CloudFlow:Create a workflow

Option A: Use CloudFlow Studio

CloudFlow Studio provides a drag-and-drop visual editor for designing workflows.

To build a workflow in CloudFlow Studio:

1. On the CloudFlow Studio tab, select a state node or flow node and drag it to the desired position on the canvas. You can also adjust the positions of nodes in an existing flow chart to modify a business flow chart.

2. Select a node on the canvas to configure it. Each node has four configuration sections:

   Section	Purpose
   Basic Settings	Set the node name and core properties
   Input Settings	Control what data the node receives
   Output Setting	Control what data the node passes to the next state
   Troubleshooting	Define retry and error-handling policies

3. (Optional) On the Configure Workflow tab, assign an execution role to the workflow.

   Note

   An execution role is required when the workflow integrates cloud services (including CloudFlow itself). See Create execution roles.

4. Click Save, then click Exit.

Example: Configure an InvokeFunction task node

The following example shows how to configure an InvokeFunction node. Use default values for any parameters not listed.

Basic Settings

Parameter	Description	Example
State Name	A descriptive name for the node.	InvokeFunction

Invocation Parameters

Specify the function to invoke, its version or alias, and the invocation method. Click Switch to YAML Editing to define these parameters in YAML instead.

Sample YAML:

invocationType: Sync
resourceArn: acs:fc:cn-beijing:1034354682****:functions/GPU-demo/LATEST
body:
  key: value

Parameter	Description	Example
Function	The function to invoke.	myfunction
Version or Alias	The function version or alias.	LATEST
Function Invocation Method	Sync Invocation or Async Invocation. See Function invocation.	Sync Invocation
Request Load	The payload that triggers the function, in YAML or JSON.	See the following example.
Next State	The state the workflow transitions to after this node completes.	Parallel

Sample request payload (JSON):

{
  "key1": "value1",
  "key2.$": "$.myStateInput.key",
  "key3": 100
}

Input Settings

Choose how to filter inputs for the node:

Option	Behavior
Use JSONPath to select some parameters	Enter a JSONPath filter expression to extract specific fields from the input.
Original inputs as node inputs	Pass the full input to the node without filtering.
Use parameter filters to construct parameters	Build a new input object in YAML or JSON.

Sample parameter filter (YAML):

FieldA: 123
FieldB: string
FieldC.$: $Input

Output Settings

Choose how to filter outputs from the node:

Option	Behavior
Use JSONPath to select some parameters	Enter a JSONPath filter expression to extract specific fields from the output.
Pass raw inputs to outputs	Forward the original input as the node output, ignoring the node result.
Use parameter filters to construct parameters	Build a new output object in YAML or JSON.

Troubleshooting

Define retry and error-capture rules for the node. These settings apply only in workflow integration scenarios. For details, see Error handling.

*Retry upon error*

Click Add Retry to add a retry policy.

Parameter	Description	Example
Error Type	One or more error types to match.	FC.BadResponse
Maximum Retries	The maximum number of retry attempts.	5
Retry Interval Seconds	Seconds between retries. Maximum: 86400.	3
Backoff Multiplication Factor	The multiplier applied to the retry interval after each attempt.	1
Maximum Backoff Seconds	The upper bound (in seconds) for exponential backoff intervals.	5

*Capture errors and exceptions*

Click Add Capture Rule to define a fallback when an error occurs.

Parameter	Description	Example
Error Type	One or more error types to match.	FC.EntitvToolLarge
Fallback Status	The state the workflow transitions to when the error is caught.	Hello World
Response	A YAML or JSON response attached to the fallback state input.	{"key": "hello world"}

Option B: Use YAML

Define your workflow by writing a YAML definition directly in the code editor.

1. On the Edit Workflow page, click the YAML tab.

2. Enter a workflow definition in the code editor on the left. The flow chart on the right updates in real time. Sample YAML workflow definition:

      Type: StateMachine
      Name: my-wkfl
      SpecVersion: v1
      StartAt: InvokeFunction
      States:
        - Type: Task
          Name: InvokeFunction
          Action: FC:InvokeFunction
          TaskMode: RequestComplete
          Parameters:
            invocationType: Sync
            resourceArn: >-
              acs:fc:{regionId}:{accountId}:functions/myfunction/LATEST
          Retry:
            - Errors: []
              Description: 'Retry policy #1'
              MaxAttempts: 1
              IntervalSeconds: 1
          Next: Hello World
          Catch:
            - Errors:
                - FC.AccessDenied
              Description: 'Error capture rule #1'
              Next: Hello World
        - Type: Pass
          Name: Hello World
          End: true

   

3. Before saving, replace the placeholder values in resourceArn with your actual function ARN: acs:fc:{regionId}:{accountId}:functions/myfunction/LATEST To get a function ARN, see Obtain the ARN of the function.

   Placeholder	Description	Example
   {regionId}	The region where your function resides.	cn-hangzhou
   {accountId}	Your Alibaba Cloud account ID.	1034354682****
   myfunction	Your function name.	GPU-demo
   LATEST	The function version. Replace with a specific version or alias as needed.	v1

4. (Optional) On the Configure Workflow tab, assign an execution role to the workflow.

   Note

   An execution role is required when the workflow integrates cloud services (including CloudFlow itself). See Create execution roles.

5. Click Save, then click Exit.