Add hostname to test scripts

[alaypatel07]

Describe what this PR does and why we need it:

Add hostname variable to test scripts and change the url

Changes proposed in this pull request

• Add hostname variable in all test scripts and take it as the last command line argument
• Change the url to reflect new changes

[eriknelson]

Nice job Alay!

[djzager]

These look like reasonable changes to me. The only concern I have is that these scripts won't work for a developer deployed broker (make deploy). Not sure how to make this more resilient.

[djzager]

This will only work for oc cluster up --enable=automation-service-broker, it won't work in the development use case (ie. after make deploy).

[eriknelson]

I'm not sure building out these developer focused scripts dynamically for all our deployment scenarios is in scope. Although to your point, make deploy is probably a developer function, so maybe it deserves attention.

[jmrodri]

If they will not work for a developer then -1 to this change. These scripts are PURELY for hooking up to a developers deployment of the broker.

[coveralls]

Coverage remained the same at 33.435% when pulling 896f38c on alaypatel07:master into 1c5cfb3 on openshift:master.

[jmrodri]

I will test these out to see if they suit the developer workflow or now.

[jmrodri]

These scripts are primarily used by a developer so the default case should work against a make deploy installation.

Developer setup

• Remove automation-service-broker from catasb
  • In ansible/roles/openshift_setup/tasks/main.yml of catasb remove ,automation-service-broker from the --enable= line.

  --enable=service-catalog,template-service-broker,router,registry,web-console,persistent-volumes,sample-templates,rhel-imagestreams

• Deploy the cluster using catasb

  • cd local/linux
  • ./reset_environment.sh

• In the broker repo, deploy it

  • cd $GOPATH/src/github.com/openshift/ansible-service-broker
  • make deploy

I should then be able to run test/catalog.sh against this deployed instance without any special parameters or flags. So I think think the url in the scripts should default to the one used by make deploy.

routes after make deploy

NAME                HOST/PORT                                               PATH      SERVICES            PORT        TERMINATION   WILDCARD
broker              broker-automation-broker.172.17.0.1.nip.io                        automation-broker   port-1338   reencrypt     None
broker-redirector   broker-redirector-automation-broker.172.17.0.1.nip.io             automation-broker   port-1337                 None

The endpoint is /automation-broker just before /v2. For example, the catalog call would look like this: https://broker-automation-broker.172.17.0.1.nip.io/automation-broker/v2/catalog

Catasb "production" setup

When a broker is launched using pure catasb, the url is completely different. I would be okay if we want the scripts to also work with that setup. In that case I would expect pass in a flag to the script to indicate it is a production or catasb setup, i.e. it is not a development setup.

For example, ./test/catalog.sh --prod or something like that. Then in the script if it sees the flag it uses the other route.

routes after catasb

NAME      HOST/PORT                                                   PATH      SERVICES                              PORT        TERMINATION   WILDCARD
asb       asb-openshift-automation-service-broker.172.17.0.1.nip.io             openshift-automation-service-broker   port-1338   reencrypt     None

The endpoint is /openshift-automation-service-broker/ just before /v2. For example, the catalog call would look like this: https://asb-openshift-automation-service-broker.172.17.0.1.nip.io/openshift-automation-service-broker/v2/catalog

The working with catasb "production" setup is OPTIONAL in my opinion.

[jmrodri]

See lengthy comment
#985 (comment)

[jmrodri]

Random testing looked good. Visual inspection looks good too.

Last operation passed

$ ./last_operation.sh 687932b6-7413-11e8-9c21-0242ac11000a c7e5f58b-d689-4c45-ba39-0d49de81a487 7f4a5e35e4af2beb70076e72fab0b7ff 1dda1477cace09730bd8ed7a6505607e
{
  "state": "succeeded",
  "description": "provision job completed"
}

Get instance passed

$ ./getinstance.sh 687932b6-7413-11e8-9c21-0242ac11000a
{
  "service_id": "687932b6-7413-11e8-9c21-0242ac11000a",
  "plan_id": "dev",
  "parameters": {
    "_apb_last_requesting_user": "admin",
    "_apb_plan_id": "dev",
    "_apb_service_class_id": "1dda1477cace09730bd8ed7a6505607e",
    "_apb_service_instance_id": "687932b6-7413-11e8-9c21-0242ac11000a",
    "postgresql_database": "admin",
    "postgresql_password": "password",
    "postgresql_user": "admin",
    "postgresql_version": "9.6"
  }
}

catalog passed

$ ./catalog.sh
{
  "services": [
    {
      "name": "dh-pyzip-demo-apb",
      "id": "0e991006d21029e47abe71acc255e807",
      "description": "Python Zip Demo APB Implementation",
      "bindable": false,
      "metadata": {
        "dependencies": [
          "docker.io/ansibleplaybookbundle/py-zip-demo:latest"
        ],
        "displayName": "Pyzip Demo (APB)",
        "documentationUrl": "https://github.com/fusor/apb-examples/tree/master/pyzip-demo-apb",
        "longDescription": "APB Implementation of the Python webapp, https://github.com/fusor/py-zip-demo",
        "providerDisplayName": "Red Hat, Inc."
      },
      "instances_retrievable": true,
      "bindings_retrievable": false,
...

bootstrap passed

$ ./bootstrap.sh 
{
  "spec_count": 42,
  "image_count": 112
}

[jmrodri]

@djzager I've verified that Alay made the changes you pointed out, they now work with make deploy which is the primary goal for this. I'm going to merge this now.