durable-execution-sdk-js.durableexecutionclient.md

December 2, 2025 · View on GitHub

Home > @aws/durable-execution-sdk-js > DurableExecutionClient

DurableExecutionClient interface

Client interface for durable execution backend operations.

This interface defines the core operations needed to manage durable execution state and checkpoints. It abstracts the underlying service calls and provides a clean contract for: - Retrieving execution state during replay scenarios - Creating checkpoints for state persistence - Managing long-running workflow state

Implementations of this interface handle the communication with AWS services to ensure durable execution capabilities including automatic retry, state recovery, and workflow orchestration.

Signature:

export interface DurableExecutionClient

Methods

Method	Description
checkpoint(params, logger)	Creates a checkpoint to persist the current execution state. This method saves the current execution progress, step results, and context data to enable recovery and replay capabilities. Checkpoints are created automatically at key points during execution (after steps complete, before waits, etc.) and can also be triggered manually for custom persistence needs. Checkpointing enables: - Automatic recovery from timeouts or failures - Resumption of long-running workflows - Replay semantics for consistent execution - State persistence across execution boundaries
getExecutionState(params, logger)	Retrieves the current execution state for a durable execution. This method fetches the persisted state data from the durable execution backend, including step history, checkpoint data, and execution metadata. It's primarily used during replay scenarios to restore the execution context and determine which operations have already been completed. The execution state contains all information needed to resume a durable function from where it left off, enabling fault tolerance and long-running workflows.

Method

Description

checkpoint(params, logger)

Creates a checkpoint to persist the current execution state.

This method saves the current execution progress, step results, and context data to enable recovery and replay capabilities. Checkpoints are created automatically at key points during execution (after steps complete, before waits, etc.) and can also be triggered manually for custom persistence needs.

Checkpointing enables: - Automatic recovery from timeouts or failures - Resumption of long-running workflows - Replay semantics for consistent execution - State persistence across execution boundaries

getExecutionState(params, logger)

Retrieves the current execution state for a durable execution.

This method fetches the persisted state data from the durable execution backend, including step history, checkpoint data, and execution metadata. It's primarily used during replay scenarios to restore the execution context and determine which operations have already been completed.

The execution state contains all information needed to resume a durable function from where it left off, enabling fault tolerance and long-running workflows.