Configuring K8s
- Eugene Fedorenko (Unlicensed)
- Hayden Mumm
- Emily Simon
For a full walkthrough of setting up a FlexDeploy Container Project deploying to Kubernetes via Helm, check out this Getting Started guide.
When a project is being deployed to a K8s Cluster, FlexDeploy offers the ability to configure a Helm Chart or Kubernetes Manifests for the deployment. Both configurations are done on the following tab:
Field Summary
Name | Description | Notes |
|---|---|---|
App Context | Optional Parameter for the application context uri if your deployment consists of a web application. | See K8s Monitoring for more information. |
Deployment Name | This is the K8s deployment name. The field is a groovy script by default so be sure to enclose literal strings in quotes. | Result after a deployment: kubectl get deployments NAME |
Manifests to deploy | Actual K8s manifest files for your deployment. Files will run from top to bottom. | Usually consists of a deployment.yaml and service.yaml. File paths can be absolute or relative to the artifacts directory. To make use of the Deployment Name field above, your deployment.yaml should use FD_KN_DEPLOYMENT_NAME flexdeploy property and can also use the FD_PROJECT_DOCKER_IMAGE_NAME property configured on the docker tab. deployment.yamlapiVersion: apps/v1beta1
kind: Deployment
metadata:
name: ${{FD_KN_DEPLOYMENT_NAME}}
spec:
replicas: 1
template:
metadata:
labels:
app: superapp
fd_deployment_name: ${{FD_KN_DEPLOYMENT_NAME}}
fd_app_name: SpringBoot
spec:
containers:
- name: superapp
image: ${{FD_PROJECT_DOCKER_IMAGE_NAME}}
ports:
- containerPort: 8080service.yamlapiVersion: v1
kind: Service
metadata:
name: superapp
spec:
selector:
app: superapp
ports:
- port: 8080
targetPort: 8080 |
Blue/Green Istio Manifest | Istio manifest file to apply when running BlueGreen deployment strategies. If using a different strategy you can leave this blank. | The BlueGreen strategy is effectively an 'On/Off Switch' for your pods. When using this strategy all of your pods will deploy with a BLUE label, meanwhile traffic is always routed to GREEN labels. Testers and QA will test BLUE pods through back doors and other means. When ready you can 'flip' the switch to label all BLUE pods as GREEN resulting in them opening up for regular traffic. As you can see in the below manifest file, the GREEN route will always receive 100 percent of the traffic through the Istio Mesh. istio-bg.yamlapiVersion: networking.istio.io/v1alpha3
kind: Gateway
metadata:
name: superapp
spec:
selector:
istio: ingressgateway
servers:
- port:
number: 80
name: http
protocol: HTTP
hosts:
- "*"
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
name: superapp
spec:
host: superapp
subsets:
- name: green
labels:
fd_version: green
- name: blue
labels:
fd_version: blue
---
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:
name: superapp
spec:
gateways:
- superapp
hosts:
- superapp
http:
- match:
- uri:
prefix: /version
route:
- destination:
port:
number: 8080
host: superapp
subset: green
weight: 100
- destination:
port:
number: 8080
host: superapp
subset: blue
weight: 0 |
Canary Istio Manifest | Istio Manifest file to apply when running the Canary deployment strategy. If using a different deployment strategy you can leave this blank. | Canary is similar to the BlueGreen strategy but instead of being an 'On/Off Switch' think of it as a 'Dimmer Switch'. Instead of cutting all traffic over in one moment you can adjust the weights as you go. For example, you can start at routing 90 percent of traffic to GREEN and 10 percent to BLUE, then 70/30, then 50/50, then full cut-over at 0/100. You will notice the the manifest file looking similar to the BlueGreen manifest with the exception that we can see FlexDeploy variables used in the weight fields (Check out the pipeline documentation to see where these weights are set): istio-canary.yaml |
AB Istio Manifest | Istio Manifest file to apply when running the AB deployment strategy. If using a different deployment strategy you can leave this blank. | The AB deployment strategy still involves BLUE and GREEN pods like BlueGreen and Canary. The difference here is that BLUE traffic is only ever open based on a 'Routing Rule' which could be an http header of your choosing. For example, below we can anyone with the 'end-user' header set to a FlexDeploy property configured on the pipeline will get sent to a BLUE pod, and everyone else is routed to GREEN. After the end-user has validated the BLUE pods, we flip them all to GREEN similar to the BlueGreen strategy. istio-ab.yaml |
Helm Chart | The Helm chart to apply during deployment. | Can be a chart in the repository configured on the deploying endpoint, a folder with a chart, a chart package or URL. |
Helm Release Name | Release Name in Helm to easily track all deployments on the endpoint. Has nothing to do with FlexDeploy Release* | If it is unspecified, Helm will generate a random name. |
Parameters Map | Parameters to pass to the Helm Chart |
In this example we will walk you through deploying your Kubernetes project through FlexDeploy. In this example we are assuming you have Kubernetes/Docker installed on a specific endpoint.
SCM Setup
For our first step we will add our Kubernetes manifest files to some sort of SCM. For this example we will be adding a deployment manifest to a git repository. Below you can see our .yaml file we added.
Workflows
We will now create a workflow to build and deploy our manifest files. For our build we will use an existing workflow so we will not need to create a new one. For our deploy workflow we will add the Kubernetes deploy operation. For the inputs on this operation you will just need to add the context which in our case is minikube.
Once we have added this we will then activate the workflow.
Topology
We will now create our topology. First we can create our endpoint if not already created. After creating an environment and instance and associating our workflow and endpoint with our new environment instance we can set all of the properties.
We will also need to create our SCM instance in topology. This integration instance will point to our repository containing our manifest files.
Project
We will now set up our project by configuring our workflows, instance and SCM instance.
We will then go to our kubernetes config tab and set some information. First we will need the deployment name. This will be the name the deployment is given. This will be returned by kubectl get deployments once the deployment is finished. We will also set our manifest files. This will be the path to the manifest files relative to our artifacts directory. In this example we will only need to give them the name of the file.
Build and deploy
We will now build and deploy the project. The build will pull our manifest files from git and put them in our artifacts directory. Our deploy will use those manifest files in the kubectl deploy command. If everything worked you should be able to see your new deployment on kubectl get deployments and see your new deployment.
K8s Monitoring
When using the FlexagonHelmPlugin deploy operation or the FlexagonKubernetesPlugin deploy operation, FlexDeploy will capture details about your deployment, such as the deployment name and service url. For configured Container Projects a link will be saved on the Current State of Project Deployments
Deployment
If your deployed application includes a service, FlexDeploy will attempt to capture the application link for your deployment and display it here. If your application resides under some URI context, such as https://hostname/myapp, you can specify the uri under the AppContext on the Project K8s configuration tab.
Often times after deployment, an external IP may not be immediately available. In this case the service link will not work. However, if a kubectl client is configured on the FlexDeploy server, FlexDeploy will periodically check back to see if the external IP has been resolved and update the link shown here.
Status
Represents the current status of your deployment. In order for FlexDeploy to accurately determine the deployment status, a kubectl client configured with the correct contexts available is required on the FlexDeploy server. If no kubectl client is found you will simply see a placeholder icon instead.
Deployment Strategies
After defining the K8s information and Deployment Strategy Manifest files in the project configuration as laid out above, the majority of the leg work is done by FlexDeploy Pipelines, though you can accomplish the same approach through traditional Projects and Workflows.
Pipeline
Pipelines now consist of a new 'Step' called 'Apply Deployment Strategy', which we can see used below throughout the stages with the K8s icon. On the step you can set the Deployment Strategy you want to use (BlueGreen, Canary, or AB) and what weights or properties you want to apply.
For more specific information on the Apply Deployment Strategy Step see here.
The below pipeline is using the Canary Strategy updating the weights as it progresses through the stages, starting at 10-90 and ultimately finishing with a full cutover at 100 - 0.
Workflows
To use deployment strategies within FlexDeploy you need to use the Kubernetes Deploy Plugin operation within your workflow. If you have your Deployment Strategy information configured in the Pipeline as outlined above, you can defer the setting of those inputs in the workflow.
Helm Configuration
In order to deploy a Helm chart to a K8s cluster, FlexDeploy uses the deploy operation of Helm plugin, which invokes helm upgrade --install command under the hood. So, the deployment workflow should contain a step as in this example:
If Chart Name and Release Name are unspecified, FlexDeploy will take the values from the Helm Configuration tab.
FlexDeploy overrides values of the chart with the values configured in the Chart Values list. These values are specified with Groovy scripts evaluated on-the-fly during the deployment. The scripts may reference any Project or Environment-Instance (available for the project) properties.
- style