Docs

Volcano Engine Function as a Service

Advanced settings

Function configuration

veFaaS allows you to configure multiple advanced attributes for each function to meet different business requirements. After a function is created, you can update its configuration items as needed on the "Function settings" tab. This topic describes each configuration item and its valid values.

warning

Only the configuration of the latest version of a function can be updated.

After you update the configurations of a function, you need to release the function again to make the updated configurations take effect.

Parameter	Configuration description	Supported function types	Related document
CPU allocation mode	CPU is provisioned only upon request: This mode only applies to event functions, web app functions, and task functions. Instances are automatically allocated and released based on requests, and can be scaled down to zero. Resource usage and function invocations are only billed for the request processing duration. Always allocate CPU: This mode applies to microservices application functions. After applications start, they cannot be scaled down to zero, and you are charged for the container lifecycle.	All types	None
Compute mode	The GPU instance and the CPU instance are independent of each other. After a function is created, switching between the instances is not supported. CPU instance: Traditional functions use CPU instances, which support basic computing scenarios. GPU acceleration: [Beta - apply for trial] This mode is good at executing highly-threaded tasks that are processed in parallel (large-scale computing tasks). It applies to scenarios such as AI model inference, AI model training, accelerated audio and video production, and graphics and image acceleration.	All types	Serverless GPU
Instance specification	You can specify this parameter only when you set the Compute mode parameter of the function to CPU instance. Select appropriate specifications of a single CPU instance, including the vCPU and memory, based on the on-screen prompts and your business requirements.	All types	None
GPU model	[Beta - apply for trial] You can specify this parameter only if you set the Compute mode parameter of the function to GPU acceleration. Select a GPU specification for a single GPU instance from the GPU types provided by veFaaS.	All types	Serverless GPU
Asynchronous tasks	You can enable the asynchronous task feature. warning After a function is created, switching between synchronous and asynchronous modes is not supported. Asynchronous task: Enable: The function responds to events in asynchronous execution mode. In this mode, event invocations do not need to block and wait for processing results. The maximum function execution duration is 3 hours. After you enable the asynchronous task feature, you can specify the Retries for failed async task parameter. When an asynchronous task fails to be executed, the task will be retried within the limit. Disable: The function responds to events in synchronous execution mode. In this mode, event invocations synchronously wait for the function to execute and return results. The maximum function execution duration is 15 minutes. Retries for failed async task: After you enable the asynchronous task feature, you can specify this parameter, which indicates the number of retries after an asynchronous task fails to be executed. You can set the value to 0 to 3. The default value is 0, which indicates that an asynchronous task will not be retried after failure.	Event function Web app function Task function	Asynchronous tasks
Execution timeout	Specify the maximum execution duration for a single request. If the request exceeds the limit, an error is returned. Valid values for the execution timeout for a synchronous invocation: 1 to 3600 seconds. Valid values for the execution timeout for an asynchronous task: 10 to 10800 seconds.	Event function Web app function Task function	None
Single-instance multi-concurrency	Select whether to enable a single instance to process multiple concurrent requests. Enable: Each instance can handle multiple concurrent requests, and memory usage is calculated only once, which significantly reduces resource costs. After you enable this feature, you need to specify the maximum number of concurrent requests that a single function instance can process. For synchronous invocations, an instance can process 10 to 1,000 concurrent requests. For asynchronous tasks, an instance can process 2 to 10 concurrent requests. Disable: Each instance can process only one request at a time.	Event function Web app function Task function	None
VPC	After enabling VPC, configure the corresponding VPC, subnets, and security groups. warning To modify the VPC or subnet of a function, you must first disable or delete all MQ triggers. VPC: Select a VPC to be accessed. Subnet: Select a subnet to be accessed. You can select multiple subnets, but you can select only one subnet in each availability zone. We recommend that you select subnets across multiple availability zones. The function instance will be deployed in the selected availability zones to ensure high availability of your business. Security group: Select a security group to which the function will be bound. A security group is a logical group that provides access policies for NICs with the same security protection requirements and mutual trust within the same VPC.	All types	None
Default network interface card to access the public network	The function uses the platform-provided public NAT and EIP to access the Internet by default. Enabled: The function can access resources in the VPC. The function can access the Internet via the default network interface. veFaaS will charge for outbound Internet traffic. Disabled: After enabling VPC, you can disable the Default network interface card to access the public network feature. The function can access resources in the VPC but cannot access the Internet via the default network interface. To access the Internet, configure a NAT gateway in the VPC. For detailed steps, see Creating public NAT gateways. veFaaS does not charge for outbound Internet traffic in this case.	Event function Web app function Task function	None
Storage	Configure remote storage for the function. After configuration, the function can access remote storage as it accesses the local system, which achieves persistent data storage and sharing across multiple functions.	All types	Mounting FileNAS Mounting TOS
Logging	Specify whether to enable the logging feature. After you enable the logging feature, the stdout and stderr logs of the function will be collected into the specified log project and log topic in Torch Log Service (TLS). You can view the execution logs of the function to check its running status and to troubleshoot. Default log project: If you select Default log project, veFaaS will automatically create the default log project and topic tls-vefaas/tls-vefaas-logstore within your account. Logs will be delivered to the log project and topic. Custom log project: If you select Custom log project, you can specify a custom log project. Logs will be delivered to the target log project and topic.	All types	Logs management
Application monitoring	Specify whether to deliver the monitoring data of the function to Application Performance Management Plus (APMPlus) of Volcano Engine. After you activate APMPlus, the monitoring data of the function will be delivered to APMPlus. The service provides integrated monitoring services for metrics, traces, and logs, draws monitoring dashboards, and monitors high-code applications of ModelArk and intelligent agents such as LangChain. For the fees incurred by the APMPlus monitoring service, see Monitoring billing overview. tip If you have not activated APMPlus, click Activate Now based on the prompts in the console to activate the APMPlus service and authorize veFaaS to access APMPlus. After you enable the application monitoring feature for the function, you can view the monitoring information of the function on the App Monitoring tab of the function details page. For details about how to query information, see Trace analysis.	Event function Web app function	Enabling server-side monitoring
Environment variables	Configure custom environment variables, which can be directly read during the runtime of the function code. Environment variables enhance code portability by separating configurations from code. You can configure environment variables by using the following methods: Edit with form: Configure environment variables by using key-value pairs. Click Add environment variable. Enter a key and a value based on the prompts to add an environment variable. Edit in JSON: Configure environment variables by using a JSON file. The JSON file needs to be in the following format: { "envKey1":"envValue1", "envKey2":"envValue2" } tip Enter key-value pairs in a JSON file as required in the console when you select Edit with form. You can click Reset in the JSON editor to revert the content back to the last parsed JSON data. You can click Formatting in the JSON editor to convert the content to standardized JSON format, remove duplicate key-value pairs, and preserve only the key-value pair that occurred last time.	All types	None
Health check	The microservice function supports the health check feature, which takes effect after the function is configured and released. Check method: Select a method of performing health checks. The following methods are supported. HTTP request check: Send an HTTP request to an instance for health checks. If the HTTP status code 200 is returned, it indicates that the instance is healthy. TCP port check: Check the health status of an instance by establishing a connection between the instance and a TCP socket. If a connection can be established, it indicates that the instance is healthy. HTTP request check: This parameter is available after you set the Check method parameter to HTTP request check. Enter the request path, port, and request header information. TCP port check: This parameter is available after you set the Check method parameter to TCP port check. Enter the TCP port. Delay: The time period after which the first health check is performed after the function starts. Timeout: The maximum wait time for a single health check request. If the function does not start within the timeout period, it is deemed to fail the health check. Check interval: The interval at which health checks are performed. Healthy threshold: The minimum number of consecutive successful checks. Unhealthy threshold: The minimum number of consecutive check failures.	Microservices app function	None
Project	Volcano Engine allows you to manage resources by project. After functions are added to projects, you can manage the functions by project in a centralized manner, which helps maintain resource independence and ensures data security. If you do not specify a specific project, the function will be added to the default project.	All types	Project management
Label	Consisting of a key-value pair, a label is used to identify resources, which facilitates searching and aggregating resources. Click Add label, and add appropriate labels to the function based on the prompts.	All types	Tag management

Last updated: 2026.09.11 16:39:48