epe105 / agile-integrations-workshop Goto Github PK

Gbenga's earlier comments
-This entire lab fails because the schema hasn't been loaded to postgresql. As a result, Fuse ignite can't perform a data mapping from the database to the webhook payload. We need to mechanism to load the locations schema and data when the postgresql pod is deployed using this https://github.com/epe105/dayinthelife-import/blob/master/location-service/src/main/resources/schema.sql
One solution could have the provisioner to utilize 'oc exec' to run pssql on the deployed pod to load the schema and populate the database. With this change I would suggest we change the provisioner to use the persistent postrgresql template so that the data won't be lost if the cluster is stopped then restarted.

I was able to proceed by applying the schema

• When testing the integration, I had some issues with onlinecurl. Postman was better but it could have been something silly i did with onlinecurl
  Update: actually this is a bit worse. The reason why I even had a postgresSQL db was because I was using an env where the dev track had been run before (the dev student deploys a postgres in dev track lab 3). If this was a freshly provisioned environment, I don't think the postgres db would exist as is. For the citizen integrator track we should probably provision a postrgres db and populate as i suggested above

add the API features to the application plans for this API so that the right features are shown on signup page in dev portal rather than the generic out of the box features

change this to be the same instructions as citizen integrator but leave up to presenter to ask them to import the json

We need to decide if we want the attendees to create the GET operation in Fuse Online in LAB 3 and use that in LAB 4 when managing the APO rather than using the provisioned backed URL.

Earlier comments:
step 1, part 14. asks to populate a fuse GET method that doesn't exist
step 3 - makes references to location service that don't exist. I'm assuming this is supposed to be the spring services deployed in the dev track. I think lab 3 should also include building an API via Fuse Online that also includes the GET operation. This operation can be used for this and the subsequent labs.

2/6/2018 Update: The service does exist in the International (backend) namespace but we can still demonstrate how Fuse Online can be used to build out an API quickly
2/6/2018 yet another update: Gary from fuse says that the API Provider currently doesn’t support collections so this could be an issue. Also API PRovider is still in tech preview

Add properties to provision for range of user numbers - start and end. This make it easier to rerun the provisioner in case problems occur without changing the ansible playbooks.

It appears the container images have changed recently for apicurio and microcks. The screenshots of the apps are quite different, particularly in the case of microcks. Would it make sense to have the provisioner point to specific versions of the images so we don't get surprised by updates to the UI?

Update 3scale to 2.4 and update workshop to demonstrate 3scale's ability to find a deployed API (fuse) and automatically "manage" the API

Instance sizing, storage, number of app nodes for X users, etc...

when provisioning a task/role, rather than deploying for one user than waiting to confirm, attempt to deploy for all the users, then wait per user..this way they can run deploy concurrently

it also seems like syndesis has a way to generate an operator. i wonder if this is supported and we can use the operator to perform the install

maybe introduce the cli during lab 3 when che is introduced. Attendees may not have the oc client downloaded locally

We have run into different issues since we started pulling this workshop together. We need to provide an instructor checklist to assist them in giving a successful workshop.

For example, check to see if container images were successfully pulled, run through the lab with 1 user for each track to verify components are functioning properly. This should include items around Provisioning and Testing the environment (provide list of specific things to check, mimic user experience by walking through labs with 1 or 2 users), Other Potential Gotchas.

1. Browser issues.  We have learned that Firefox and or Safari are more reliable and provide a consistent user experience (Chrome and IE might experience some issues with certain URLs and domains especially with the 3scale admin and developer portal).
   

2. Before starting the hands-on portion of the workshop, let the participants know that it’s important to follow the instructions carefully. Missing a step might cause problems later in the lab work due incorrect user details and mappings:
   

• Make sure you are using the correct User ID.
• Make sure the URL is correct and the user ID is part of the path you are trying to access.
• Network performance may vary, so be patient and allow the application\tool to launch, load and enable all its dependencies.
• etc...

Add details about what needs to be configured each of these files under Workshop Configuration section.

delete all resources and pvcs within the namespace before deleting project

..honestly don't know. if we leave this in, then we can close this issue

The POST operation should probably have a request body rather than a response body

Need to add "edit the inventory file for the OCP hostname variables before running the install" to the instructions.

step 5 adds variables that are never used. We can probably remove them

Dependencies:
Ansible
Python (boto, boto3, botocore, lxml)
AWS cli

Include versions and and any related configuration details.

deploy a local nexus repo with the maven artifacts. update the stack to pull from this repo OR maybe create a stack that already has the repos downloaded locally or a factory that does this once the workspace is created

remove all reference to gogs. Just ask attendees to download the postman test json from a github repo