Hello World Codelab 3

We've made it to Part 3! Thanks for keeping at it! What we've covered so far:

Hello World Part 1 We created the Hello World Workflow.

Hello World Part 2 We created V2 of Hello World (learning about versioning) and added a HTTP Task to query information about the user's IP address.

Part 3

In Hello World Part 3, we'll introduce the Fork and Join tasks to break our workflow into parallel tracks that run asynchronously, and then combine back into a single workflow.

Where we stand

At the end of Part 2, our workflow appears as:

Now, these two tasks are very simple, and do not take long to run, but what if each of these workflows took several seconds to complete? The overall workflow processing time would take the sum of their execution times to complete.

Neither of these tasks are dependant on one another, and can run independently. In this section, we'll introduce the Fork & Join tasks that allows us to run independent tasks in parallel.

Fork

The Fork ann Join tasks run on the Conductor server, and thus do not require a special task definition (or any unique identifier).

Each 'tine' of the fork runs independently and concurrently to the other 'tines'. Each parallel set of tasks is defined as an array attribute inside the Fork task.

Since hello_world and get_IP are independent, we can place them in separate parallel forks in version 3 of our workflow.

Join

The JOIN task tells the workflow that when the independent paths are completed, the workflow can continue on. In this case, we are waiting for get_IP and hello_world_ref. These two tasks are added to the joinOn parameter - telling the Join that once they are completed, the workflow can move ahead.

Updating the workflow

Changes made:

1. Version set to 3.
2. Fork Task added, and the existing hello_world_<uniqueId> and get_IP tasks are placed into arrays.
3. Join task is added, and the joinOn attributes set.

{
  "name": "hello_world_<uniqueId>",
  "description": "hello world Workflow",
  "version": 3,
  "tasks": [
    {"name":"hello_world_fork",
     "taskReferenceName":"hw_fork_ref",
     "type":"FORK_JOIN",
     "forkTasks":[
         [
           {
                "name": "hello_world_<uniqueid>",
                "taskReferenceName": "hello_world_ref",
                "inputParameters": {},
                "type": "SIMPLE",
                "decisionCases": {},
                "defaultCase": [],
                "forkTasks": [],
                "startDelay": 0,
                "joinOn": [],
                "optional": false,
                "defaultExclusiveJoinTask": [],
                "asyncComplete": false,
                "loopOver": []
            }     

         ],[
            {
                "name": "Get_IP",
                "taskReferenceName": "get_IP",
                "inputParameters": {
                    "http_request": {
                    "uri": "http://ip-api.com/json/${workflow.input.ipaddress}?fields=status,message,country,countryCode,region,regionName,city,zip,lat,lon,timezone,offset,isp,org,as,query",
                    "method": "GET"
                    }
                },
            "type": "HTTP"
            }
         ]
     ]},
     {
        "name": "hello_world_join",
        "taskReferenceName": "hw_join_ref",
        "type": "JOIN",
        "joinOn": [
        "get_IP",
        "hello_world_ref"
        ]

     }

  ],
  "outputParameters": {

    "hw_response": "${hello_world_ref.output.hw_response}",
    "hw_location": "We hope the weather is nice near ${get_IP.output.response.body.city}"

  },
  "schemaVersion": 2,
  "restartable": true,
  "workflowStatusListenerEnabled": true,
  "ownerEmail": "devrel@orkes.io",
  "timeoutPolicy": "ALERT_ONLY",
  "timeoutSeconds": 0,
  "variables": {},
  "inputTemplate": {}
}

When this version of the workflow is submitted, we have a new diagram showing the power of the FORK task:

Running Version 3

We can now run the workflow version 3 with similar input. Since we didn't change the output, the response should be the same.

We'll leave running the workflow to the user to complete (but it is identical to part 2 if any issues arise).

Next Steps

We've completed part 3 of the codelab.

In Part 1, we created a workflow using the Netflix Conductor in the Orkes Playground

In Part 2, we extended the workflow using versioning, and added a HTTP Task.

In Part 3, we created parallel workflows using the FORK task.

In Part 4, we'll extend one of the forks and add an Inline Task. Ready to go? On to Part 4!