Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

docs: restructure deploy doc #567

Merged
merged 11 commits into from
Jun 26, 2020
13 changes: 12 additions & 1 deletion content/docs/faq.md
Original file line number Diff line number Diff line change
Expand Up @@ -159,4 +159,15 @@ The `project.yaml` file is validated when a new Appsody project is initialized.
* moved to a different directory
* are being initialized into the same directory as a previous project

Their entry is removed from the `project.yaml` file and the associated volumes are deleted. This validation ensures that volumes exist for current projects only and that volumes are not reused between different projects of the same name.
Their entry is removed from the `project.yaml` file and the associated volumes are deleted. This validation ensures that volumes exist for current projects only and that volumes are not reused between different projects of the same name.

## 12. How do I deploy my application to a local Minikube cluster?
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this title is right. AFAIK, appsody can deploy to minikube fine - no special config should be needed. Please can you confirm this once?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The intent was to document how to use Docker daemon from minikube vs Docker on host.


Currently, Appsody supports you using a local Docker daemon only. If you use the Minikube Docker daemon, the following commands do not work as expected:
- `eval $(minikube docker-env)`
- [appsody extract](docs/cli-commands/#appsody-extract)
- [appsody build](docs/cli-commands/#appsody-build)

To work around this issue, you can use your local Docker daemon as follows:
1. Unset the `DOCKER_HOST`, `DOCKER_CERT_PATH` and `DOCKER_TLS_VERIFY` environment variables.
2. Use [appsody deploy --push](docs/cli-commands/#appsody-deploy) to push your image to an external Docker registry, and configure Minikube to pull from this registry.
72 changes: 31 additions & 41 deletions content/docs/using-appsody/deploying.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,55 +9,59 @@ The Appsody CLI provides the [appsody deploy](/docs/cli-commands/#appsody-deploy

You can delegate the build and deployment steps to an external pipeline, such as a Tekton pipeline that consumes the source code of your Appsody project after you push it to a GitHub repository. Within the pipeline, you can run [appsody build](/docs/using-appsody/building), which builds the application image and generates a deployment manifest. You can use the manifest to deploy your application to a Kubernetes environment where the Appsody operator is installed.

> Currently, Appsody supports you using a local Docker daemon only. To work around this, see the [FAQ](/docs/faq/#12-how-do-i-deploy-my-application-to-a-local-minikube-cluster).

These deployment options are covered in more detail in the following sections.

## Deploying your app directly to a Kubernetes cluster
## Deploying your application to a Kubernetes cluster
There are many options to deploy your Appsody applications to a Kubernetes cluster. The best approach depends on the specific scenario:
- If you are testing your app on a locally installed cluster, using `appsody deploy` is your best bet
- If you intend to have your app deployed on a shared cluster for integration testing or production, you are probably going to rely on CI/CD pipelines, and have the app built and deployed from its source.
- If you intend to deploy your application on a shared cluster for integration testing or production, you are probably going to rely on CI/CD pipelines. In this case, the application is built and deployed from its source.
- If you are testing your application on a locally installed cluster, you would want to use [appsody deploy](/docs/cli-commands/#appsody-deploy).

The `appsody deploy` command provides a way for you to deploy your application directly to a Kubernetes cluster. The stack contains a deployment manifest that can be consumed by the [Appsody operator](https://operatorhub.io/operator/appsody-operator). `appsody deploy` will install the operator, if necessary, and deploy your application to the cluster using that deployment manifest.
If your development workstation has a Kubernetes cluster installed, you can use your local Docker image cache instead of pushing the image to Docker Hub. To do this, you need to configure your Kubernetes cluster to use images from the local Docker [cache](https://kubernetes.io/docs/concepts/containers/images/#pre-pulled-images).

If you want to deploy your application without rebuilding the application image, or modifying the deployment manifest, you can run
To deploy your Appsody project locally, run:
```
appsody deploy --no-build
appsody deploy
```
This command completes the following actions:

- Calls `appsody build` and creates a *deployment* Docker image and a deployment manifest file named `app-deploy.yaml`.
- If you specified the `--knative` flag, or if Knative is the only deployment option for your stack, the command tags the image with the prefix `dev.local`, making it accessible to your Kubernetes cluster (assuming you followed [these directions](/docs/using-appsody/installing-knative-locally)).
- The deployment manifest issues a `kubectl apply -f` command against the target Kubernetes cluster so that the application can be deployed by the [Appsody operator](https://operatorhub.io/operator/appsody-operator).

> To deploy your application without rebuilding the application image, or modifying the deployment manifest, run `appsody deploy --no-build`.

### Ensuring the latest application code changes get deployed
Some users have noticed that their code changes do not seem to be published to the target Kubernetes cluster after an initial deployment of the Appsody project through `appsody deploy`.

If you issue `appsody deploy` without explicitly tagging the image, you end up with an identical deployment manifest (`app-deploy.yaml` file) to the one that was initially used to deploy the application. Therefore, Kubernetes will detect no differences in the deployment manifest, and will not update your application.

To ensure the latest version of your application is pushed to the cluster, use the `-t` flag to add a unique tag every time you redeploy your application. Kubernetes then detects a change in the deployment manifest, and pushes your application to the cluster again. For example: `appsody deploy -t dev.local/my-image:0.x`, where x is a number that you increment every time you redeploy.

> If you are running multiple Appsody projects on your workstation, you can use the [appsody deploy](/docs/cli-commands/#appsody-deploy) and [appsody operator](/docs/cli-commands/#appsody-operator) commands to deploy them to a Kubernetes cluster. However, do not run these commands concurrently as they create temporary files that might lead to conflicts.

### Deployment via the Appsody Operator
Kubernetes operators offer a powerful way to provide full lifecycle maintenance of a wide range of resources on Kubernetes clusters. In particular, they can install, upgrade, remove, and monitor application deployments. The recently published [Appsody operator](https://operatorhub.io/operator/appsody-operator) automates the installation and maintenance of a special type of Custom Resource Definitions (CRDs), called **AppsodyApplication**.

The Appsody stacks that are currently available include a template of such a CRD manifest. When you run `appsody deploy` on a project created from one of the stacks enabled with those manifests, the CLI customizes the manifest with information that is specific to the deployment (such as namespace and project name), and submits the manifest to the Appsody operator on the Kubernetes cluster.
The currently available Appsody stacks include a template of such a CRD manifest. When you run `appsody deploy` on a project created from one of the stacks enabled with those manifests, the CLI customizes the manifest with information that is specific to the deployment (e.g. namespace and project name), and submits the manifest to the Appsody operator on the Kubernetes cluster.

In fact, if your cluster does not already provide an operator, `appsody deploy` will install one for you. You can also use the Appsody CLI to install an instance of the Appsody operator, without installing any applications. This can be achieved by running the `appsody operator install` command.

To find out more about the Appsody operator, see [here](/docs/reference/appsody-operator).

### Deployment as a Knative Service

You can deploy your application as a Knative service on your target Kubernetes cluster by using the `--knative` flag with the `appsody build` or `appsody deploy` commands. This action sets the flag `createKnativeService` in the deployment manifest to `true`.
You can deploy your application as a Knative service on your target Kubernetes cluster by using the `--knative` flag. This flag is available to the `appsody build` or `appsody deploy` commands. This action sets the `createKnativeService` value in the deployment manifest to `true`.

For your app to work as a Knative service, the following **pre-requisites** apply:
To deploy your application as a Knative service, the following **pre-requisites** apply:

- You must have access to a Kubernetes cluster, with Knative Serving installed and running. To install Knative locally, use the Kubernetes feature in Docker for Desktop, see [Installing Knative Locally](/docs/faq#10-how-do-i-set-up-knative-serving-for-local-kubernetes-development). To install Knative on other Kubernetes clusters, see the [Knative Install Documentation](https://knative.dev/docs/install/).
- You must configure your `kubectl` CLI to point to your Kubernetes cluster.
- If you intend to push the Docker image containing your application to Docker Hub, your target cluster must be configured to pull images from Docker Hub.

Once the `appsody deploy --knative` command completes successfully, the Knative Service is operable at the URL specified in the command output.

### Deploying your application to a local Kubernetes cluster

If you have installed a Kubernetes cluster on your development workstation and want to use your local Docker image cache instead of pushing the image to Docker Hub, make sure you set up your Kubernetes cluster to consume images from the local Docker cache.

To deploy your Appsody project locally, run:
```
appsody deploy
```
This command completes the following actions:

- Calls `appsody build` and creates a *deployment* Docker image and a manifest file named `app-deploy.yaml`, as described in the previous section.
- If you specified the `--knative` flag, or if Knative is the only deployment option for your stack, the command tags the image with the special prefix `dev.local`, making it accessible to your Kubernetes cluster (assuming you followed [these directions](/docs/using-appsody/installing-knative-locally))
- The deployment manifest, `app-deploy.yaml`, is used to issue a `kubectl apply -f` command against the target Kubernetes cluster so that the application can be deployed by the Appsody Operator.

### Deploying your application through Docker Hub

If your cluster is configured to pull images from Docker Hub, use the following command to deploy your application:
Expand Down Expand Up @@ -87,26 +91,12 @@ appsody deploy -t <mynamespace/myrepository[:tag]> --push-url <external-registry
```
> Note: The pull registry url gets injected into the deployment manifest for Kubernetes to pull the correct image.

### Deploying multiple projects
If you are running multiple Appsody projects on your workstation, you can use `appsody deploy` and `appsody operator` commands to get them deployed to a Kubernetes cluster. However, make sure that you run these commands one at a time, because those commands create temporary files that might lead to conflicts if created concurrently.

### Ensuring the latest application code changes get deployed
Some users have noticed that their code changes do not seem to be published to the target Kubernetes cluster after an initial deployment of the Appsody project through `appsody deploy`.
The sequence of actions that leads to this behavior is as follows:
1. You create an initial version of your app, and then use `appsody deploy` to publish it to your test Kubernetes cluster.
1. You test your app, and make code changes. The code changes appear as you re-test your app using `appsody run`.
1. You decide to re-publish your app to your target cluster, and run `appsody deploy` again.
1. The command succeeds, yet nothing seems to change on the Kubernetes cluster: you still observe the older version of your app.

This behavior can be explained by the fact that - if you simply issue `appsody deploy` without explicitly tagging the image - you end up with a deployment manifest (the `app-deploy.yaml` file) that is identical to the one that was used to deploy the application the first time. Therefore, Kubernetes will detect no differences in the deployment yaml, and will do nothing to update your app.

To ensure that the latest version of your app is pushed to the cluster, use the -t flag to add a unique tag every time you redeploy your app. Kubernetes then detects a change in the deployment manifest, and pushes your app to the cluster again. For example: appsody deploy -t dev.local/my-image:0.x, where x is a number that you increment every time you redeploy.
## Deploying your app through a Tekton pipeline
## Deploying your application through a Tekton pipeline

> This deployment option is under development

Most likely, the deployment of apps created with the Appsody CLI is going to occur through the invocation of a CI/CD build pipeline.
Most likely, the deployment of applications that are created with the Appsody CLI is going to occur through the invocation of a CI/CD build pipeline.

As a developer, you develop your app using the Appsody CLI, and when you are ready to deploy, you push your code to a repo or create a pull request on GitHub.
As a developer, you develop your application using the Appsody CLI, and when you are ready to deploy, you push your code to a repository or create a pull request on GitHub.

This [example](https://github.com/appsody/tekton-example) shows you how to use Tekton pipelines to deploy your app to a Kubernetes cluster. More details on running the Tekton pipeline example for Appsody can be found in the repo [readme file] (https://github.com/appsody/tekton-example/blob/master/README.md). The example uses a [customized Buildah image with the Appsody CLI installed](https://github.com/appsody/appsody-buildah). For more information on using Appsody with Buildah, see the [FAQ](/docs/faq#9-can-i-use-appsody-without-docker).
This [example](https://github.com/appsody/tekton-example) shows you how to use Tekton pipelines to deploy your application to a Kubernetes cluster. More details on running the Tekton pipeline example for Appsody can be found in the repository [README](https://github.com/appsody/tekton-example/blob/master/README.md). The example uses a [customized Buildah image](https://github.com/appsody/appsody-buildah) with the Appsody CLI installed. For more information on using Appsody with Buildah, see the [FAQ](/docs/faq#9-can-i-use-appsody-without-docker).