If standard runtimes in Function Compute cannot meet your business requirements, you use Custom Container runtimes to create functions. This topic describes how to create Custom Container functions in the Function Compute console or by using Serverless Devs.
Usage notes
The image that you want to use to create a Custom Container function must be an image from a Container Registry repository that resides in the same region as your function. If you use a Mac computer that has Apple chips or a computer that runs on the Arm architecture, you must specify the Linux/Amd64-based platform to compile images. Sample code: docker build --platform linux/amd64 -t $IMAGE_NAME .
Prerequisites
Container Registry
A Container Registry Enterprise Edition instance is created. For more information, see Create a Container Registry Enterprise Edition instance.
NoteYou can create a Container Registry Personal Edition instance or a Container Registry Enterprise Edition instance. We recommend that you create a Container Registry Enterprise Edition instance.
A namespace is created. For more information, see Use Container Registry Enterprise Edition instances to build images.
An image repository is created. For more information, see the "Step 3: Create an image repository" section of the Use Container Registry Enterprise Edition instances to build images topic.
Serverless Devs
Use the Function Compute console to create a function
Step 1: Create a service and configure permissions
Create a function in the Function Compute console.
For more information, see the "Create a service" section of the Manage service topic.
Attach the AliyunContainerRegistryReadOnlyAccess or AliyunContainerRegistryFullAccess policy to the service.
For more information, see Grant Function Compute permissions to access other Alibaba Cloud services. The preceding policies allow Function Compute to obtain the temporary account for the default instance in Container Registry. Then, Function Compute uses the temporary account to push the image from your private image repository.
Step 2: Create a function
Log on to the Function Compute console. In the left-side navigation pane, click Services & Functions.
In the top navigation bar, select a region. On the Services page, click the desired service.
- On the Functions page, click Create Function.
On the Create Function page, select Use Container Image, configure the following parameters based on your business requirements, and then click Create.
Basic Settings: Configure basic information about the function that you want to create, including the function name, web server mode, and the handler type.
Parameter
Description
Function Name
Enter a name for the function. If you leave this parameter empty, Function Compute automatically creates a name for your function.
Web Server Mode
Specifies whether to enable the web server mode. Valid values:
Yes: enables the web server mode. You need to implement a web server in your container image to listen on ports and process requests.
No: disables the web server mode. Then, you can set Handler Type only to Event Handler. You must actively exit processes after requests are processed and set
ExitCode
to 0. In addition, the requested event parameters are passed to the function in the form of environment variables.
Handler Type
Specify the type of requests that are processed by the function. The requests can be event requests or HTTP requests. Valid values:
Event Handler: triggers the execution of the function by using a timer, calling API operations, using SDKs, or using triggers that are integrated with other Alibaba Cloud services.
HTTP Handler: triggers function execution by sending an HTTP request or creating a WebSocket connection. For a web scenario, we recommend that you select Use Custom Runtime when you create the function.
Image Configurations: Configure an image for the function.
Parameter
Description
Image Selection Mode
Select Use Sample Images or select your own image to create the function.
Use Sample Image: Select a built-in sample image of Function Compute.
Use Container Registry Images: Click Select a Container Registry image below Container Image. In the Select Container Image panel, configure Container Image Instance and Container Registry Repository. Find an image among the available images and click Select in the Actions column.
NoteFor Container Registry Enterprise Edition, Function Compute uses the tag of the image you selected to create function instances.
If you use an image of Container Registry Enterprise Edition (Advanced Edition or Standard Edition), we recommend that you turn on Immutable for Image Version. Otherwise, the image tag may be updated elsewhere and Function Compute pulls the updated image data to start function instances. For more information, see Configure a repository to be immutable.
If you are using an image of Container Registry Enterprise Edition (Basic Edition) or Container Registry Personal Edition, you must update the function with the image information in the Function Compute console after you update the image tag with the same name in the image repository. Then, Function Compute starts function instances with the new image.
Startup Command
Specify the startup command of the container. If you do not configure this parameter, the Entrypoint or CMD on the image is used.
Listening Port
Specify the port on which the HTTP server in the container image listens. The default port is 9000. If you do not specify this parameter, the container image runs in non-interactive mode.
Advanced Settings: Configure the instance-related information, execution timeout period, and handler of the function.
Parameter
Description
GPU Acceleration
Specifies whether to use GPU-accelerated instances based on your business requirements. By default, elastic instances are used and GPU-accelerated instances are not used. For more information, see Instance types and instance modes. For information about the billing of each instance type, see Billing overview. Valid values:
Enable GPU
Disable GPU
Specifications
Enable GPU:
Select a value from the GPU Type drop-down list and select appropriate GPU Memory Size based on your business requirements. Function Compute automatically configures vCPU Capacity and Memory Capacity based on the GPU memory that you selected.
Disable GPU:
Enter or select vCPUs or memory within valid scopes.
For more information about the billing of resources, see Billing overview.
NoteThe ratio of vCPU to memory capacity (in GB) is 1: N. N must be a value that ranges from 1 to 4.
Size of Temporary Disk
Select the disk size based on your business requirements. Function Compute provides you with a free disk capacity of up to 512 MB. For more information, see Billing overview.
Instance Concurrency
If you set Web Server Mode to No, this parameter is not required. Specify the concurrency to process requests for each instance of the function. For more information, see Configure instance concurrency.
Execution Timeout Period
Configure the timeout period for the execution of the function. The default value is 60 seconds. The maximum value is 86,400 seconds. We recommend that you set this parameter to 600 seconds.
NoteIf the execution time of the function exceeds the specified timeout period, the function fails to be executed. If you want to apply a longer timeout period, join the DingTalk group (ID: 11721331) to contact the technical support.
Time zone
Select the time zone of the function. After you select the time zone of the function, the environment variable TZ is automatically added to the function. The value of the variable is the time zone that you select.
Environment Variables: Configure the environment variables in the runtime of the function. For more information, see Environment variables.
Trigger Configurations: Configure a trigger for execution of the function. For more information, see Manage triggers.
After the function is created, you can view and update the function in the function list of the service.
When you update a function, you can only change the specified listening port. You cannot delete or add extra listening ports. If a listening port is configured when you create the function, the listening port is retained when you update the function if you do not specify the listening port.
Use Serverless Devs to create a function
You can use Serverless Devs to build and push container images and deploy functions with a few clicks.
Run the following command to initialize the project: You must enter the address of the image repository that you created in Container Registry as prompted.
sudo s init start-fc-custom-container-event-nodejs14
Sample command output:
Serverless Awesome: https://github.com/Serverless-Devs/package-awesome Please input your project name (init dir) start-fc-custom-container-event-nodejs14 file decompression completed ____ _ _ ___ _ _ _ _____ ____ / _ \/ \ / \\ \/// \ /\/ \ /| / // _\ | / \|| | | | \ / | | ||| |\ || | __\| / | |-||| |_/\| | / / | \_/|| | \|| | | | \__ \_/ \|\____/\_//_/ \____/\_/ \| \_/ \____/ please select credential alias default Welcome to the Aliyun FC start application This application requires to open these services: FC : https://fc.console.aliyun.com/ ACR: https://cr.console.aliyun.com/ * Note: Actions are declared in the s.yaml file: Run the sudo s build --use-docker --dockerfile ./code/Dockerfile command before the project is deployed. If you do not need to build the project, you can use annotations to skip the preceding command. > Note: Before the project deployment, replace the value of the image parameter in the s.yaml file with the address of the image repository in Container Registry. * The project is initialized. You can go to the project directory and run the s deploy command to deploy the project. Thanks for using Serverless-Devs You could [cd /test/test1/start-fc-custom-container-event-nodejs14] and enjoy your serverless journey! If you need help for this example, you can use [s -h] after you enter folder. Document Star: https://github.com/Serverless-Devs/Serverless-Devs Do you want to deploy the project immediately? No
Run the following command to go to the project directory:
cd start-fc-custom-container-event-nodejs14
Edit the
s.yaml
file. For more information about the parameters in the .yaml file, see the "YAML syntax and permission management" section of the Serverless Devs commands topic.Run the following command to deploy the project:
sudo s deploy
Sample command output:
[2021-12-15 07:54:30] [INFO] [S-CLI] - Start ... [2021-12-15 07:54:30] [INFO] [S-CLI] - Start the pre-action [2021-12-15 07:54:30] [INFO] [S-CLI] - Action: s build --use-docker --dockerfile ./code/Dockerfile [2021-12-15 07:54:31] [INFO] [S-CLI] - Start ... [2021-12-15 07:54:32] [INFO] [FC-BUILD] - Build artifact start... [2021-12-15 07:54:32] [INFO] [FC-BUILD] - Use docker for building. [2021-12-15 07:54:32] [INFO] [FC-BUILD] - Building image... Sending build context to Docker daemon 5.12kB Step 1/7 : FROM node:14.5.0-alpine3.11 ---> 5d97b3d11dc1 ...... Step 7/7 : ENTRYPOINT [ "node", "server.js" ] ---> Using cache ---> a5ef1c015e7e Successfully built a5ef1c015e7e Successfully tagged registry.cn-hangzhou.aliyuncs.com/fc-example/test:nginx SECURITY WARNING: You are building a Docker image from Windows against a non-Windows Docker host. All files and directories added to build context will have '-rwxr-xr-x' permissions. It is recommended to double check and reset permissions for sensitive files and directories. Build image(registry.cn-hangzhou.aliyuncs.com/fc-example/test:nginx) successfully [2021-12-15 07:54:33] [INFO] [FC-BUILD] - Build artifact successfully. Tips for next step ====================== * Invoke Event Function: s local invoke * Invoke Http Function: s local start * Deploy Resources: s deploy End of method: build [2021-12-15 07:54:33] [INFO] [S-CLI] - End the pre-action [2021-12-15 07:54:34] [INFO] [FC-DEPLOY] - Using region: cn-hangzhou [2021-12-15 07:54:34] [INFO] [FC-DEPLOY] - Using access alias: default [2021-12-15 07:54:34] [INFO] [FC-DEPLOY] - Using accessKeyID: LTAI4G4cwJkK4Rza6xd9**** [2021-12-15 07:54:34] [INFO] [FC-DEPLOY] - Using accessKeySecret: eCc0GxSpzfq1DVspnqqd6nmYNN**** [2021-12-15 07:54:34] [INFO] [FC-DEPLOY] - Checking Service hello-world-service exists [2021-12-15 07:54:35] [INFO] [FC-DEPLOY] - Setting role: AliyunFCDefaultRole [2021-12-15 07:54:35] [INFO] [RAM] - Checking Role AliyunFCDefaultRole exists [2021-12-15 07:54:35] [INFO] [RAM] - Updating role: AliyunFCDefaultRole [2021-12-15 07:54:35] [INFO] [RAM] - Checking Plicy AliyunFCDefaultRolePolicy exists [2021-12-15 07:54:35] [INFO] [FC-DEPLOY] - Checking Function nodejs14-event-function exists [2021-12-15 07:54:36] [INFO] [FC-DEPLOY] - Using image registry: registry.cn-hangzhou.aliyuncs.com [2021-12-15 07:54:36] [INFO] [FC-DEPLOY] - Try to use a temporary token for docker login Login to registry: registry.cn-hangzhou.aliyuncs.com with user: cr_temp_user Pushing docker image: registry.cn-hangzhou.aliyuncs.com/fc-example/test:nginx... The push refers to repository [registry.cn-hangzhou.aliyuncs.com/fc-example/test] cdf38e7753b7: Layer already exists 43128f71725b: Layer already exists 0fb36a16ab83: Layer already exists dd966b9fd474: Layer already exists a1915d7a1111: Layer already exists c4491b3ee709: Layer already exists 9fb10d900487: Layer already exists 3e207b409db3: Layer already exists nginx: digest: sha256:02b69157def85ceb72f32cb1c5845d00e1d8df19caf6eaf720a9bc77bb57db76 size: 1991 √ Make service hello-world-service success. √ Make function hello-world-service/nodejs14-event-function success. [2021-12-15 07:54:39] [INFO] [FC-DEPLOY] - Checking Service hello-world-service exists [2021-12-15 07:54:39] [INFO] [FC-DEPLOY] - Checking Function nodejs14-event-function exists There is auto config in the service: hello-world-service Tips for next step ====================== * Display information of the deployed resource: s info * Display metrics: s metrics * Display logs: s logs * Invoke remote function: s invoke * Remove Service: s remove service * Remove Function: s remove function * Remove Trigger: s remove trigger * Remove CustomDomain: s remove domain helloworld: region: cn-hangzhou service: name: hello-world-service function: name: nodejs14-event-function runtime: custom-container handler: not-used memorySize: 256 timeout: 60
Run the following command to debug the function:
sudo s invoke -e "{\"key\":\"val\"}"
Sample command output:
[2021-12-15 08:00:17] [INFO] [S-CLI] - Start ... ========= FC invoke Logs begin ========= FC Invoke Start RequestId: 768945c8-f92d-428e-89c2-ecd50883**** {"key":"val"} FC Invoke End RequestId: 768945c8-f92d-428e-89c2-ecd50883**** Duration: 3.05 ms, Billed Duration: 4 ms, Memory Size: 256 MB, Max Memory Used: 10.77 MB ========= FC invoke Logs end ========= FC Invoke Result: OK End of method: invoke
References
If you want to use GPU-accelerated instances, you must create Custom Container functions. For more information about how to configure specifications for GPU-accelerated instances, see Manage functions.
When you use Custom Container functions, container images rely on the execution environment and require more time for download and decompression. For more information, see Best practices for cold start optimization.
You can also call the Function Compute API to create a function. For more information, see CreateFunction.
For information about the common scenarios and differences among built-in runtimes, custom runtimes, and Customer Container runtimes, see Function runtime selection.