Meet Brigade 2
For some time now, the Brigade maintainers, building upon the community’s experiences using the product, as well as our own, have been hard at work planning and implementing a major revision of the platform. Today, we’re proud to announce a milestone in that effort – the release of Brigade v2.0.0-alpha.1. While we’re still working out the kinks and cultivating an extended Brigade 2 ecosystem, we feel it is time to begin introducing the community to Brigade 2.
If you’re unfamiliar with Brigade, now might be a good time to head over to brigade.sh and get acquainted, as the rest of this post assumes at least a passing familiarity with the platform.
What Has Changed? What Hasn’t?
Two things that Brigade’s maintainers have learned over time are that success with the platform requires a fair degree of Kubernetes competency – and that with a better abstraction, users not particularly versed in Kubernetes could also find value in Brigade. So, paradoxically, Brigade 2’s biggest change is also subtle in some respects. Those familiar with Brigade 1.x might also be familiar with its tagline – “event-driven scripting for Kubernetes.” Brigade 2’s tagline might be written more appropriately as “event-driven scripting (for Kubernetes).” This reflects our effort to abstract end-users away from Kubernetes as cleanly as possible whilst not hiding it entirely from those who are both familiar and have access to the underlying cluster.
Kubernetes fading into the background as an implementation detail, necessarily, has had a broad impact on Brigade’s overall architecture. Where the Brigade 1.x CLI and event gateways all interacted directly with the Kubernetes API server for instance, their Brigade 2 counterparts interact with an all-new Brigade API. This has been an incredibly positive change, as having our very own API has granted the project latitude to do some really amazing things that weren’t previously possible.
A non-exhaustive list of highlights includes:
- Access to the underlying cluster is no longer a prerequisite for end-users.
- Support for authenticating with OpenID Connect and a compatible identity provider such as Azure Active Directory or Google Cloud Identity Platform.
- A subscription-based event bus. Gateways can emit events into Brigade without knowing who the event is for. All subscribers receive a copy of the event.
- Event data (including logs) now persists beyond the lifetime of the Kubernetes pod(s) that handle the event.
- Support for jobs composed of multiple OCI containers (i.e. jobs having sidecar containers).
- Easier and substantially improved integrations:
- Custom workers (container-based event processors) and event gateways are easier than ever to create.
- Event gateways can now follow up on the status of events they have emitted into Brigade, perhaps to report status upstream.
Other noteworthy advancements not directly attributable to the revised architecture include dramatically improved UX and support for handling events using TypeScript.
To address one final, less technical change – naming things is hard. In Brigade 1.x, a single event spawned a single build, so unsurprisingly, the term “build” came to be used almost synonymously with “event.” Using the CLI for instance, a user may have queried for “builds” rather than “events.” Over time, the maintainers have come to realize that this nomenclature promoted an inaccurate notion that Brigade is a CI platform. While CI is a notable and popular use case that Brigade addresses well, Brigade has never been constrained to such uses. In Brigade 2, we’ve stricken the term “build” from our vernacular. Now there are just events. Project subscribe to events. Brigade spawns workers to handle events.
So what hasn’t changed then?
Despite the many changes under the hood, the maintainers believe we have remained faithful to Brigade’s original vision, and as such most general knowledge of Brigade 1.x should carry over to Brigade 2. At the end of the day, Brigade 2, like its predecessor is simply about connecting events from any arbitrary source (provided a gateway exists or can be developed) with any arbitrary action (script).
A lot of love has gone into this exciting product refresh and we hope you’ll love it as much as we do.
We’re not done. With this first alpha release behind us, we’re anticipating further alpha and beta releases, and eventually release candidates, on a regular cadence.
At present, we’re very focuses on:
- Squashing bugs
- Refreshing documentation
- Feature development
- Building integrations to both broaden the ecosystem and identify bugs and
missing API features:
- A new GitHub gateway, with more gateways to come
- A team of students from Brigham Young University is working independently on a refreshed, Brigade 2 compatible Kasthi (Brigade UI).
Brigade 2 may be a work-in-progress, but if you’d like to start getting familiar, it’s not too early, and it’s the perfect time to provide feedback or even begin contributing!
So how does one get started?
Kicking the Tires
To get started, you need a Kubernetes cluster. Although we’ve worked hard to abstract Kubernetes, Brigade’s control plane is nevertheless packaged as a Helm chart for ease of deployment to Kubernetes and it uses that same Kubernetes cluster as a “workload execution substrate” (data plane).
We cannot emphasize strongly enough that if you are just kicking the tires, Brigade 2 should only be installed on a private cluster – either minikube or kind on your local machine, or a remote cluster that you do not share with anyone else. The reason for this is that the instructions that follow will install Brigade on your cluster with common, insecure defaults and no support for OpenID Connect, which is required for securing a multi-user environment.
All commands below assume you’re working within a POSIX-compliant shell such as bash, zsh, or the Windows Subsystem for Linux.
First, you’ll need Helm 3 with experimental support for OCI registries enabled:
$ export HELM_EXPERIMENTAL_OCI=1
Next, pull the chart from the GitHub Container Registry and export it to save
it somewhere on your local file system. In the example below, we store it at
$ helm chart pull ghcr.io/brigadecore/brigade:v2.0.0-alpha.1 $ helm chart export ghcr.io/brigadecore/brigade:v2.0.0-alpha.1 -d ~/charts
Now create a Kubernetes namespace to install Brigade into. We use
$ kubectl create namespace brigade2
And install Brigade with default configuration:
$ helm install brigade2 ~/charts/brigade --namespace brigade2
It can take a few minutes for everything to come up and some components may “flap” until other components upon which they depend are running and available.
If you’re well-versed in Helm, feel free to inspect and tweak configuration values, but doing so is beyond the scope of these instructions.
Exposing the API Server
Because you are presumably following these steps in a local cluster, the best method of exposing Brigade 2’s API server is to do something like this after installation:
$ kubectl --namespace brigade2 port-forward \ service/brigade2-apiserver 8443:443 &>/dev/null &
Install the CLI
Next, download the appropriate, pre-built
brig CLI (command line interface)
from our releases page and
move it to any location on your path, such as
/usr/local/bin, for instance.
If you’re already a frequent Brigade 1.x user, you may want to rename the binary
to something like
brig2 to avoid confusing it with your existing Brigade 1.x
You may also have to enabled execution of the downloaded file with
chmod 755 <path/to/file> and your OS may enforce other security restrictions
that require you to explicitly allow execution of this new binary.
Log in as the “root” user, using the default root password
F00Bar!!!. Be sure
to use the
-k option to indicate tolerance for the API server’s self-signed
$ brig -k login --server https://localhost:8443 --root
For security reasons, root user sessions are invalidated one hour after they are created. If you play with Brigade 2 for more than an hour, or you walk away and come back, you will have to log in again.
Remember, for drastically improved security, we support authentication using Open ID Connect and third-party identity providers like Azure Active Directory or Google Cloud Identity Platform, but configuring that is a bit more involved and doesn’t work well if you’re taking Brigade 2 for a test drive in a local environment like minikube or kind, so it’s beyond the scope of this introduction.
Creating a Project
Your next step is to create a Brigade project. Unlike Brigade 1.x, this is not accomplished by means of an onerous, interactive process. Rather, it is accomplished using a file that looks suspiciously like a Kubernetes manifest (but isn’t).
You can download an example from here:
With this file stored locally, at a location such as
instance, you can direct Brigade to create a new project from this file:
$ brig -k project create --file ~/pipeline-demo.yaml
Creating an Event
With your first project set up, it’s time to create your first event:
$ brig -k event create --project pipeline-demo
On success, this step will reveal the ID of the new event, which will be handled asynchronously by Brigade 2.
Watch the Event
To view the status of the event:
$ brig -k event get --id <event id from previous step>
Eventually, the worker spawned to process the event, and any jobs spawned by
the worker, should all display a
Congratulations! You’re using Brigade 2!
We’re really excited about Brigade 2 and we hope you are as well!
We invite everyone to continue beyond the simple tutorial above and begin getting familiar with this first alpha release to see what it can do – it can do a lot already. More importantly, please file issues to tell us about problems you encounter or even to make feature requests! Please also feel free to engage us in the #brigade channel on the Kubernetes Slack!
We’ll see you on the internet!