The Durable Task Framework is a lightweight, embeddable engine for writing durable, fault-tolerant business logic (orchestrations) as ordinary code. The engine itself is written in Go and intended to be embedded into other Go-based processes. It exposes a gRPC endpoint to support writing durable flows in any language. There are currently SDKs that consume this gRPC endpoint for .NET and Java, with more to come. It's also possible to write orchestrations directly in Go and run them in the local process.
This project is largely a Go clone of the .NET-based Durable Task Framework, which is used by various cloud service teams at Microsoft for building reliable control planes and managing infrastructure. It also takes inspiration from the Go Workflows project, which itself is a Go project that borrows heavily from both the Durable Task Framework and Temporal. The main difference is that the Durable Task engine is designed to be used in sidecar architectures.
The Durable Task engine is also intended to be used as the basis for the Dapr embedded workflow engine.
This project is a work-in-progress and should not be used for production workloads. The public API surface is also not yet stable. The project itself is also in the very early stages and is missing some of the basics, such as contribution guidelines, etc.
This project includes a sqlite storage provider for persisting app state to disk.
// Persists state to a file named test.sqlite3. Use "" for in-memory storage.
options := sqlite.NewSqliteOptions("test.sqlite3")
be := sqlite.NewSqliteBackend(options, backend.DefaultLogger())Additional storage providers can be created by extending the Backend interface.
See the main.go file for an example of how to create a standalone gRPC sidecar that embeds the Durable Task engine. In short, you must create an Backend (for storage), an Executor (for executing user code), and host them as a TaskHubWorker.
The following code creates a TaskHub worker with sqlite Backend and a gRPC Executor implementations.
// Use the default logger or provide your own
logger := backend.DefaultLogger()
// Configure the sqlite backend that will store the runtime state
sqliteOptions := sqlite.NewSqliteOptions(sqliteFilePath)
be := sqlite.NewSqliteBackend(sqliteOptions, logger)
// Create a gRPC server that the language SDKs will connect to
grpcServer := grpc.NewServer()
executor := backend.NewGrpcExecutor(grpcServer, be, logger)
// Construct and start the task hub worker object, which polls the backend for new work
orchestrationWorker := backend.NewOrchestrationWorker(be, executor, logger)
activityWorker := backend.NewActivityTaskWorker(be, executor, logger)
taskHubWorker := backend.NewTaskHubWorker(be, orchestrationWorker, activityWorker, logger)
taskHubWorker.Start(context.Background())
// Start listening.
lis, _ := net.Listen("tcp", "localhost:4001")
fmt.Printf("server listening at %v\n", lis.Addr())
grpcServer.Serve(lis)Note that the Durable Task gRPC service implementation is designed to serve one client at a time, just like with any sidecar architecture. Scale out is achieved by adding new pod replicas that contain both the app process and the sidecar (connected to a common database).
The Durable Task Framework for Go currently supports writing orchestrations in the following languages:
| Language/Stack | Package | Project Home | Samples |
|---|---|---|---|
| .NET |  |
GitHub | Samples |
| Java |  |
GitHub | Samples |
| Python |  |
GitHub | Samples |
More language SDKs are planned to be added in the future. In particular, SDKs for Python and JavaScript/TypeScript. Anyone can theoretically create an SDK using a language that supports gRPC. However, there is not yet a guide for how to do this, so developers would need to reference existing SDK code as a reference. Starting with the Java implementation is recommended. The gRPC API is defined here.
It's also possible to create orchestrations in Go and run them in the local process. You can find code samples in the samples directory. The full set of Durable Task features is not yet available as part of the Go SDK, but will be added over time.
Activity sequences like the following are the simplest and most common pattern used in the Durable Task Framework.
// ActivitySequenceOrchestrator makes three activity calls in sequence and results the results
// as an array.
func ActivitySequenceOrchestrator(ctx *task.OrchestrationContext) (any, error) {
var helloTokyo string
if err := ctx.CallActivity(SayHelloActivity, task.WithActivityInput("Tokyo")).Await(&helloTokyo); err != nil {
return nil, err
}
var helloLondon string
if err := ctx.CallActivity(SayHelloActivity, task.WithActivityInput("London")).Await(&helloLondon); err != nil {
return nil, err
}
var helloSeattle string
if err := ctx.CallActivity(SayHelloActivity, task.WithActivityInput("Seattle")).Await(&helloSeattle); err != nil {
return nil, err
}
return []string{helloTokyo, helloLondon, helloSeattle}, nil
}
// SayHelloActivity can be called by an orchestrator function and will return a friendly greeting.
func SayHelloActivity(ctx task.ActivityContext) (any, error) {
var input string
if err := ctx.GetInput(&input); err != nil {
return "", err
}
return fmt.Sprintf("Hello, %s!", input), nil
}You can find the full sample here.
The next most common pattern is "fan-out / fan-in" where multiple activities are run in parallel, as shown in the snippet below (note that the GetDevicesToUpdate and UpdateDevice activity definitions are left out of the snippet below for brevity):
// UpdateDevicesOrchestrator is an orchestrator that runs activities in parallel
func UpdateDevicesOrchestrator(ctx *task.OrchestrationContext) (any, error) {
// Get a dynamic list of devices to perform updates on
var devices []string
if err := ctx.CallActivity(GetDevicesToUpdate).Await(&devices); err != nil {
return nil, err
}
// Start a dynamic number of tasks in parallel, not waiting for any to complete (yet)
tasks := make([]task.Task, 0, len(devices))
for _, id := range devices {
tasks = append(tasks, ctx.CallActivity(UpdateDevice, task.WithActivityInput(id)))
}
// Now that all are started, wait for them to complete and then return the success rate
successCount := 0
for _, task := range tasks {
var succeeded bool
if err := task.Await(&succeeded); err == nil && succeeded {
successCount++
}
}
return float32(successCount) / float32(len(devices)), nil
}The full sample can be found here.
Sometimes orchestrations need asynchronous input from external systems. For example, an approval workflow may require a manual approval signal from an authorized user. Or perhaps an orchestration pauses and waits for a command from an operator. The WaitForSingleEvent method can be used in an orchestrator function to pause execution and wait for such inputs. You an even specify a timeout value indicating how long to wait for the input before resuming execution (use -1 to indicate infinite timeout).
// ExternalEventOrchestrator is an orchestrator function that blocks for 30 seconds or
// until a "Name" event is sent to it.
func ExternalEventOrchestrator(ctx *task.OrchestrationContext) (any, error) {
var nameInput string
if err := ctx.WaitForSingleEvent("Name", 30*time.Second).Await(&nameInput); err != nil {
// Timeout expired
return nil, err
}
return fmt.Sprintf("Hello, %s!", nameInput), nil
}Sending an event to a waiting orchestration can be done using the RaiseEvent method of the task hub client. These events are durably buffered in the orchestration state and are consumed as soon as the target orchestration calls WaitForSingleEvent with a matching event name. The following code shows how to use the RaiseEvent method to send an event with a payload to a running orchestration. See Managing local orchestrations for more information on how to interact with local orchestrations in Go.
id, _ := client.ScheduleNewOrchestration(ctx, ExternalEventOrchestrator)
// Prompt the user for their name and send that to the orchestrator
go func() {
fmt.Println("Enter your first name: ")
var nameInput string
fmt.Scanln(&nameInput)
client.RaiseEvent(ctx, id, "Name", api.WithEventPayload(nameInput))
}()The full sample can be found here.
The following code snippet provides an example of how you can configure and run orchestrations. The TaskRegistry type allows you to register orchestrator and activity functions, and the TaskHubClient allows you to start, query, terminate, suspend, resume, and wait for orchestrations to complete.
The code snippet below demonstrates how to register and start a new instance of the ActivitySequenceOrchestrator orchestrator and wait for it to complete. The initialization of the client and worker are left out for brevity.
r := task.NewTaskRegistry()
r.AddOrchestrator(ActivitySequenceOrchestrator)
r.AddActivity(SayHelloActivity)
ctx := context.Background()
client, worker := Init(ctx, r)
defer worker.Shutdown(ctx)
id, err := client.ScheduleNewOrchestration(ctx, ActivitySequenceOrchestrator)
if err != nil {
panic(err)
}
metadata, err := client.WaitForOrchestrationCompletion(ctx, id)
if err != nil {
panic(err)
}
fmt.Printf("orchestration completed: %v\n", metadata)Each sample linked above has a full implementation you can use as a reference.
The Durable Task Framework for Go supports publishing distributed traces to any configured Open Telemetry-compatible exporter. Simply use otel.SetTracerProvider(tp) to register a global TracerProvider as part of your application startup and the task hub worker will automatically use it to emit OLTP trace spans.
The following example code shows how you can configure distributed trace collection with Zipkin, a popular open source distributed tracing system. The example assumes Zipkin is running locally, as shown in the code.
func ConfigureZipkinTracing() *trace.TracerProvider {
// Inspired by this sample: https://github.com/open-telemetry/opentelemetry-go/blob/main/example/zipkin/main.go
exp, err := zipkin.New("http://localhost:9411/api/v2/spans")
if err != nil {
log.Fatal(err)
}
// NOTE: The simple span processor is not recommended for production.
// Instead, the batch span processor should be used for production.
processor := trace.NewSimpleSpanProcessor(exp)
// processor := trace.NewBatchSpanProcessor(exp)
tp := trace.NewTracerProvider(
trace.WithSpanProcessor(processor),
trace.WithSampler(trace.AlwaysSample()),
trace.WithResource(resource.NewWithAttributes(
"durabletask.io",
attribute.KeyValue{Key: "service.name", Value: attribute.StringValue("sample-app")},
)),
)
otel.SetTracerProvider(tp)
return tp
}You can find this code in the distributedtracing.go sample. The following is a screenshot showing the trace for the sample's orchestration, which calls an activity, creates a 2-second durable timer, and uses another activity to make an HTTP request to bing.com:
Note that each orchestration is represented as a single span with activities, timers, and sub-orchestrations as child spans. The generated spans contain a variety of attributes that include information such as orchestration instance IDs, task names, task IDs, etc.
This repository contains submodules. Be sure to clone it with the option to include submodules. Otherwise you will not be able to generate the protobuf code.
git clone --recurse-submodules https://github.com/microsoft/durabletask-go This project requires go v1.18.x or greater. You can build a standalone executable by simply running go build at the project root.
Use the following command to regenerate the protobuf from the submodule. Use this whenever updating the submodule reference.
# NOTE: assumes the .proto file defines: option go_package = "/internal/protos"
protoc --go_out=. --go-grpc_out=. -I submodules/durabletask-protobuf/protos orchestrator_service.protoTest mocks were generated using mockery. Use the following command at the project root to regenerate the mocks.
mockery --dir ./backend --name="^Backend|^Executor|^TaskWorker" --output ./tests/mocks --with-expecterAll automated tests are under ./tests. A separate test package hierarchy was chosen intentionally to prioritize black box testing. This strategy also makes it easier to catch accidental breaking API changes.
Run tests with the following command.
go test ./tests/... -coverpkg ./api,./task,./client,./backend/...,./internal/helpersYou can run pre-built container images to run full integration tests against the durable task host over gRPC.
Use the following docker command to run tests against a running worker.
docker run -e GRPC_HOST="host.docker.internal" cgillum/durabletask-dotnet-tester:0.5.0-betaNote that the test assumes the gRPC server can be reached over localhost on port 4001 on the host machine. These values can be overridden with the following environment variables:
GRPC_HOST: Use this to change from the default127.0.0.1to some other value, for examplehost.docker.internal.GRPC_PORT: Set this environment variable to change the default port from4001to something else.
If successful, you should see output that looks like the following:
Test run for /root/out/bin/Debug/Microsoft.DurableTask.Tests/net6.0/Microsoft.DurableTask.Tests.dll (.NETCoreApp,Version=v6.0)
Microsoft (R) Test Execution Command Line Tool Version 17.3.1 (x64)
Copyright (c) Microsoft Corporation. All rights reserved.
Starting test execution, please wait...
A total of 1 test files matched the specified pattern.
[xUnit.net 00:00:00.00] xUnit.net VSTest Adapter v2.4.3+1b45f5407b (64-bit .NET 6.0.10)
[xUnit.net 00:00:00.82] Discovering: Microsoft.DurableTask.Tests
[xUnit.net 00:00:00.90] Discovered: Microsoft.DurableTask.Tests
[xUnit.net 00:00:00.90] Starting: Microsoft.DurableTask.Tests
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.ExternalEvents(eventCount: 100) [6 s]
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.ExternalEvents(eventCount: 1) [309 ms]
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.LongTimer [8 s]
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.SubOrchestration [1 s]
...
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.ActivityFanOut [914 ms]
[xUnit.net 00:01:01.04] Finished: Microsoft.DurableTask.Tests
Passed Microsoft.DurableTask.Tests.OrchestrationPatterns.SingleActivity_Async [365 ms]
Test Run Successful.
Total tests: 33
Passed: 33
Total time: 1.0290 Minutes
You can run the engine locally by pressing F5 in Visual Studio Code (the recommended editor). You can also simply run go run main.go to start a local Durable Task gRPC server that listens on port 4001.
go run main.go --port 4001 --db ./test.sqlite3The following is the expected output:
2022/09/14 17:26:50 backend started: sqlite::./test.sqlite3
2022/09/14 17:26:50 server listening at 127.0.0.1:4001
2022/09/14 17:26:50 orchestration-processor: waiting for new work items...
2022/09/14 17:26:50 activity-processor: waiting for new work items...
At this point you can use one of the language SDKs mentioned earlier in a separate process to implement and execute durable orchestrations. Those SDKs will connect to port 4001 by default to interact with the Durable Task engine.
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.
This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.