Function Compute:Asynchronous invocation

Use cases

Example use cases for asynchronous invocation include:

• Audio and video processing

  When you use Function Compute for audio and video workloads, you often run time-consuming tasks such as encoding, decoding, and transcoding. By invoking these tasks asynchronously, they run in the background without blocking the front end, improving the user experience. If a project is large, you can also split it into multiple tasks for parallel processing, or convert a video into multiple formats by invoking multiple functions asynchronously. This shortens the overall processing time.

• Data ETL processing

  In an extract, transform, and load (ETL) pipeline, the process of extracting data from a source, transforming it, and loading it into a destination system may involve multiple independent operations, such as database queries, file I/O, or data cleansing. Asynchronous invocation allows these operations to run in parallel, which reduces processing time and improves system performance. For long-running tasks, such as processing large datasets or complex data transformations, asynchronous invocation allows these tasks to run in the background, which improves the user experience by not blocking the front end.

• Developing web applications

  You can use Function Compute with other cloud services to quickly build web applications. When users submit forms, run search queries, or load content-heavy pages, asynchronous invocation can prevent the page from freezing due to long waits for back-end responses. Users can continue to interact with other parts of the page without experiencing noticeable latency. In high-concurrency scenarios, such as when a large number of users access the application simultaneously, asynchronous invocation can distribute the request load and prevent service overload.

Deferred invocation

You can configure Function Compute to trigger an asynchronous invocation after a specified delay.

Add the HTTP request header x-fc-async-delay to your code. The value must be in seconds and within the range of (0, 3600). Function Compute starts a timer when the execution is triggered and invokes the function after the delay specified by x-fc-async-delay.

Retry policy

The asynchronous invocation mechanism provides built-in error handling and retries. If a step fails, the system can reschedule the task without affecting the overall workflow. When an asynchronous function invocation fails, Function Compute automatically retries the invocation.

Retry mechanism

The following table describes the default retry policy for common errors.

Configure the retry policy

Function Compute allows you to customize the maximum number of retries and the maximum message lifetime.

1. Log on to the Function Compute console. In the left-side navigation pane, choose Function Management > Functions.

2. In the top navigation bar, select a region. On the Functions page, click the target function.

3. On the function details page, click the Tasks tab, and then click Task Mode to the right of Modify. In the Task Mode panel, set the following parameters. After you complete the configuration, click Deploy.

   Parameter	Description
   Task Mode	Specifies whether to enable Task Mode for an asynchronous task. For more information, see Asynchronous tasks.
   Maximum Retries	The maximum number of retries for an asynchronous invocation. The value must be an integer from 0 to 8. By default, Function Compute retries a failed asynchronous invocation 3 times. You can adjust the number of retries based on your business requirements.
   Maximum Message Lifetime	The maximum lifetime of a message in an asynchronous invocation queue. The value must be an integer from 1 to 604,800. The default value is 86,400. Unit: seconds. This duration is calculated from the time the asynchronous invocation is triggered. If the message lifetime exceeds the configured value, the message is discarded. Discarded messages are recorded by the Asynchronous Invocation Events monitoring metric in CloudMonitor. For more information about metrics, see Monitoring metrics.

Result callback

When Function Compute receives an asynchronous invocation request, it persists the request and immediately returns a response without waiting for the request to execute. To retain requests that are discarded after all retries fail, or to notify a downstream service of the asynchronous invocation result, configure the result callback feature. After you configure a destination service, Function Compute automatically calls back the corresponding service based on the execution result when the asynchronous invocation is complete.

How it works

The following figure shows the result callback workflow.

Use cases

• Save discarded events for subsequent processing

  When an asynchronous request fails and fails all retries based on the configured policy, Function Compute discards the request. If you have configured a failure destination, Function Compute automatically pushes the context information of the failed request to a messaging service such as ApsaraMQ for RocketMQ for subsequent processing. You can also set another function as the destination service. In this case, Function Compute automatically pushes the context information of the failed request to that function to execute your custom error-handling logic.

• Automatically notify downstream services of execution results

  After a successful execution, if a success destination is configured, the Function Compute system automatically pushes the request's context to the downstream destination service. For example, if you configure Function Compute to automatically decompress a ZIP file uploaded to OSS and want to receive a message notification upon completion, you can configure a result callback destination service for the target function.

Supported destination services

The configuration for asynchronous invocation destinations includes the following:

• Event content for asynchronous invocation destinations

  When Simple Message Queue (formerly MNS), Function Compute, or ApsaraMQ for RocketMQ is the destination for an asynchronous function invocation, the event content has the following format.

  {
    "timestamp": 1660120276975,
    "requestContext": {
      "requestId": "xxx",
      "functionArn": "acs:fc:{regionid}:{accountid}:functions/xxxx",
         "condition": "FunctionResourceExhausted", 
          "approximateInvokeCount": 3
    },
    "requestPayload": "",
    "responseContext": {
      "statusCode": 200,
      "functionError": ""
    },
    "responsePayload": ""
  }

  Table 1. Parameter descriptions

  Parameter	Description
  timestamp	The timestamp of the invocation.
  requestContext	The request context.
  requestContext.requestId	The request ID of the asynchronous invocation.
  requestContext.functionArn	The Alibaba Cloud Resource Name (ARN) of the asynchronously invoked function.
  requestContext.condition	The error code of the invocation.
  requestContext.approximateInvokeCount	The number of times the asynchronous invocation was executed. A value greater than 1 indicates that Function Compute has retried the invocation.
  requestPayload	The original payload of the function request.
  responseContext	The response context.
  responseContext.statusCode	The system status code returned by the function invocation. If the status code is not 200, a system error occurred.
  responseContext.functionError	The error message from the invocation.
  responsePayload	The original payload returned by the function execution.

  When EventBridge is used as the destination for an asynchronous function invocation, the event is in the following format. For more information, see Event Overview.

  {
      "datacontenttype": "application/json",
      "aliyunaccountid": "143xxxx",
      "data": {
          "requestContext": {
              "condition": "",
              "approximateInvokeCount": 1,
              "requestId": "0fcb7f0c-xxxx",
              "functionArn": "acs:fc:{regionid}:{accountid}:functions/xxxx"
          },
          "requestPayload": "",
          "responsePayload": "",
          "responseContext": {
              "functionError": "",
              "statusCode": 200
          },
          "timestamp": 1660120276975
      },
      "subject": "acs:fc:{regionid}:{accountid}:functions/xxxx",
      "source": "acs:fc",
      "type": "fc:AsyncInvoke:succeeded",
      "aliyunpublishtime": "2021-01-03T09:44:31.233Asia/Shanghai",
      "specversion": "1.0",
      "aliyuneventbusname": "xxxxxxx",
      "id": "ecc4865xxxxxx",
      "time": "2021-01-03T01:44:31Z",
      "aliyunregionid": "cn-shanghai-vpc",
      "aliyunpublishaddr": "199.99.xxx.xxx"
  }

• Payload limits

  The maximum payload limits for supported destination services are as follows:

  • Simple Message Queue (formerly MNS): 64 KB

  • Function Compute: subject to the limits described in Function runtime resource limits.

  • EventBridge: 64 KB

  • ApsaraMQ for RocketMQ: 4 MB

• Avoid invocation loops

  When configuring a destination for an asynchronous execution, avoid creating an invocation loop. For example, you configure Function B as the destination for successful invocations of Function A, and Function A as the destination for successful invocations of Function B. In this case, if you asynchronously trigger Function A and it executes successfully, an invocation loop from A to B and then back to A may occur.

FAQ

How to trigger an asynchronous invocation?

You can use one of the following methods to asynchronously invoke a function in Function Compute:

• Log on to the Function Compute console, find the target function, go to the Test tab, select I want to make the call asynchronously., and then click Test Function.

• Call the InvokeFunction API and set the x-fc-invocation-type parameter to Async.

• Use Serverless Devs to configure an asynchronous invocation. Set the invocation-type parameter to Async. For more information, see Invoke a function.

• Create a trigger that supports asynchronous invocation to asynchronously trigger the function. For more information, see Event-driven invocations.

• When you use an HTTP trigger or a custom domain for an invocation, set the X-Fc-Invocation-Type request header to Async, as shown in the following example.

  curl -v -H "X-Fc-Invocation-Type: Async" https://example.cn-shenzhen.fcapp.run/$path

  Note

  • Call a Web Function: For example, if you use Flask and want to test a Python function with a route defined as @app.route('/test'), replace $path with test. When you test a Python function with a route defined as @app.route('/'), call the public endpoint of the HTTP trigger directly.

  • Invoke an Event Function: Directly call the public endpoint of the HTTP trigger.

Next steps

To track the status of an asynchronous request at each stage, enable Task Mode. For more information, see Asynchronous tasks.