huawei-cloud-functiongraph-function-create

FunctionGraph Function Creation & Invocation Skill

Overview

This skill manages FunctionGraph functions on Huawei Cloud, supporting creation with advanced configuration (VPC, environment variables) and function invocation (sync/async).

Architecture: Uses Huawei Cloud FunctionGraph Python SDK to interact with FunctionGraph service.

Applicable Scenarios:

• Quickly create cloud functions without manual console login
• Batch creation of multiple functions with VPC or environment variable configuration
• Foundation component for workflow deployment
• CI/CD integration for function deployment
• Invoke and test deployed functions

Typical Use Cases:

1. "Create a Python function named my_handler"
2. "Deploy this code to FunctionGraph with VPC configuration"
3. "Create a function with environment variables"
4. "Invoke the function I just created"

Prerequisites

1. Python Environment

• Python 3.9+
• Verify: python --version

2. SDK Installation

pip install huaweicloudsdkfunctiongraph

3. Authentication Configuration

• Valid Huawei Cloud credentials (AK/SK mode)
• Configure via environment variables:

export HUAWEI_AK="your-access-key"
export HUAWEI_SK="your-secret-key"
export HUAWEI_REGION="cn-north-4"
export HUAWEI_PROJECT_ID="your-project-id"

Security Rules:

• NEVER expose AK/SK values in code or logs
• NEVER let users input AK/SK directly in conversation
• ALWAYS use environment variables for credentials
• Recommend using IAM user instead of main account

4. IAM Permission Requirements

• functiongraph:function:create - Create function
• functiongraph:function:get - Query function details
• functiongraph:function:list - List functions
• functiongraph:function:invoke - Invoke function

See IAM Policies for detailed permission configuration.

Workflow

User Request → Parse Parameters → Validate Input
  → Create Function (with optional VPC/Env Vars) → Return Function URN
  → (Optional) Invoke Function → Return Invocation Result

Usage

Create a Function

cd scripts

# Basic creation
python create_function.py create \
  --name my_function \
  --runtime Python3.9 \
  --handler index.handler \
  --code main.py

# With VPC configuration
python create_function.py create \
  --name my_function \
  --runtime Python3.9 \
  --handler index.handler \
  --code main.py \
  --vpc-id vpc-xxx \
  --subnet-id subnet-xxx \
  --security-group-id sg-xxx

# With environment variables
python create_function.py create \
  --name my_function \
  --runtime Python3.9 \
  --handler index.handler \
  --code main.py \
  --env-vars '{"DB_HOST":"localhost","DB_PORT":"3306"}'

# Full configuration
python create_function.py create \
  --name my_function \
  --runtime Python3.9 \
  --handler index.handler \
  --code main.py \
  --memory 256 \
  --timeout 30 \
  --description "My production function" \
  --vpc-id vpc-xxx \
  --subnet-id subnet-xxx \
  --security-group-id sg-xxx \
  --env-vars '{"DEBUG":"true","LOG_LEVEL":"info"}'

Invoke a Function

Note: The function URN uses the real FunctionGraph URN prefix urn:fss:. Replace PROJECT_ID in the examples below with your actual project ID (the same value as the HUAWEI_PROJECT_ID environment variable). The function package is default and the version is latest (or a specific version/alias). You can copy the URN returned by the create command and append :latest for invocation.

# Synchronous invocation (default)
python create_function.py invoke \
  --function-urn urn:fss:cn-north-4:PROJECT_ID:function:default:my_function:latest \
  --body '{"key1":"value1","source":"test"}'

# Asynchronous invocation
python create_function.py invoke \
  --function-urn urn:fss:cn-north-4:PROJECT_ID:function:default:my_function:latest \
  --body '{"async":true}' \
  --invocation-type async

# Invoke with log tail (returns last 4KB of log)
python create_function.py invoke \
  --function-urn urn:fss:cn-north-4:PROJECT_ID:function:default:my_function:latest \
  --body '{"test":"data"}' \
  --log-type tail

Parameter Confirmation

Create Parameters

Parameter	Required/Optional	Description	Default
--name	Required	Function name, must follow naming rules	-
--runtime	Required	Runtime environment (Python3.9, etc.)	-
--handler	Required	Function entry point (e.g., index.handler)	-
--code	Required	Code file path or inline code content	-
--memory	Optional	Memory size in MB	128
--timeout	Optional	Timeout in seconds	3
--description	Optional	Function description	-
--vpc-id	Optional	VPC ID for VPC configuration (must be set together with --subnet-id and --security-group-id)	-
--subnet-id	Optional	Subnet ID for VPC configuration (must be set together with --vpc-id and --security-group-id)	-
--security-group-id	Optional	Security group ID for VPC configuration (must be set together with --vpc-id and --subnet-id)	-
--env-vars	Optional	Environment variables as JSON string	-

Invoke Parameters

Parameter	Required/Optional	Description	Default
--function-urn	Required	Function URN to invoke	-
--body	Optional	Event payload as JSON string	{}
--invocation-type	Optional	sync or async	sync
--log-type	Optional	tail to return last 4KB log	-

Runtime Options

Runtime	Handler Format	Example
Python3.9	index.handler	def handler(event, context):
Node.js14.18	index.handler	exports.handler = (event, context) => {}
Java8	com.example.Handler::handleRequest	Java class method
Go1.x	handler	Go function name

Verification Method

See Verification Method for detailed procedures.

Quick Verification

# Verify function exists using SDK

from huaweicloudsdkfunctiongraph.v2.functiongraph_client import FunctionGraphClient
from huaweicloudsdkfunctiongraph.v2.model import ListFunctionsRequest

# List functions to verify creation
client = FunctionGraphClient.new_builder().build()
request = ListFunctionsRequest()
response = client.list_functions(request)
print(f"Total functions: {len(response.functions)}")

Output Format

Create Output

The script returns a structured JSON object:

{
  "status": "success",
  "function_urn": "urn:fss:cn-north-4:project_id:function:default:my_function",
  "function_name": "my_function",
  "runtime": "Python3.9",
  "memory_size": 128,
  "timeout": 3,
  "handler": "index.handler",
  "code_size": 1234,
  "message": "Function created successfully"
}

Invoke Output

{
  "status": "success",
  "function_urn": "urn:fss:cn-north-4:project_id:function:default:my_function",
  "invocation_type": "sync",
  "result": {
    "output": "..."
  },
  "request_id": "a1b2c3d4-...",
  "message": "Function invoked successfully"
}

Best Practices

1. Test before production: Always test in a development environment first
2. Monitor resources: Set up monitoring for function invocations and errors
3. Use environment variables: Store sensitive data in environment variables
4. Implement error handling: Add proper error handling in function code
5. Set appropriate timeouts: Configure timeout based on function logic
6. VPC for private resources: Use VPC configuration when functions need access to private network resources

Error Handling

Error Code	Description	Resolution
InvalidParameter	Invalid input parameters	Check parameter format and values
InsufficientPermission	Insufficient permissions	Check IAM permissions
QuotaExceeded	Resource quota exceeded	Request quota increase or delete unused functions
FunctionAlreadyExists	Function with same name exists	Use different function name or delete existing function
FunctionNotFound	Function URN not found (invoke)	Verify the function URN is correct

References

• FunctionGraph Documentation
• Python SDK Documentation
• IAM Policies Guide
• Verification Method
• Acceptance Criteria