bulldozer
is a GitHub App that
automatically merges pull requests (PRs) when (and only when) all required
status checks are successful and required reviews are provided.
Additionally, bulldozer
can:
Bulldozer might be useful if you:
bulldozer
will only merge pull requests that GitHub allows non-admin
collaborators to merge. This means that all branch protection settings,
including required status checks and required reviews, are respected. It also
means that you must enable branch protection to prevent bulldozer
from
immediately merging every pull request.
Only pull requests matching the trigger conditions (or not matching
ignore conditions) are considered for merging. bulldozer
is event-driven,
which means it will usually merge a pull request within a few seconds of the
pull request satisfying all preconditions.
The behavior of the bot is configured by a .bulldozer.yml
file at the root of
the repository. The file name and location are configurable when running your
own instance of the server.
The file is read from the most recent commit on the target branch of each pull request.
The file may contain a reference to a configuration in a different repository (see Remote Configuration.)
If the file does not exist in the repository, bulldozer
tries to load a
shared bulldozer.yml
file at the root of the .github
repository in the
same organization. You can change this path and repository name when running
your own instance of the server.
You can also define a global default configuration when running your own instance of the server. This is used if the repository or the shared organization repository do not define any configuration.
If configuration does not exist in the repository or in the shared
organization repository and the server does not have a default configuration,
bulldozer
does not act on the pull request. This means it is safe to enable
bulldozer
on all repositories in an organization.
The .bulldozer.yml
file supports the following keys.
# "version" is the configuration version, currently "1".
version: 1
# "merge" defines how and when pull requests are merged. If the section is
# missing, bulldozer will consider all pull requests and use default settings.
merge:
# "trigger" defines the set of pull requests considered by bulldozer. If
# the section is missing, bulldozer considers all pull requests not excluded
# by the ignore conditions.
trigger:
# Pull requests with any of these labels (case-insensitive) are added to
# the trigger.
labels: ["merge when ready"]
# Pull requests where the body or any comment contains any of these
# substrings are added to the trigger.
comment_substrings: ["==MERGE_WHEN_READY=="]
# Pull requests where any comment matches one of these exact strings are
# added to the trigger.
comments: ["Please merge this pull request!"]
# Pull requests where the body contains any of these substrings are added
# to the trigger.
pr_body_substrings: ["==MERGE_WHEN_READY=="]
# Pull requests targeting any of these branches are added to the trigger.
branches: ["develop"]
# Pull requests targeting branches matching any of these regular expressions are added to the trigger.
branch_patterns: ["feature/.*"]
# Pull requests with auto merge enabled are added to the trigger.
auto_merge: true
# "ignore" defines the set of pull request ignored by bulldozer. If the
# section is missing, bulldozer considers all pull requests. It takes the
# same keys as the "trigger" section.
ignore:
labels: ["do not merge"]
comment_substrings: ["==DO_NOT_MERGE=="]
# "method" defines the merge method. The available options are "merge",
# "rebase", "squash", and "ff-only".
method: squash
##### branch_method has been DEPRECATED in favor of merge_method #####
#
# Allows the merge method that is used when auto-merging a PR to be different
# target branch. The keys of the hash are the target branch name, and the values are the merge method that
# will be used for PRs targeting that branch. The valid values are the same as for the "method" key.
# Note: If the target branch does not match any of the specified keys, the "method" key is used instead.
branch_method:
develop: squash
master: merge
##### branch_method has been DEPRECATED in favor of merge_method #####
# Allows the merge method that is used when auto-merging a PR to be different
# based on trigger criteria. The first method where ALL triggers match will
# be used. Otherwise, the method specified previously in "merge.method" will
# be used.
# - ALL trigger criteria must match, unlike merge/trigger where ANY match
# will trigger bulldozer.
# - This will override any branch_method logic if one of the methods is
# triggered
# - If no trigger criteria is provided the method is ignored
merge_method:
# "method" defines the merge method. The available options are "merge",
# "rebase", "squash", and "ff-only".
- method: squash
trigger:
# All methods from merge/trigger are supported. Additionally, the
# following additional methods are provided:
# Pull requests which a number of commits less than or equal to this value are added to the trigger.
max_commits: 3
# "options" defines additional options for the individual merge methods.
options:
# "squash" options are only used when the merge method is "squash"
squash:
# "title" defines how the title of the commit message is created when
# generating a squash commit. The options are "pull_request_title",
# "first_commit_title", and "github_default_title". The default is
# "pull_request_title".
title: "pull_request_title"
# "body" defines how the body of the commit message is created when
# generating a squash commit. The options are "pull_request_body",
# "summarize_commits", and "empty_body". The default is "empty_body".
body: "empty_body"
# If "body" is "pull_request_body", then the commit message will be the
# part of the pull request body surrounded by "message_delimiter"
# strings. This is disabled (empty string) by default.
message_delimiter: ==COMMIT_MSG==
# "required_statuses" is a list of additional status contexts that must pass
# before bulldozer can merge a pull request. This is useful if you want to
# require extra testing for automated merges, but not for manual merges.
required_statuses:
- "ci/circleci: ete-tests"
# If true, bulldozer will delete branches after their pull requests merge.
delete_after_merge: true
# If true, bulldozer will merge pull requests with no required checks. This
# helps to protect against merging branches which inadvertently do not have
# required status checks.
allow_merge_with_no_checks: false
# "update" defines how and when to update pull request branches. Unlike with
# merges, if this section is missing, bulldozer will not update any pull requests.
update:
# "trigger" defines the set of pull requests that should be updated by
# bulldozer. It accepts the same keys as the trigger in the "merge" block.
trigger:
labels: ["WIP", "Update Me"]
# "ignore" defines the set of pull requests that should not be updated by
# bulldozer. It accepts the same keys as the ignore in the "merge" block.
ignore:
labels: ["Do Not Update"]
# If true, bulldozer will ignore updating draft pull requests, unless they
# explicitly match a configured trigger condition.
ignore_drafts: false
# "required_statuses" is a list of additional status contexts that must pass
# before bulldozer will update a pull request, unless the pull request
# explicitly matches a configured trigger condition. This is useful if you want
# to require certain statuses to pass before automated updates are made.
required_statuses:
- "policy-bot: develop"
You can use a remote configuration by specifying a repository and an optional
path and Git reference. Place the following in the repository's
.bulldozer.yml
file instead of the normal configuration:
# The remote repository to read the configuration file from. This is required,
# and must be in "org/repo-name" form. Must be a public repository.
remote: org/repo-name
# The path to the configuration file in the remote repository. If not set,
# uses the default configuration path.
path: path/to/bulldozer.yml
# The branch (or tag, or commit hash) that should be used on the remote
# repository. If not set, uses the default branch of the repository.
ref: main
The remote file must contain bulldozer
configuration and cannot be another
remote reference. However, the organization-level default configuration may be
a remote reference.
ignore
and trigger
?Yes. If both ignore
and trigger
are specified, bulldozer will attempt to match
on both. In cases where both match, ignore
will take precedence.
squash
strategy?Yes. When the merge strategy is squash
, you can set additional options under the
options.squash
property, including how to render the commit body.
merge:
method: squash
options:
squash:
body: summarize_commits # or `pull_request_body`, `empty_body`
You can also define part of pull request body to pick as a commit message when
body
is pull_request_body
.
merge:
method: squash
options:
squash:
body: pull_request_body
message_delimiter: ==COMMIT_MSG==
Anything that's contained between two ==COMMIT_MSG==
strings will become the
commit message instead of whole pull request body.
You can add default repository configuration in your bulldozer config file.
It will be used only when your repo config file does not exist.
options:
default_repository_config:
ignore:
labels: ["do not merge"] # or any other available config.
Bulldozer will attempt to merge a branch whenever it passes the trigger/ignore criteria. GitHub may prevent it from merging a branch in certain conditions, some of which are to be expected, and others that may be caused by mis-configuring Bulldozer.
.bulldozer.yml
is not allowed by your
repository settingsbulldozer[bot]
from pushing to the
branch. Github apps can be added to the list of restricted
push users, so you can allow Bulldozer specifically for your repo.allow_merge_with_no_checks
is set to
the default value (false
).When using the branch update functionality, Bulldozer only performs an update after updates are enabled when:
For example:
develop
develop
update me
comment to the first pull requestdevelop
Note that the update does not happen when the update me
comment is added,
even though there is a new commit on develop
that is not part of the pull
request.
As mentioned above, as of Github ~2.19.x, GitHub Apps can be added to the list of users associated with push restrictions. If you don't want to do this, or if you're running an older version of Github that doesn't support this behaviour, you may work around this:
Use another app like policy-bot to implement approval restrictions as required status checks instead of using push restrictions. This effectively limits who can push to a branch by requiring changes to go through the pull request process and be approved.
Configure Bulldozer to use a personal access token for a regular user to
perform merges in this case. The token must have the repo
scope and the
user must be allowed to push to the branch. In the server configuration
file, set:
options:
push_restriction_user_token: <token-value>
The token is only used if the target branch has push restrictions enabled. All other merges are performed as the normal GitHub App user.
bulldozer is easy to deploy in your own environment as it has no dependencies other than GitHub. It is also safe to run multiple instances of the server, making it a good fit for container schedulers like Nomad or Kubernetes.
We provide both a Docker container and a binary distribution of the server:
A sample configuration file is provided at config/bulldozer.example.yml
.
Certain values may also be set by environment variables; these are noted in the
comments in the sample configuration file. By default, the environment
variables for server values are prefixed with BULLDOZER_
(e.g.
BULLDOZER_PORT
). For configuring options the prefix is
BULLDOZER_OPTIONS
(e.g. BULLDOZER_OPTIONS_APP_NAME
)
This prefix can be overridden by setting the
BULLDOZER_ENV_PREFIX
environment variable.
To configure Bulldozer as a GitHub App, these general options are required:
http(s)://<your-bulldozer-domain>/api/github/hook
github.app.webhook_secret
property in the server configurationThe app requires these permissions:
Permission | Access | Reason |
---|---|---|
Repository administration | Read-only | Determine required status checks |
Checks | Read-only | Read checks for ref |
Repository contents | Read & write | Read configuration, perform merges |
Issues | Read & write | Read comments, close linked issues |
Repository metadata | Read-only | Basic repository data |
Pull requests | Read & write | Merge and close pull requests |
Commit status | Read-only | Evaluate pull request status |
The app should be subscribed to these events:
bulldozer uses go-baseapp and go-githubapp, both of which emit standard metrics and structured log keys. Please see those projects for details.
Example .bulldozer.yml
files can be found in config/examples
The server configuration for bulldozer allows you to specify configuration_v0_path
, which is a list of paths
to check for 0.4.X
style bulldozer configuration. When a 1.X
style configuration file does not appear
at the configured path, bulldozer will attempt to read from the paths configured by configuration_v0_path
,
converting the legacy configuration into an equivalent v1
configuration internally.
The upgrade process is therefore to deploy the latest version of bulldozer with both configuration_path
and
configuration_v0_path
configured, and to enable the bulldozer GitHub App on all organizations where it was
previously installed.
The Bulldozer "Update" feature is useful when keeping branches up-to-date, but can cause lots of pressure on CI build systems and Github when there are many pull requests open on a single repository.
To reduce pressure on CI systems and Github, the update feature can be disabled at the server level by specifying the following server option:
options:
disable_update_feature: true
To develop bulldozer
, you will need a Go installation.
Run style checks and tests
./godelw verify
Running the server locally
# copy and edit the server config
cp config/bulldozer.example.yml config/bulldozer.yml
./godelw run bulldozer server
config/bulldozer.yml
is used as the default configuration filehttp://localhost:8080/
Running the server via Docker
# copy and edit the server config
cp config/bulldozer.example.yml config/bulldozer.yml
# build the docker image
./godelw docker build --verbose
docker run --rm -v "$(pwd)/config:/secrets/" -p 8080:8080 palantirtechnologies/bulldozer:latest
config
directory (which should contain the bulldozer.yml
configuration file) at the expected locationhttp://localhost:8080/
Contributions and issues are welcome. For new features or large contributions, we prefer discussing the proposed change on a GitHub issue prior to a PR.
This application is made available under the Apache 2.0 License.