[DurableExecution]: Invalid checkpoint token misclassified as terminal; execution fails instead of invocation

[yaythomas]

Describe the bug

Amazon.Lambda.DurableExecution matches the stale checkpoint token rejection with a case-sensitive prefix check against "Invalid Checkpoint Token". The service emits "Invalid checkpoint token" (lowercase c and t). The check never matches, so a stale token is treated as a terminal 4xx and the whole durable execution fails instead of just the invocation.

Regression Issue

• Select this option if this issue appears to be a regression.

Expected Behavior

When the service rejects a checkpoint call with InvalidParameterValueException because the checkpoint token is stale (a newer invocation superseded this one), IsTerminalCheckpointError returns false. The exception escapes to the host, Lambda retries the invocation, the service issues a fresh token, and the execution continues.

Current Behavior

IsTerminalCheckpointError returns true. The catch in DurableFunction.FunctionHandlerAsync returns Status = InvocationStatus.Failed. The execution is failed every time a stale token is seen.

Root cause, in Libraries/src/Amazon.Lambda.DurableExecution/DurableFunction.cs:

aws-lambda-dotnet/Libraries/src/Amazon.Lambda.DurableExecution/DurableFunction.cs

Lines 202 to 207 in 8ed93ce

	if (ex.ErrorCode == "InvalidParameterValueException"
	&& ex.Message != null
	&& ex.Message.StartsWith("Invalid Checkpoint Token", StringComparison.Ordinal))
	{
	return false;
	}

if (ex.ErrorCode == "InvalidParameterValueException"
    && ex.Message != null
    && ex.Message.StartsWith("Invalid Checkpoint Token", StringComparison.Ordinal))
{
    return false;
}

The service constant (ValidationErrorMessages.INVALID_CHECKPOINT_TOKEN, internal) is "Invalid checkpoint token".

The existing test uses the SDK's string rather than the service's, so it cannot catch this:

aws-lambda-dotnet/Libraries/test/Amazon.Lambda.DurableExecution.Tests/DurableFunctionTests.cs

Line 488 in 8ed93ce

new object[] { MakeServiceException("InvalidParameterValueException", HttpStatusCode.BadRequest, "Invalid Checkpoint Token: stale") },

Reproduction Steps

1. Construct an AmazonServiceException with ErrorCode = "InvalidParameterValueException", StatusCode = 400, and message "Invalid checkpoint token: ..." (the real service message).
2. Call IsTerminalCheckpointError with it (or run a workflow whose checkpoint flush throws it).
3. Observe it returns true and the invocation output is Failed.

Possible Solution

1. Change the literal to "Invalid checkpoint token", or use StringComparison.OrdinalIgnoreCase to protect against future drift.
2. Update DurableFunctionTests to use the real service message.

Additional Information/Context

The same defect exists in the JS, Python, Java, Rust, and Go durable execution SDKs; each has its own issue. Links are in a comment below.

AWS .NET SDK and/or Package version used

Amazon.Lambda.DurableExecution 2.0.0 (master at 8ed93ce)

Targeted .NET Platform

Any

Operating System and version

Any

[yaythomas]

Sibling issues (same bug, one per SDK):

• JS: [Bug]: Invalid checkpoint token misclassified as execution failure (message case mismatch) aws-durable-execution-sdk-js#917
• Python: [Bug]: Invalid checkpoint token misclassified as execution failure (message case mismatch) aws-durable-execution-sdk-python#721
• Java: [Bug]: Invalid checkpoint token misclassified as execution failure (message case mismatch) aws-durable-execution-sdk-java#705
• Rust: [Bug]: Invalid checkpoint token misclassified as execution failure (message case mismatch) aws-durable-execution-sdk-rust#66
• Go: https://github.com/aws/aws-durable-execution-sdk-go/issues/6
• .NET: [DurableExecution]: Invalid checkpoint token misclassified as terminal; execution fails instead of invocation #2571
  Prerequisite relationship: [DurableExecution]: Return Pending when a checkpoint response has no CheckpointToken #2572 depends on this fix.