac', 'win', or 'linux'
  },

  # Optional: where to upload test results. Valid values include:
  #   'public_server' for test-results.appspot.com
  #   'staging_server' for test-results-test.appspot.com
  #   'no_server' to disable upload
  'test_results_config': 'public_server',

  # There are a variety of other options; most of them are either
  # unnecessary in most cases or are deprecated. If you think one
  # may be applicable, please reach out or ask your reviewer.
}

For chromium try builders, you'll also want to set up mirroring. You can do so by adding your new try builder to trybots.py.

A typical entry will just reference the matching CI builder, e.g.:

TRYBOTS = freeze({
  # ...

  'tryserver.chromium.example': {
    'builders': {
      # If you want to build and test the same targets as one
      # CI builder, you can just do this:
      'your-new-builder': simple_bot({
        'mastername': 'chromium.example',
        'buildername': 'your-new-builder'
      }),

      # If you want to build the same targets as one CI builder
      # but not test anything, you can do this:
      'your-new-compile-builder': simple_bot({
        'mastername': 'chromium.example',
        'buildername': 'your-new-builder',
      }, analyze_mode='compile'),

      # If you want to build and test the same targets as a builder/tester
      # CI pair, you can do this:
      'your-new-tester': simple_bot({
        'mastername': 'chromium.example',
        'buildername': 'your-new-builder',
        'tester': 'your-new-tester',
      }),

      # If you want to mirror multiple try bots, please reach out.
    },
  },

  # ...
})

Chromium configuration

Lastly, you need to configure a variety of things in the chromium repo. It's generally ok to land all of them in a single CL.

LUCI services

LUCI services used by chromium are configured in //infra/config.

Buildbucket

Buildbucket is responsible for taking a build scheduled by a user or an agent and translating it into a swarming task. Its configuration includes things like:

  • ACLs for scheduling and viewing builds
  • Swarming dimensions
  • Recipe name and properties

Buildbucket‘s configuration schema is here. Chromium’s buildbucket configuration is here.

A typical chromium builder won‘t need to configure much. Adding a builders entry to the appropriate bucket (luci.chromium.ci for CI / waterfall, luci.chromium.try for try) with the new builder’s name, the mixin containing the appropriate master name, and perhaps one or two dimensions should be sufficient, e.g.:

buckets {
  name: "luci.chromium.ci"
  ...

  swarming {
    ...

    builders {
      name: "your-new-builder"

      # To determine what you should include here, look for an
      # existing mixin containing
      #
      #   recipe {
      #     properties: "mastername:$MASTER_NAME"
      #   }
      #
      mixins: "$MASTER_NAME_MIXIN"

      # If you're running a bunch of bots on GCE, you probably don't
      # want those bots to be keyed by buildername. Rather, they should
      # share the large pool with all the other bots using similar hardware.
      # To enable this, use the builderless mixin:
      mixins: builderless

      # Add other mixins and dimensions as necessary. You will
      # usually at least want an os dimension configured, so if
      # none of your included mixins have one, consider adding one.
    }
  }
}

Milo

Milo is responsible for displaying builders and build histories on a set of consoles. Its configuration includes the definitions of those consoles.

Milo‘s configuration schema is here. Chromium’s milo configuration is here.

A typical chromium builder should be added to one or two consoles at most: one corresponding to its master, and possibly the main console, e.g.

consoles {
  ...
  name: "$MASTER_NAME"
  ...
  builders {
    name: "buildbucket/$BUCKET_NAME/$BUILDER_NAME"

    # A builder's category is a pipe-delimited list of strings
    # that determines how a builder is grouped on a console page.
    category: "$LARGE_GROUP|$MEDIUM_GROUP|$SMALL_GROUP"

    # A builder's short name is a string up to three characters
    # long that lets someone uniquely identify it among builders
    # in the same category.
    short_name: "$ID"
  }
}

Scheduler (CI / waterfall builders only)

The scheduler is responsible for triggering CI / waterfall builders.

Scheduler‘s configuration schema is here. Chromium’s scheduler configuration is here.

A typical chromium builder will need a job configuration. A chromium builder that's triggering on new commits or on a regular schedule (as opposed to being triggered by a parent builder) will also need a trigger entry.

trigger {
  id: "master-gitiles-trigger"

  ...

  # Adding your builder to the master-gitiles-trigger
  # will cause your builder to be triggered on new commits
  # to chromium's master branch.
  triggers: "your-new-ci-builder"
}

job {
  id: "your-new-ci-builder"

  # acl_sets should either be
  #  - "default" for builders that are triggered by the scheduler
  #     (i.e. anything triggering on new commits or on a cron)
  #  - "triggered-by-parent-builders" for builders that are
  #    triggered by other builders
  acl_sets: "default"

  buildbucket: {
    server: "cr-buildbucket.appspot.com"
    bucket: "luci.chromium.ci"
    builder: "your-new-ci-builder"
  }
}

Recipe-specific configurations

chromium & chromium_trybot

The build and test configurations used by the main chromium and chromium_trybot recipes are stored src-side:

  • Build configuration: the gn configuration used by chromium recipe builders is handled by MB. MB's configuration is documented here. You only need to modify it if your new builder will be compiling.

  • Test configuration: the test configuration used by chromium recipe builders is in a group of .pyl and derived .json files in //testing/buildbot. The format is described here.

Questions? Feedback?

If you‘re in need of further assistance, if you’re not sure about one or more steps, or if you found this documentation lacking, please reach out to infra-dev@chromium.org or file a bug!