Skip to content

Latest commit

 

History

History
309 lines (214 loc) · 11.3 KB

complete-application-deployment.md

File metadata and controls

309 lines (214 loc) · 11.3 KB
Raw

Complete Application Deployment

The Quick Start shows how you can run the application locally, but that may not be enough if you want to experiment by scaling individual services, look at tracing data in Jaeger, monitor services via Prometheus and Grafana, or make API calls directly via Swagger UI.

To do all of the above, you need to deploy the services into a managed Kubernetes cluster in the cloud, by following the same set of steps described above (except for port forwarding, which is not necessary), and performing a few additional steps.

Additional Pre-Requisites

  1. Install envsubst

    You must download and install envsubst for your platform from https://github.com/a8m/envsubst and make it available in your PATH.

Install Prometheus and Grafana

Install the Prometheus Operator, as documented in the Prometheus Operator Quick Start page.

Prometheus can then be accessed as documented in the Access Prometheus section of the Quick Start page.

IMPORTANT If installing Prometheus into a RBAC enabled k8s cluster you may need to create the required RBAC resources as described in the Prometheus RBAC documentation.

In some environments, installation can fail with the error message "path /sys is not shared or slave mount". For this demo, if necessary, it will be sufficient to comment out mountPropagation: HostToContainer in the manifests/nodeExporter-daemonset.yaml file.

IMPORTANT If you installed the Sock Shop back-end into K8s before you installed Prometheus Operator, you must run the following to delete and re-add the deployments for Prometheus to pick up the Pods. This is because the Coherence Operator will not have been able to create Prometheus ServiceMonitor resources before Prometheus was installed.

kubectl delete -k k8s/coherence --namespace sockshop

kubectl apply -k k8s/coherence --namespace sockshop

Import the Grafana Dashboards

A set of Grafana dashboards can be downloaded from the Coherence Operator GitHub repository and imported into Grafana. Full instructions for importing dashboards into Grafana can be found in the Coherence Operator documentation.

There is an additional Sock Shop specific Grafana dashboard located in the project source k8s/optional/grafana-dashboards/sockshop-dashboard.json This file can be manually imported into Grafana using the instructions in the Add a Datasource section of the Grafana documentation.

Expose Application via a Load Balancer

Note: This is assuming you have deployed the back-ends via the instructions in the previous section.

Note: This step can be skipped if one wants to access application and Grafana/Prometheus/Jaeger services just by using localhost. In that case add port forwarding setup for each service:

$ kubectl --namespace sockshop port-forward svc/front-end 8079:80
$ kubectl --namespace monitoring port-forward svc/prometheus-k8s 9090
$ kubectl --namespace monitoring port-forward svc/alertmanager-main 9093
$ kubectl --namespace monitoring port-forward svc/grafana 3000

  1. Create the Load Balancer

    kubectl apply -f k8s/optional/ingress-controller.yaml

    You can get the status of the Load Balancer using:

    kubectl get services --namespace ingress-nginx

    The command should return a result similar to the following:

    NAME            TYPE           CLUSTER-IP       EXTERNAL-IP       PORT(S)                      AGE
ingress-nginx   LoadBalancer   AAA.BBB.CCC.DDD   WWW.XXX.YYY.ZZZ  80:31475/TCP,443:30578/TCP   17s

    Once you have been assigned an external IP address, continue to the next step.

  2. Setup Domains

    You must have access to a top level domain for which you can create sub-domains to allow access to the application via a Load Balancer (LB).

    For example if your top level domain is mycompany.com then you should create a single wildcard DNS entry *.sockshop.mycompany.com to point to your external load balancer IP address.

    Tip: You can also run Ingress just on localhost. However, in that case you must update you hosts file with the used urls.

    127.0.0.1       sockshop.mycompany.com
127.0.0.1       coherence.sockshop.mycompany.com
127.0.0.1       jaeger.coherence.sockshop.mycompany.com
127.0.0.1       mp.coherence.sockshop.mycompany.com
127.0.0.1       grafana.sockshop.mycompany.com
127.0.0.1       prometheus.sockshop.mycompany.com

  3. Create the ingress

    In your terminal, export (or SET for Windows) your top level domain and the backend you are using.

    For example for domain sockshop.mycompany.com use the following

    export SOCKSHOP_DOMAIN=sockshop.mycompany.com

envsubst -i k8s/optional/ingress.yaml | kubectl apply --namespace sockshop -f -

kubectl get ingress --namespace sockshop

NAME               HOSTS                                                                                                           ADDRESS           PORTS   AGE
mp-ingress         mp.coherence.sockshop.mycompany.com                                                                             XXX.XXX.XXX.XXX   80      12d
sockshop-ingress   coherence.sockshop.mycompany.com,jaeger.coherence.sockshop.mycompany.com,api.coherence.sockshop.mycompany.com   XXX.XXX.XXX.XXX   80      12d

  4. Create the ingress for Grafana and Prometheus

    Ensuring you have the SOCKSHOP_DOMAIN environment variable set and issue the following:

    envsubst -i k8s/optional/ingress-grafana.yaml | kubectl apply --namespace monitoring -f -

kubectl get ingress --namespace monitoring

NAME              HOSTS                                                             ADDRESS          PORTS   AGE
grafana-ingress   grafana.sockshop.mycompany.com,prometheus.sockshop.mycompany.com  XXX.YYY.XXX.YYY  80      12s

    The following URLs can be used to access Grafana and Prometheus. Please use the admin user password that was setup in the section Install Prometheus Operator above when using the URLs below:

  5. Access the application

    Access the application via the endpoint http://coherence.sockshop.mycompany.com/ (or http://localhost:8079)

Install the Jaeger Operator

  1. Install the Jaeger Operator

    The Jaeger Operator requires cert-manager to be installed. If it's missing, cert-manager can be installed with the following command:

    kubectl apply -f https://github.com/cert-manager/cert-manager/releases/download/v1.14.2/cert-manager.yaml

    If you have not done, yet, create the monitoring namespace:

    kubectl create namespace monitoring

    The command below will install the Jaeger Operator. You only need to do this once, regardless of the number of backends you want to deploy.

    kubectl create -f k8s/optional/jaeger-operator.yaml

  2. Deploy All-in-One Jaeger Instance

    kubectl create -f k8s/optional/jaeger.yaml --namespace sockshop

    Note: To access Jaeger UI over localhost add port forwarding setup:

    $ kubectl --namespace sockshop port-forward svc/jaeger-query 16686

    Note: to view not just Sockshop traces but also Coherence traces edit application.yaml and set coherence.tracing.ratio to 1.

    coherence:
  tracing:
    ratio: 1

    Due to known issues between Coherence and Spring, Coherence traces will not be properly associated with the Spring traces.

  3. Enable Jaeger tracing in the app.yaml files and rebuild and redeploy services

    Set management.tracing.enabled to true, e.g.:

    application:
  main: com.oracle.coherence.examples.sockshop.spring.carts.CartsApp
  args:
    - "--port=8080"
    - "--management.tracing.enabled=true"

  4. Exercise the Application and access Jaeger

    Accessing the Jaeger UI at http://jaeger.coherence.sockshop.mycompany.com/, you should see the trace information similar to the images below, allowing you to see how long each individual operation in the call tree took.

    Jaeger Search

    Jaeger Detail

Access Swagger

  1. Deploy Swagger UI

    kubectl create -f k8s/optional/swagger.yaml --namespace sockshop

    Access the Swagger UI at http://mp.coherence.sockshop.mycompany.com/swagger/.

    Enter /carts/swagger/carts-1.0.yml into the Explore field at the top of the screen and click on Explore button. You should see the screen similar to the following, showing you all the endpoints for the Carts service (and their payloads), and allowing you to make API requests to it directly from your browser.

    Swagger UI

Complete 🍻

Once you are done, you should have the following URLs available:

Cleanup

  1. Cleanup the ingress for applications

    To clean up the ingress for your deployment, execute the following:

    export SOCKSHOP_DOMAIN=sockshop.mycompany.com

envsubst -i k8s/optional/ingress.yaml| kubectl delete -f - --namespace sockshop

  2. Cleanup the ingress for Grafana and Prometheus

    If you installed Prometheus Operator, execute the following:

    envsubst -i k8s/optional/ingress-grafana.yaml | kubectl delete --namespace monitoring -f -

  3. Remove the deployed services

    To clean up the deployed services, execute the following:

    export SOCKSHOP_DOMAIN=sockshop.mycompany.com
kubectl delete -k k8s/coherence --namespace sockshop

  4. Remove the Load Balancer

    If you wish to remove your load balancer, execute the following:

    kubectl delete -f k8s/optional/ingress-controller.yaml

  5. Remove Jaeger

    kubectl delete -f k8s/optional/jaeger-operator.yaml

    Execute the following:

    kubectl delete -f k8s/optional/jaeger.yaml --namespace sockshop

  6. Remove Swagger

    Execute the following:

    kubectl delete -f k8s/optional/swagger.yaml --namespace sockshop

  7. Remove Prometheus and Grafana

    To remove the Prometheus Operator follow the instructions in the Remove kube-prometheus section of the Prometheus Operator Quick Start.