Navigation :

Building and Deploying Gloo Edge from Source

You can build and deploy Gloo Edge Open Source from the source code.

Before you begin

Follow the setup guide to clone the Gloo Edge repository and install the project dependencies.
In your terminal, navigate to the Gloo Edge project.
```
cd ${GOPATH}/src/github.com/solo-io/gloo
```
Continue with this guide to learn how you can build Gloo Edge from the top-level Makefile.

Set up build dependencies and code generation

From the Gloo Edge project directory, run:

make -B install-go-tools generated-code

The install-go-tools target installs the dependencies to build Gloo Edge in _output/.bin.

The generated-code target does multiple things:

It generates the Go source code from the .proto files in the api directories. For example, .proto files in projects/gloo/api will be generated in projects/gloo/pkg/api.
It generates the custom resource definitions (CRDs) so that Gloo Edge can interact with Kubernetes. These CRDs can be found in install/helm/gloo/crds.
It generates solo-kit resources, event loops, emitters, snapshots, and resource clients. These tools are denoted by .sk.go, and are built using solo-kit.json configuration files.

Build Gloo Edge and Docker Images

The Gloo Edge project has several products. The code for gloo itself can be found in projects/gloo. Each product has its own set of build targets.

Gloo Edge binary

To compile the gloo binary to the _output/projects/gloo/ directory, run:

make -B gloo

Docker image

To build the Docker image, run the following command. Review the following table to understand the command options.

VERSION=0.0.1 make gloo-docker -B

Option	Description
`VERSION`	An optional version number for the Docker image tag, such as `0.0.1`. The format must be valid semantic versioning.
`IMAGE_REGISTRY`	The image registry for the image, such as the local host. This value is required on `arm64` or `m1` machines.
`gloo-docker`	The name for the Docker image.

Example output:

Successfully tagged quay.io/solo-io/gloo:0.0.1

Deploy Gloo Edge

You can choose from several Gloo Edge installation options. This guide assumes you deploy Gloo Edge into a Kubernetes cluster that runs locally in kind.

Install gloo into the cluster, if not already present.
```
glooctl install gateway
```
Make the image accessible to the kind cluster.
- x86_64
- arm64 or m1
```
kind load docker-image quay.io/solo-io/gloo:0.0.1
```
```
kind load docker-image quay.io/solo-io/gloo:0.0.1
```
Update the kind cluster to use the new image as the template. Note that the image tag varies depending on your machine and the image registry and tag version that you previously used.
- x86_64
- arm64 or m1
```
kubectl -n gloo-system set image deployments/gloo gloo=quay.io/solo-io/gloo:0.0.1
```
```
kubectl -n gloo-system set image deployments/gloo gloo=quay.io/solo-io/gloo:0.0.1
```

Verify that Kubernetes removes the old pod and spins up a new pod.

Tip: You might find it easier to observe the update in real time by using a live monitoring tool like k9s.

$ kubectl -n gloo-system get pods
NAME                            READY   STATUS    RESTARTS   AGE
discovery-77d7cd499d-vg2rl      1/1     Running   0          82s
gateway-proxy-5b88d8995-hzt27   1/1     Running   0          82s
gloo-6578f95696-4fdth           1/1     Running   0          82s
gloo-bdd7ddb9d-l977c            0/1     Running   0          9s

After a few seconds, the old pod is deleted:

$ kubectl -n gloo-system get pods
NAME                            READY   STATUS    RESTARTS   AGE
discovery-77d7cd499d-vg2rl      1/1     Running   0          89s
gateway-proxy-5b88d8995-hzt27   1/1     Running   0          89s
gloo-bdd7ddb9d-l977c            1/1     Running   0          16s

Make and debug changes

Now that you can build and deploy your own builds, you can start editing Gloo Edge itself. An in-depth discussion on making code changes is outside the scope of this guide. However, a brief overview of Gloo’s software architecture will make things easier to understand.

Gloo Edge is primarily composed of plugins. You can find the code for these plugins in projects/gloo/pkg/plugins. Gloo Edge has a lot of them! Generally speaking, these plugins take Gloo Edge configuration as input, and translate Envoy configuration as output. You can test any changes you make to plugins by following two main steps:

Unit tests, including dynamic and static breakpoints
Deploying the code

Unit tests

The fastest way to test and debug changes is by using the unit tests in these directories. The unit tests are fast to recompile and re-execute, and integrate smoothly with the debugger of your choice. You can find the unit tests for a plugin in its directory.

Navigate to the directory of a plugin, such as cross-site request forgery (CSRF).
```
cd projects/gloo/pkg/plugins/csrf/
```
Run the plugin unit tests. You might use ginkgo or gdb.
- ginkgo
- gdb
```
ginkgo
```
1. Create a test file for the plugin by using Go.
  go test -c
  This command creates a a file called csrf.test.
2. Run gdb against the test file.
  gdb csrf.test

Dynamic breakpoints

Dynamic breakpoints can be added via gdb without having to rebuild the unit tests.

List the available source files. The plugin.go argument is optional but helpful to filter down all the source files available.
```
info sources plugin.go
```
Specify the breakpoint by copying the full path of the filename, followed by the line on which to place it. For example:
```
b /home/user/go/src/github.com/solo-io/gloo/projects/gloo/pkg/plugins/csrf/plugin.go:91
```

Static breakpoints

Breakpoints can be added statically by editing the source code and adding runtime.Breakpoint() at the desired point in the code. Then rebuild and re-run in gdb. gdb breaks when it reaches that line. Then from that line, you can go up the stack to inspect variables.

Note that adding runtime.Breakpoint() in the code will likely cause the Go runtime to abort, because Go does not know how to handle the break signal. Therefore, you can only run the binary in gbd, not Go or ginkgo.

Testing in deployment

Testing in deployment can take time to set up and run. Make sure that you get the unit tests updated and passing before deploying.

Rebuild the Docker image.
Update the cluster.
Check any output to stdout by viewing the logs.
```
kubectl -n gloo-system logs deployments/gloo
```