Exercise: Hello World API

Build a minimal saga on Train RPC that returns {message: "Hello World"}.

This exercise creates a tiny saga and exposes it as an API endpoint. You will run it through the Train RPC runner. The flow uses only three steps: Start → Transform → Success.

Before you start

• You can access the Devops app and the Saga screen.

• The training deployment (including train_rpc) is installed and running.

• You know your API base URL: https://[YOUR_ADMIN_API_DOMAIN].

What you’ll build

• Saga path: /HelloWorld

• Runner: train_rpc (via Allowed For: Train RPC)

• Response: HTTP 200 with body {message: "Hello World"}

1

Open the Saga screen

Open the Saga screen from the Devops app.

Unless you changed routing, the UI is at https://[YOUR_ADMIN_UI_DOMAIN]/app/devops/common/saga.

Saga Screen

2

Create a new saga

Click CREATE NEW in the left menu area. This clears the editor and opens a blank saga.

You’ll define the saga metadata first. Then you’ll design the step graph.

Create New Button

3

Assign an ID

Click the pencil next to ID: and set a unique identifier.

For this exercise, use: hello_world-0001.

Saga ID Assignment

ID conventions (optional)

IDs can be auto-generated if an ID generator is configured. For core assets like sagas and queries, human-readable IDs make debugging and audits much easier.

If you assign IDs manually, prefer alphanumerics plus - and _. For high-volume assets, consider a 4‑digit partition suffix (like -0001). For low-volume assets, you can skip partitioning.

4

Define the saga (name, path, allowed runner)

Click the Definition icon (circled edit icon) to open saga metadata.

Definition Button

Fill in these fields:

Hello World Saga

• Saga Name: Hello World

• Saga Domain (optional): util

• Status: ACTIVE

• Saga Description (optional): Returns "Hello World"

• Saga Path: /HelloWorld

• Version: 0

• Allowed For: Train RPC

• Auto Fail: true

This means requests to /HelloWorld routed through train_rpc will execute this saga.

Allowed For limits which runner can execute the saga. Leaving it empty allows any runner to execute it.

5

Add steps (Start → Transform → Success)

Drag a Start step from the stencil.

Saga Stencil

Drag a Transform step and a Success step.

Your canvas should look like this:

Saga Steps

Connect Start → Transform.

Saga Link

Connect Transform → Success.

Your graph should look like this:

Saga Links

6

Configure the Transform step

The flow runs now, but it does not yet return "Hello World". You’ll add a static value to the payload.

Select the Transform step. Click its pencil icon to edit.

Transform Step Icons

Set these values:

Transform Definition Screen

• Step Name: Say Hello World

• Step Description (optional): This step adds a static "Hello World" message.

• Transform Class: Add Value to Payload

• Transform Parameters:

  • Key: value

  • Value: {message: "Hello World"}

Click Apply to save the step configuration.

Why does Transform have a “Class”?

Instead of adding a separate stencil shape for every transformation type, the step stays the same. You switch behavior by selecting a different Class (or Action) in the step config.

This makes it safer to evolve flows over time. You can change behavior without rebuilding the graph and breaking tracing continuity.

The step label should update on the canvas:

Defined Transform Step

7

Save the saga

Click SAVE in the top-right corner.

Save Button

Wait for the runner to pick up changes. In most training setups this is quick. If your environment has a reload interval, give it ~10–30 seconds.

You should see a confirmation notification.

8

Test the endpoint

Call the API endpoint from your REST client:

https://[YOUR_ADMIN_API_DOMAIN]/api/request/train_rpc/HelloWorld

Expected result:

• HTTP 200 OK

• Body: {message: "Hello World"}

Remember the saga path is case-sensitive. Use /HelloWorld exactly as configured.

Troubleshooting

• 404 Not Found: confirm Saga Path and that you saved the saga. Also confirm it’s allowed for Train RPC.

• 403/401: gateway auth is blocking the request. Check your token/session.

• Saga does not trigger: confirm the request is routed through train_rpc (URL contains /train_rpc/).