Implementation:BerriAI Litellm Custom Logger Registry

Attribute	Value
Sources	litellm/litellm_core_utils/custom_logger_registry.py
Domains	Logging, Callbacks, Integrations
Last Updated	2026-02-15 16:00 GMT

Overview

Central registry that maps callback name strings (e.g., "datadog", "prometheus") to their corresponding logger class types.

Description

The CustomLoggerRegistry class maintains a class-level dictionary CALLBACK_CLASS_STR_TO_CLASS_TYPE that associates short string identifiers with logger/callback class implementations. It covers over 30 integrations including DataDog, Prometheus, Langfuse, OpenTelemetry, GCS, Argilla, Opik, and many more. Multiple string identifiers can map to the same class (e.g., "logfire", "arize", "otel" all map to OpenTelemetry). Enterprise loggers (PagerDuty, GenericAPI, email loggers) are conditionally added if the litellm_enterprise package is installed. The class provides methods for bidirectional lookup: string-to-class and class-to-string(s).

Usage

Import this registry when you need to resolve a callback name string from configuration to its actual class, or when you need to discover which string identifier(s) correspond to a logger class. This is used internally by LiteLLM's callback initialization pipeline.

Code Reference

Source Location

litellm/litellm_core_utils/custom_logger_registry.py (178 lines)

Signature

class CustomLoggerRegistry:
    CALLBACK_CLASS_STR_TO_CLASS_TYPE: dict  # str -> type

    @classmethod
    def get_callback_str_from_class_type(cls, class_type: type) -> Union[str, None]

    @classmethod
    def get_all_callback_strs_from_class_type(cls, class_type: type) -> list[str]

    @classmethod
    def get_class_type_for_custom_logger_name(
        cls,
        custom_logger_name: _custom_logger_compatible_callbacks_literal,
    ) -> type

Import

from litellm.litellm_core_utils.custom_logger_registry import CustomLoggerRegistry

I/O Contract

`get_callback_str_from_class_type`

Direction	Name	Type	Description
Input	class_type	`type`	Logger class to look up
Output	return	`Union[str, None]`	First matching callback string, or `None`

`get_all_callback_strs_from_class_type`

Direction	Name	Type	Description
Input	class_type	`type`	Logger class to look up
Output	return	`list[str]`	All callback strings mapping to this class

`get_class_type_for_custom_logger_name`

Direction	Name	Type	Description
Input	custom_logger_name	`_custom_logger_compatible_callbacks_literal`	String identifier for the logger
Output	return	`type`	The class type for the named logger

Usage Examples

from litellm.litellm_core_utils.custom_logger_registry import CustomLoggerRegistry
from litellm.integrations.opentelemetry import OpenTelemetry

# Resolve a callback string to its class
cls = CustomLoggerRegistry.get_class_type_for_custom_logger_name("datadog")
# cls == DataDogLogger

# Find the string for a class
name = CustomLoggerRegistry.get_callback_str_from_class_type(OpenTelemetry)
# name == "opentelemetry" (first match)

# Find all strings that map to OpenTelemetry
all_names = CustomLoggerRegistry.get_all_callback_strs_from_class_type(OpenTelemetry)
# ["opentelemetry", "logfire", "arize", "langfuse_otel", "arize_phoenix", "langtrace", "weave_otel", "levo", "otel"]

Related Pages

BerriAI_Litellm_DD_Tracing -- DataDog tracing integration referenced in the registry

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment