Azure
/
orkestra
зеркало из https://github.com/Azure/orkestra.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
orkestra/docs/examples.md

9.4 KiB
Исходник Постоянная ссылка Ответственный История

layout title nav_order
default Examples 3

Examples

Prerequisites

1. Bookinfo without Quality Gates

In this example we deploy an application group consisting of two demo applications,

  • Istio bookinfo app (with subcharts) : source
  • Ambassador : source

Orkestra Helm Chart without Keptn

From the root directory of the repository, run:

helm upgrade --install orkestra chart/orkestra -n orkestra --create-namespace

Install the ApplicationGroup:

kubectl apply -f examples/simple/bookinfo.yaml

applicationgroup.orkestra.azure.microsoft.com/bookinfo created

The orkestra controller logs should look as follows on success,

orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T07:53:24.452Z       INFO    setup   starting manager
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T07:53:24.453Z       INFO    controller-runtime.manager      starting metrics server {"path": "/metrics"}
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T07:53:24.453Z       INFO    controller-runtime.controller   Starting EventSource    {"controller": "applicationgroup", "source": "kind source: /, Kind="}
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T07:53:24.554Z       INFO    controller-runtime.controller   Starting Controller     {"controller": "applicationgroup"}
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T07:53:24.554Z       INFO    controller-runtime.controller   Starting workers        {"controller": "applicationgroup", "worker count": 1}
... truncated for brevity ...
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T08:04:18.875Z       DEBUG   controllers.ApplicationGroup    workflow ran to completion and succeeded        {"appgroup": "bookinfo"}
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T08:04:18.901Z       DEBUG   controller-runtime.controller   Successfully Reconciled {"controller": "applicationgroup", "request": "/bookinfo"}
orkestra-885c5ff4-kh7n9 orkestra 2021-03-23T08:04:18.902Z       DEBUG   controller-runtime.manager.events       Normal  {"object": {"kind":"ApplicationGroup","name":"bookinfo","uid":"52c5095e-0aa1-4067-a434-f1155ebbbdcd","apiVersion":"orkestra.azure.microsoft.com/v1alpha1","resourceVersion":"30145"}, "reason": "ReconcileSuccess", "message": "Successfully reconciled ApplicationGroup bookinfo"}

workflow

Verify that the Application helm release have been successfully deployed

helm ls

NAME            NAMESPACE       REVISION        UPDATED                                 STATUS    CHART            APP VERSION
orkestra        orkestra        1               2021-03-23 08:02:15.0044864 +0000 UTC   deployed  orkestra-0.1.0   0.1.0
ambassador      ambassador      1               2021-03-23 08:02:35.0044864 +0000 UTC   deployed  ambassador-6.6.0 1.12.1     
bookinfo        bookinfo        1               2021-03-23 08:04:08.6088786 +0000 UTC   deployed  bookinfo-v1      0.16.2     
details         bookinfo        1               2021-03-23 08:03:26.1043919 +0000 UTC   deployed  details-v1       1.16.2     
productpage     bookinfo        1               2021-03-23 08:03:47.4150589 +0000 UTC   deployed  productpage-v1   1.16.2     
ratings         bookinfo        1               2021-03-23 08:03:25.9770024 +0000 UTC   deployed  ratings-v1       1.16.2     
reviews         bookinfo        1               2021-03-23 08:03:36.9634599 +0000 UTC   deployed  reviews-v1       1.16.2

Send request to productpage via Ambassador gateway/proxy

kubectl -n default exec curl -- curl -ksS https://ambassador.ambassador:443/bookinfo/ | grep -o "<title>.*</title>"
<title>Simple Bookstore App</title>

2. Bookinfo with Quality Gates

Prequisites

  • A Kubernetes cluster with sufficient resources to run the Keptn Controller

⚠️ Avoid using a cluster with a low number of nodes and low CPU/RAM or a KinD, Minikube or microk8s cluster

In this example we will show how to use Keptn to perform a promotion based on the evaluation of a Quality Gate.

The Application Group deploys an application that relies on multiple supporting layers to be deployed and running before the application can be started. The bookinfo application relies on the following layers:

  • Istio CRDs - This layer (istio-base) contains the Istio CRDs that are used to configure the Istio service mesh.
  • Prometheus - This layer contains the Prometheus component that is used to monitor the application and leveraged by Keptn to perform the evaluation of the Quality Gate.
  • Istio - This layer (*istiod) )contains the Istio service mesh that is used to deploy the application.
  • Istio Ingress Gateway - This layer contains the Istio Ingress Gateway that is used to expose the application.
  • Bookinfo - This layer contains the bookinfo application.

The scenarios that we will use in this example are:

The bookinfo application is deployed with the Istio sidecar injection enabled. The application is configured to use the Keptn executor to perform evaluation of the Quality Gate.

Orkestra Workflow

  1. The productpage sidecar is configured to serve traffic without any issues. We expect the Workflow and subsequently the ApplicationGroup to succeed.

Keptn dashboard - Success

⚠️ monitoring failed is a known, benign issue when submitting the ApplicationGroup multiple times. Authenticate with Keptn Controller for the dashboard:

export KEPTN_ENDPOINT=http://$(kubectl get svc api-gateway-nginx -n orkestra -ojsonpath='{.status.loadBalancer.ingress[0].ip}')/api \
export KEPTN_ENDPOINT=http://$(kubectl get svc api-gateway-nginx -n orkestra -ojsonpath='{.status.loadBalancer.ingress[0].ip}')/api \
keptn auth --endpoint=$KEPTN_ENDPOINT --api-token=$KEPTN_API_TOKEN

Retrieve the dashboard URL, Username and Password:

The IPs and password will differ for each cluster.

keptn configure bridge --output
Your Keptn Bridge is available under: http://20.75.119.32/bridge

These are your credentials
user: keptn
password: UxUqN6XvWMpsrLqp6BeL

Keptn Dashboard

  1. The productpage sidecar is configured to inject faults (return status code 500, 80% of the time) using the VirtualService. We expect the Workflow and subsequently the ApplicationGroup to fail & rollback to the previous ApplicationGroup spec (i.e. Scenario 1).

Keptn dashboard - Failure

Keptn Dashboard

Installation

Orkestra Helm Chart

From the root directory of the repository, run:

helm upgrade --install orkestra chart/orkestra -n orkestra --create-namespace --set=keptn.enabled=true --set=keptn-addons.enabled=true

💡 Note: If prometheus is expected to run in a different namespace the user must specify the namespace in the --set option as follows,

export PROM_NS=prometheus
helm upgrade --install orkestra chart/orkestra -n orkestra --create-namespace --set=keptn.enabled=true --set=keptn-addons.enabled=true --set=keptn-addons.prometheus.namespace=$PROM_NS

Scenario 1 : Successful Reconciliation

The bookinfo application is deployed using the following Kubernetes manifests:

The ConfigMap is used to configure the Keptn executor and contains the following:

  • keptn-config.yaml - This file contains the Keptn configuration.
  • sli.yaml - This file contains the SLI configuration.
  • slo.yaml - This file contains the SLO configuration.
  • config.yaml - This file contains the configuration for the hey load generator.
kubectl create -f examples/keptn/bookinfo.yaml -n orkestra \
kubectl create -f examples/keptn/bookinfo-keptn-cm.yaml -n orkestra

Scenario 2 : Failed Reconciliation leading to Rollback

kubectl apply -f examples/keptn/bookinfo-with-faults.yaml -n orkestra

Cleanup

  1. Delete the bookinfo ApplicationGroup and wait for the reverse workflow to complete
kubectl delete -f examples/keptn/bookinfo.yaml -n orkestra
  1. Once the ApplicationGroup is deleted, delete the Keptn configuration configMap

⚠️ Deleting the Keptn ConfigMap before the ApplicationGroup will cause the reverse Workflow to fail causing cleanup to fail.

kubectl delete -f examples/keptn/bookinfo-keptn-cm.yaml -n orkestra

Manual Testing

Trigger evaluation

keptn create project bookinfo --shipyard=./shipyard.yaml
keptn create service bookinfo --project=bookinfo
keptn configure monitoring prometheus --project=bookinfo --service=bookinfo
keptn add-resource --project=bookinfo --service=bookinfo --resource=slo.yaml --resourceUri=slo.yaml --stage=dev
keptn add-resource --project=bookinfo --service=bookinfo --resource=prometheus/sli.yaml  --resourceUri=prometheus/sli.yaml --stage=dev
keptn add-resource --project=bookinfo --service=bookinfo --resource=job/config.yaml  --resourceUri=job/config.yaml --stage=dev
keptn trigger evaluation --project=bookinfo --service=bookinfo --timeframe=5m --stage dev --start $(date -u +"%Y-%m-%dT%T")