Rewind support seems broken

[kaibocai]

Description

A clear and concise description of what the bug is. Please make an effort to fill in all the sections below; the information will help us investigate your issue.

We are working out the rewind support on durabletask-java, at microsoft/durabletask-java#123. In our end-to-end test, we found an exception thrown out (more details are microsoft/durabletask-java#123 (review)), to me this comes from the extension. Note that this exception is thrown out both using the POST HTTP request or using the client.rewind method.

Already had a quick chat with @jviau, it seems the extension drops the value that needed to be set in the grpc message somewhere.
I spent sometime look into the extension but with no luck on this due to lack of knowledge in extension, so open a issue here to track it.

I am open to have a session on this if it is preferred.

cc @davidmrdavid as well.

NOTE: JavaScript issues should be reported here: https://github.com/Azure/azure-functions-durable-js

Expected behavior

A clear and concise description of what you expected to happen.

Actual behavior

A clear and concise description of what happened.

Relevant source code snippets

// insert code snippet here

Known workarounds

Provide a description of any known workarounds you used.

App Details

• Durable Functions extension version (e.g. v1.8.3):
• Azure Functions runtime version (1.0 or 2.0):
• Programming language used:

Screenshots

If applicable, add screenshots to help explain your problem.

If deployed to Azure

We have access to a lot of telemetry that can help with investigations. Please provide as much of the following information as you can to help us investigate!

• Timeframe issue observed:
• Function App name:
• Function name(s):
• Azure region:
• Orchestration instance ID(s):
• Azure storage account name:

If you don't want to share your Function App or storage account name GitHub, please at least share the orchestration instance ID. Otherwise it's extremely difficult to look up information.

[kaibocai]

Related update on proto microsoft/durabletask-protobuf#18. Though, still need to find why reason is dropped.

[kaibocai]

With @davidmrdavid 's help, we debug into the extension and I found here

azure-functions-durable-extension/src/WebJobs.Extensions.DurableTask/Bindings/OrchestrationTriggerAttributeBindingProvider.cs

Lines 171 to 172 in 9f750eb

	PastEvents = { remoteContext.PastEvents.Select(ProtobufUtils.ToHistoryEventProto) },
	NewEvents = { remoteContext.NewEvents.Select(ProtobufUtils.ToHistoryEventProto) },

theremoteContext.PastEvents already contains a few GenericEvents that have null as value of Data, and that causing the exception in ProtobufUtils.ToHistoryEventProto. I am not quite sure where are those GenericEvents come from.

The GenericEvent created by rewind API seems in the remoteContext.newEvents and has the correct data set to rewind reason.

Also, I am told the logic for handling HTTP requests for java is totally different from js and python, java is using grpc while js and python is using HTTP. That may explain why I can run rewind successfully using js.

Hopefully the PR here microsoft/durabletask-protobuf#18 will have the issue resolved

cc @cgillum , @davidmrdavid , @jviau

[CSharpFiasco]

What's the status on this?

[cgillum]

@CSharpFiasco this work got put on pause for a while, but it's on our backlog to get the API implemented in a way that will make it available to all language SDKs.