Tackle-196 landscape report

[jonathanvila]

Issue : https://github.com/konveyor/tackle-pathfinder/issues/66

Features covered

• New endpoint "landscape" that will answer with a List of assessmentIds + assessment risk based on the answers
• Only considers assessments Complete

Unit , Integration and End2End Tests cases covered

• Unit Test ( AssessmentSvcTest.java )

  • given_AssessmentWithNoRedAnswersSeveralAmber_When_landscape_Then_ResultIsAMBER
  • given_AssessmentWithNoRedAnswersFewAmber_When_landscape_Then_ResultIsGREEN
  • given_AssessmentWithRedAnswers_When_landscape_Then_ResultIsRED

• Integration Test ( AssessmentsResourceTest.java )

  • given_ApplicationsAssessed_When_LandscapeRequested_Then_ExpectedJSONIsReturned

• End2End Test on minikube with native image ( check_api.sh )

  • Tests 15

Pre steps Minikube:

1. Install minikube : https://minikube.sigs.k8s.io/docs/start/
2. Start minikube : minikube start --kubernetes-version=v1.20.2 --cpus 4 --memory 12000
3. Enable ingress addons : minikube addons enable ingress
4. Create tackle namespace

   kubectl create namespace tackle

5. Deploy tackle ( to have the keycloak instance running ) on your K8s cluster

   $ kubectl apply -f  https://raw.githubusercontent.com/konveyor/tackle/main/kubernetes/kubernetes-tackle.yaml -n tackle

6. Edit the Ingress object to allow direct request to Pathfinder API

   $ kubectl edit ingress tackle -n tackle

   Add this below paths:

    - backend:
         service:
           name: tackle-pathfinder
           port:
             number: 8080 
       path: /pathfinder
       pathType: ImplementationSpecific

7. Build and push

   $ ./mvnw -U -B package -Dquarkus.container-image.push=true -Dquarkus.container-image.group={your quay user} -Dquarkus.container-image.registry=quay.io -Dquarkus.container-image.username={your quay user} -Dquarkus.container-image.password={your quay pwd} -Pnative

8. Change the image in the deployment tackle-pathfinder to be the one you have pushed
9. Rollout to update the image

   $ kubectl rollout restart deployment tackle-pathfinder -n tackle

Test case with Minikube : Do the full API test

$ .github/scripts/check_api.sh

Expect

+++++ API CHECK SUCCESSFUL ++++++

Pre steps non containerised local test:

1. run keycloak

   podman run -it --name keycloak --rm \     
           -e KEYCLOAK_USER=admin -e KEYCLOAK_PASSWORD=admin -e KEYCLOAK_IMPORT=/tmp/keycloak/quarkus-realm.json \
           -e DB_VENDOR=h2 -p 8180:8080 -p 8543:8443 -v ./src/main/resources/keycloak:/tmp/keycloak:Z \
           jboss/keycloak:12.0.2

2. run postgresql

   podman run -it \                          
           --name postgres-pathfinder -e POSTGRES_USER=pathfinder \                                               
           -e POSTGRES_PASSWORD=pathfinder -e POSTGRES_DB=pathfinder_db \                              
           -p 5433:5432 postgres:10.6

3. run application

   ./mvnw quarkus:dev

Test case with local non containerised : Do the full API test

$ .github/scripts/check_api.sh localhost:8085 localhost:8180

Expect

+++++ API CHECK SUCCESSFUL ++++++

[carlosthe19916]

Adding some notes after an internal chat with @jonathanvila :

• The endpoint GET /assessments/landscape will NEVER return a risk whose value is UNKNOWN even if the enum used in the model includes it.
• The "unassesed" application will need to be calculated in the UI rather than the backend: the number of unasseed application is equal to the total number of applications sent to the endpoint minus the length of the response of the current endpoint => totalAppsRequestedInUI - response.length

[carlosthe19916]

@jonathanvila I just want to clarify/summarize all my previous comments:

• About the 4th donut chart "Unassesed": This is something the UI can do with the current data the backend is providing => this PR doesn't need extra action on this topic.
• About risk UNKNOWN: The endpoint GET /assessments/landscape will NEVER return a risk whose value is UNKNOWN even if the enum used in the model includes it => this PR doesn't need extra action on this topic.
• About whether or not the landscape totals should be calculated in the backend or in the UI: The current response of the backend is enough for the UI to create the donut charts. However, I think there is still a minor decision we need to make in this topic. I have 2 options we can choose:

1. Option 1: Currently the backend is returning an array. Can the response return an object containing the total values?. E.g.

   {
   "low": 1,
   "medium": 0,
   "high": 1
   }

2. Option 2: Let's keep the endpoint as it is but change the name of the endpoint from /assessments/landscape to something like /assessments/assessment-risks (the name is up to you) since the endpoint itself is not creating the Landscape but providing data that will be processed/converted to the landscape report. If you ask me, I would go for this option but I'd like to know your opinion.

• About the Rest endpoint GET: As discussed in the internal meetings, changing the verb GET to POST is something we can do in order to reduce the risk of having a big URL calling the endpoint.

[jonathanvila]

@carlosthe19916 :

1. Calculation : I agree with Option 2. Endpoint = /assessments/assessment-risk
2. Verb, changing to POST + body

[carlosthe19916]

Thanks @jonathanvila, this PR looks good to me.