Skip to main content
Flow control actions allow you to manage the execution order and concurrency of your workflow. The parallel action enables you to run multiple action branches at the same time, improving performance when tasks are independent.

​
parallel

Executes multiple action branches simultaneously. Each branch runs independently and can have its own chain of subsequent actions.

​
Steps

A list of action steps to execute in parallel. Each step is a reference to another action in the workflow.
YAML Keysteps
Typearray
RequiredYes
Each step follows the format action.action_id where action_id is the identifier of another action in your workflow.

​
Stop On Failure

Whether to stop all parallel branches when any branch fails. When true, remaining branches are cancelled on the first failure.
YAML KeystopOnFailure
Typeboolean
RequiredNo
Defaulttrue
Setting this to false allows all branches to complete even if some fail, useful when you want partial results.

​
Examples

​
Fetch Multiple Data Sources

Retrieve data from multiple tables simultaneously: 
actions:
  fetch_all_data:
    type: parallel
    config:
      steps:
        - action.fetch_users
        - action.fetch_orders
        - action.fetch_products
      stopOnFailure: false
    next: response.combined_data
    fail: response.partial_error

  fetch_users:
    type: fetch
    config:
      integrationID: my_database
      table: users

  fetch_orders:
    type: fetch
    config:
      integrationID: my_database
      table: orders

  fetch_products:
    type: fetch
    config:
      integrationID: my_database
      table: products
After the parallel action completes, you can access results from each branch using their action IDs:
  • {{ .fetch_users }} — Results from the users query
  • {{ .fetch_orders }} — Results from the orders query
  • {{ .fetch_products }} — Results from the products query

​
Parallel API Calls

Call multiple external APIs at the same time: 
actions:
  fetch_external_data:
    type: parallel
    config:
      steps:
        - action.get_weather
        - action.get_news
        - action.get_stocks
      stopOnFailure: true
    next: action.combine_results

  get_weather:
    type: http
    config:
      url: "https://api.weather.com/current"
      method: GET
      headers:
        Authorization: "Bearer {{ secret \"WEATHER_API_KEY\" }}"

  get_news:
    type: http
    config:
      url: "https://api.news.com/headlines"
      method: GET
      headers:
        Authorization: "Bearer {{ secret \"NEWS_API_KEY\" }}"

  get_stocks:
    type: http
    config:
      url: "https://api.stocks.com/quotes"
      method: GET
      headers:
        Authorization: "Bearer {{ secret \"STOCKS_API_KEY\" }}"

  combine_results:
    type: javascript
    config:
      script: |
        function servflowRun(vars) {
          return {
            weather: vars.get_weather,
            news: vars.get_news,
            stocks: vars.get_stocks
          };
        }
    next: response.dashboard

​
Parallel with Chained Actions

Each parallel branch can have its own chain of subsequent actions: 
actions:
  process_all:
    type: parallel
    config:
      steps:
        - action.process_images
        - action.process_documents
    next: response.complete

  process_images:
    type: fetch
    config:
      integrationID: my_database
      table: images
      filters:
        - field: status
          operator: eq
          value: pending
    next: action.resize_images

  resize_images:
    type: javascript
    config:
      script: |
        function servflowRun(vars) {
          // Process images
          return vars.process_images.map(img => ({
            ...img,
            status: 'processed'
          }));
        }

  process_documents:
    type: fetch
    config:
      integrationID: my_database
      table: documents
      filters:
        - field: status
          operator: eq
          value: pending
    next: action.analyze_documents

  analyze_documents:
    type: agent
    config:
      integrationID: my_openai
      systemPrompt: "Analyze and summarize each document."
      userPrompt: "{{ jsonout .process_documents }}"

​
Graceful Degradation

Allow partial failures to continue the workflow: 
actions:
  fetch_optional_data:
    type: parallel
    config:
      steps:
        - action.fetch_required
        - action.fetch_optional_cache
        - action.fetch_optional_analytics
      stopOnFailure: false
    next: action.build_response

  fetch_required:
    type: fetch
    config:
      integrationID: my_database
      table: users
      filters:
        - field: id
          operator: eq
          value: "{{ param \"user_id\" }}"
      single: true

  fetch_optional_cache:
    type: memory_fetch
    config:
      key: "user_preferences_{{ param \"user_id\" }}"

  fetch_optional_analytics:
    type: http
    config:
      url: "https://analytics.example.com/user/{{ param \"user_id\" }}"
      method: GET

  build_response:
    type: javascript
    config:
      script: |
        function servflowRun(vars) {
          return {
            user: vars.fetch_required,
            preferences: vars.fetch_optional_cache || {},
            analytics: vars.fetch_optional_analytics || null
          };
        }
    next: response.success

​
Best Practices

Independent Operations: Only use parallel for actions that don’t depend on each other’s results. If action B needs the result of action A, they must run sequentially.
Error Handling: Set stopOnFailure: false when some branches are optional and you want the workflow to continue even if they fail.
Rate Limits: Be mindful of rate limits when making parallel API calls to the same external service. You may need to implement throttling or use sequential processing instead.

​
Next Steps

Data Operations

Learn about fetching and storing data from databases.

HTTP Requests

Make parallel calls to external APIs.

Transformation

Combine and transform parallel results with JavaScript.

Actions Overview

Learn the fundamentals of actions in ServFlow.