Retry Configuration

Configure automatic retry with exponential backoff for transient failures.

Configuration

Add retry fields to any target:

targets:
  - name: azure_base
    provider: azure
    endpoint: ${{ AZURE_OPENAI_ENDPOINT }}
    api_key: ${{ AZURE_OPENAI_API_KEY }}
    model: ${{ AZURE_DEPLOYMENT_NAME }}
    max_retries: 5
    retry_initial_delay_ms: 2000
    retry_max_delay_ms: 120000
    retry_backoff_factor: 2
    retry_status_codes: [500, 408, 429, 502, 503, 504]

Fields

Field	Default	Description
max_retries	—	Maximum number of retry attempts
retry_initial_delay_ms	—	Initial delay before first retry (milliseconds)
retry_max_delay_ms	—	Maximum delay between retries (milliseconds)
retry_backoff_factor	—	Multiplier for exponential backoff
retry_status_codes	—	HTTP status codes that trigger a retry

Behavior

• Retries use exponential backoff with jitter to avoid thundering herd
• Rate limit errors (429) and transient server errors (5xx) are automatically retried
• Network failures trigger retries
• The delay between retries doubles each attempt (up to retry_max_delay_ms)