to build the application and the [TriggerMesh CLI](https://github.com/triggermesh/tm) to simplify the
deployment of knative services and functions.
1.**`serverless.yml`** (for [functions only](#deploying-functions)): When using serverless to deploy functions, the `serverless.yml` file
1.**`serverless.yaml`** (for [functions only](#deploying-functions)): When using serverless to deploy functions, the `serverless.yaml` file
will contain the information for all the functions being hosted in the repository as well as a reference to the
runtime being used.
1.**`Dockerfile`:** Knative requires a `Dockerfile` in order to build your application. It should be included
1.**`Dockerfile`** (for [applications only](#deploying-serverless-applications): Knative requires a `Dockerfile` in order to build your application. It should be included
at the root of your project's repo and expose port `8080`.
## Installing Knative via GitLab's Kubernetes integration
...
...
@@ -75,11 +73,16 @@ The minimum recommended cluster size to run Knative is 3-nodes, 6 vCPUs, and 22.
> Introduced in GitLab 11.6.
Using functions is useful for initiating, responding, or triggering independent
Using functions is useful for dealing with independent
events without needing to maintain a complex unified infrastructure. This allows
you to focus on a single task that can be executed/scaled automatically and independently.
In order to deploy functions to your Knative instance, the following templates must be present:
Currently the following [runtimes](https://gitlab.com/triggermesh/runtimes) are offered:
- node.js
- kaniko
In order to deploy functions to your Knative instance, the following files must be present:
1.`.gitlab-ci.yml`: This template allows to define the stage, environment, and
image to be used for your functions. It must be included at the root of your repository:
...
...
@@ -97,11 +100,16 @@ In order to deploy functions to your Knative instance, the following templates m
2.`serverless.yml`: This template contains the metadata for your functions,
such as name, runtime, and environment. It must be included at the root of your repository:
The `gitlab-ci.yml` template creates a `Deploy` stage with a `functions` job that invokes the `tm` CLI with the required parameters.
2.`serverless.yaml`: This file contains the metadata for your functions,
such as name, runtime, and environment. It must be included at the root of your repository. The following is a sample `echo` function which shows the required structure for the file.
NOTE: **Note:**
The file extension for the `serverless.yaml` file must be specified as `.yaml` in order to the file to be parsed properly. Specifying the extension as `.yml` will not work.
After the templates have been created, each function must be defined as a single
file in your repository. Committing a function to your project will result in a
The `serverless.yaml` file contains three sections with distinct parameters:
### `service`
| Parameter | Description |
|-----------|-------------|
| `service` | Name for the Knative service which will serve the function. |
| `description` | A short description of the `service`. |
### `provider`
| Parameter | Description |
|-----------|-------------|
| `name` | Indicates which provider is used to execute the `serverless.yaml` file. In this case, the TriggerMesh `tm` CLI. |
| `registry-secret` | Indicates which registry will be used to store docker images. The sample function is using the GitLab Registry (`gitlab-registry`). A different registry host may be specified using `registry` key in the `provider` object. If changing the default, update the permission and the secret value on the `gitlab-ci.yml` file |
| `environment` | Includes the environment variables to be passed as part of function execution for **all** functions in the file, where `FOO` is the variable name and `BAR` are he variable contents. You may replace this with you own variables. |
### `functions`
In the `serverless.yaml` example above, the function name is `echo` and the subsequent lines contain the function attributes.
| Parameter | Description |
|-----------|-------------|
| `handler` | The function's file name. In the example above, both the function name and the handler are the same. |
| `runtime` | The runtime to be used to execute the function. |
| `description` | A short description of the function. |
| `buildargs` | Pointer to the function file in the repo. In the sample the function is located in the `echo` directory. |
| `environment` | Sets an environment variable for the specific function only. |
After the `gitlab-ci.yml` template has been added and the `serverless.yaml` file has been
created, each function must be defined as a single file in your repository. Committing a
function to your project will result in a
CI pipeline being executed which will deploy each function as a Knative service.
Once the deploy stage has finished, additional details for the function will
appear under **Operations > Serverless**.
...
...
@@ -149,6 +181,10 @@ The function details can be retrieved directly from Knative on the cluster:
kubectl -n"$KUBE_NAMESPACE" get services.serving.knative.dev
```
The sample function can now be triggered from any HTTP client using a simple `POST` call.
Currently, the Serverless page presents all functions available in all clusters registered for the project with Knative installed.
## Deploying Serverless applications
...
...
@@ -190,17 +226,12 @@ deploy:
## Deploy the application with Knative
With all the pieces in place, you can create a new CI pipeline to deploy the Knative application. Navigate to
**CI/CD > Pipelines** and click the **Run Pipeline** button at the upper-right part of the screen. Then, on the
Pipelines page, click **Create pipeline**.
With all the pieces in place, the next time a CI pipeline runs, the Knative application will be deployed. Navigate to
**CI/CD > Pipelines** and click the most recent pipeline.
## Obtain the URL for the Knative deployment
There are two ways to obtain the URL for the Knative deployment.
### Using the CI/CD deployment job output
Once all the stages of the pipeline finish, click the **deploy** stage.
Use the CI/CD deployment job output to obtain the deployment URL. Once all the stages of the pipeline finish, click the **deploy** stage.
