Initial plan

Co-authored-by: torosent <[email protected]>

Author: Copilot

internal/durabletask-protobuf/protos/orchestrator_service.proto

@@ -41,6 +41,7 @@ message TaskFailureDetails {
     google.protobuf.StringValue stackTrace = 3;
     TaskFailureDetails innerFailure = 4;
     bool isNonRetriable = 5;
+    map<string, google.protobuf.Value> properties = 6;
 }
 
 enum OrchestrationStatus {
@@ -95,6 +96,7 @@ message TaskScheduledEvent {
     google.protobuf.StringValue version = 2;
     google.protobuf.StringValue input = 3;
     TraceContext parentTraceContext = 4;
+    map<string, string> tags = 5;
 }
 
 message TaskCompletedEvent {
@@ -192,7 +194,7 @@ message EntityOperationCalledEvent {
 }
 
 message EntityLockRequestedEvent {
-    string criticalSectionId = 1;  
+    string criticalSectionId = 1;
     repeated string lockSet = 2;
     int32 position = 3;
     google.protobuf.StringValue parentInstanceId = 4; // used only within messages, null in histories 
@@ -217,7 +219,14 @@ message EntityUnlockSentEvent {
 message EntityLockGrantedEvent {
     string criticalSectionId = 1;
 }
- 
+
+message ExecutionRewoundEvent {
+    google.protobuf.StringValue reason = 1;
+    google.protobuf.StringValue parentExecutionId = 2; // used only for rewinding suborchestrations, null otherwise
+    google.protobuf.StringValue instanceId = 3; // used only for rewinding suborchestrations, null otherwise
+    TraceContext parentTraceContext = 4; // used only for rewinding suborchestrations, null otherwise
+}
+
 message HistoryEvent {
     int32 eventId = 1;
     google.protobuf.Timestamp timestamp = 2;
@@ -244,25 +253,29 @@ message HistoryEvent {
         ExecutionResumedEvent executionResumed = 22;
         EntityOperationSignaledEvent entityOperationSignaled = 23;
         EntityOperationCalledEvent entityOperationCalled = 24;
-        EntityOperationCompletedEvent entityOperationCompleted = 25; 
-        EntityOperationFailedEvent entityOperationFailed = 26; 
+        EntityOperationCompletedEvent entityOperationCompleted = 25;
+        EntityOperationFailedEvent entityOperationFailed = 26;
         EntityLockRequestedEvent entityLockRequested = 27;
         EntityLockGrantedEvent entityLockGranted = 28;
         EntityUnlockSentEvent entityUnlockSent = 29;
+        ExecutionRewoundEvent executionRewound = 30;
     }
 }
 
 message ScheduleTaskAction {
     string name = 1;
     google.protobuf.StringValue version = 2;
     google.protobuf.StringValue input = 3;
+    map<string, string> tags = 4;
+    TraceContext parentTraceContext = 5;
 }
 
 message CreateSubOrchestrationAction {
     string instanceId = 1;
     string name = 2;
     google.protobuf.StringValue version = 3;
     google.protobuf.StringValue input = 4;
+    TraceContext parentTraceContext = 5;
 }
 
 message CreateTimerAction {
@@ -282,6 +295,7 @@ message CompleteOrchestrationAction {
     google.protobuf.StringValue newVersion = 4;
     repeated HistoryEvent carryoverEvents = 5;
     TaskFailureDetails failureDetails = 6;
+    map<string, string> tags = 7;
 }
 
 message TerminateOrchestrationAction {
@@ -312,6 +326,11 @@ message OrchestratorAction {
     }
 }
 
+message OrchestrationTraceContext {
+    google.protobuf.StringValue spanID = 1;
+    google.protobuf.Timestamp spanStartTime = 2;   
+}
+
 message OrchestratorRequest {
     string instanceId = 1;
     google.protobuf.StringValue executionId = 2;
@@ -320,6 +339,8 @@ message OrchestratorRequest {
     OrchestratorEntityParameters entityParameters = 5;
     bool requiresHistoryStreaming = 6;
     map<string, google.protobuf.Value> properties = 7;
+
+    OrchestrationTraceContext orchestrationTraceContext = 8;
 }
 
 message OrchestratorResponse {
@@ -331,6 +352,17 @@ message OrchestratorResponse {
     // The number of work item events that were processed by the orchestrator.
     // This field is optional. If not set, the service should assume that the orchestrator processed all events.
     google.protobuf.Int32Value numEventsProcessed = 5;
+    OrchestrationTraceContext orchestrationTraceContext = 6;
+
+    // Whether or not a history is required to complete the original OrchestratorRequest and none was provided.
+    bool requiresHistory = 7;
+
+    // True if this is a partial (chunked) completion. The backend must keep the work item open until the final chunk (isPartial=false).
+    bool isPartial = 8;
+
+    // Zero-based position of the current chunk within a chunked completion sequence.
+    // This field is omitted for non-chunked completions.
+    google.protobuf.Int32Value chunkIndex = 9;
 }
 
 message CreateInstanceRequest {
@@ -343,6 +375,7 @@ message CreateInstanceRequest {
     google.protobuf.StringValue executionId = 7;
     map<string, string> tags = 8;
     TraceContext parentTraceContext = 9;
+    google.protobuf.Timestamp requestTime = 10;
 }
 
 message OrchestrationIdReusePolicy {
@@ -453,6 +486,7 @@ message PurgeInstancesRequest {
     oneof request {
         string instanceId = 1;
         PurgeInstanceFilter purgeInstanceFilter = 2;
+        InstanceBatch instanceBatch = 4;
     }
     bool recursive = 3;
 }
@@ -468,6 +502,15 @@ message PurgeInstancesResponse {
     google.protobuf.BoolValue isComplete = 2;
 }
 
+message RestartInstanceRequest {
+    string instanceId = 1;
+    bool restartWithNewInstanceId = 2;
+}
+
+message RestartInstanceResponse {
+    string instanceId = 1;
+}
+
 message CreateTaskHubRequest {
     bool recreateIfExists = 1;
 }
@@ -490,10 +533,12 @@ message SignalEntityRequest {
     google.protobuf.StringValue input = 3;
     string requestId = 4;
     google.protobuf.Timestamp scheduledTime = 5;
+    TraceContext parentTraceContext = 6;
+    google.protobuf.Timestamp requestTime = 7;
 }
 
 message SignalEntityResponse {
-     // no payload
+    // no payload
 }
 
 message GetEntityRequest {
@@ -575,6 +620,7 @@ message OperationRequest {
     string operation = 1;
     string requestId = 2;
     google.protobuf.StringValue input = 3;
+    TraceContext traceContext = 4;
 }
 
 message OperationResult {
@@ -591,10 +637,14 @@ message OperationInfo {
 
 message OperationResultSuccess {
     google.protobuf.StringValue result = 1;
+    google.protobuf.Timestamp startTimeUtc = 2;
+    google.protobuf.Timestamp endTimeUtc = 3;
 }
 
 message OperationResultFailure {
     TaskFailureDetails failureDetails = 1;
+    google.protobuf.Timestamp startTimeUtc = 2;
+    google.protobuf.Timestamp endTimeUtc = 3;
 }
 
 message OperationAction {
@@ -610,6 +660,8 @@ message SendSignalAction {
     string name = 2;
     google.protobuf.StringValue input = 3;
     google.protobuf.Timestamp scheduledTime = 4;
+    google.protobuf.Timestamp requestTime = 5;
+    TraceContext parentTraceContext = 6;
 }
 
 message StartNewOrchestrationAction {
@@ -618,6 +670,8 @@ message StartNewOrchestrationAction {
     google.protobuf.StringValue version = 3;
     google.protobuf.StringValue input = 4;
     google.protobuf.Timestamp scheduledTime = 5;
+    google.protobuf.Timestamp requestTime = 6;
+    TraceContext parentTraceContext = 7;
 }
 
 message AbandonActivityTaskRequest {
@@ -644,6 +698,17 @@ message AbandonEntityTaskResponse {
     // Empty.
 }
 
+message SkipGracefulOrchestrationTerminationsRequest {
+    InstanceBatch instanceBatch = 1;
+    google.protobuf.StringValue reason = 2;
+}
+
+message SkipGracefulOrchestrationTerminationsResponse {
+    // Those instances which could not be terminated because they had locked entities at the time of this termination call,
+    // are already in a terminal state (completed, failed, terminated, etc.), are not orchestrations, or do not exist (i.e. have been purged)
+    repeated string unterminatedInstanceIds = 1;
+}
+
 service TaskHubSidecarService {
     // Sends a hello request to the sidecar service.
     rpc Hello(google.protobuf.Empty) returns (google.protobuf.Empty);
@@ -657,18 +722,21 @@ service TaskHubSidecarService {
     // Rewinds an orchestration instance to last known good state and replays from there.
     rpc RewindInstance(RewindInstanceRequest) returns (RewindInstanceResponse);
 
+    // Restarts an orchestration instance.
+    rpc RestartInstance(RestartInstanceRequest) returns (RestartInstanceResponse);
+
     // Waits for an orchestration instance to reach a running or completion state.
     rpc WaitForInstanceStart(GetInstanceRequest) returns (GetInstanceResponse);
-    
+
     // Waits for an orchestration instance to reach a completion state (completed, failed, terminated, etc.).
     rpc WaitForInstanceCompletion(GetInstanceRequest) returns (GetInstanceResponse);
 
     // Raises an event to a running orchestration instance.
     rpc RaiseEvent(RaiseEventRequest) returns (RaiseEventResponse);
-    
+
     // Terminates a running orchestration instance.
     rpc TerminateInstance(TerminateRequest) returns (TerminateResponse);
-    
+
     // Suspends a running orchestration instance.
     rpc SuspendInstance(SuspendRequest) returns (SuspendResponse);
 
@@ -714,6 +782,10 @@ service TaskHubSidecarService {
 
     // Abandon an entity work item
     rpc AbandonTaskEntityWorkItem(AbandonEntityTaskRequest) returns (AbandonEntityTaskResponse);
+
+    // "Skip" graceful termination of orchestrations by immediately changing their status in storage to "terminated".
+    // Note that a maximum of 500 orchestrations can be terminated at a time using this method.
+    rpc SkipGracefulOrchestrationTerminations(SkipGracefulOrchestrationTerminationsRequest) returns (SkipGracefulOrchestrationTerminationsResponse);
 }
 
 message GetWorkItemsRequest {
@@ -732,6 +804,16 @@ enum WorkerCapability {
     // When set, the service may return work items without any history events as an optimization.
     // It is strongly recommended that all SDKs support this capability.
     WORKER_CAPABILITY_HISTORY_STREAMING = 1;
+
+    // Indicates that the worker supports scheduled tasks.
+    // The service may send schedule-triggered orchestration work items,
+    // and the worker must handle them, including the scheduledTime field.
+    WORKER_CAPABILITY_SCHEDULED_TASKS = 2;
+
+    // Signals that the worker can handle large payloads stored externally (e.g., Blob Storage).
+    // Work items may contain URI references instead of inline data, and the worker must fetch them.
+    // This avoids message size limits and reduces network overhead.
+    WORKER_CAPABILITY_LARGE_PAYLOADS = 3;
 }
 
 message WorkItem {
@@ -750,7 +832,7 @@ message CompleteTaskResponse {
 }
 
 message HealthPing {
-	// No payload
+    // No payload
 }
 
 message StreamInstanceHistoryRequest {
@@ -764,3 +846,8 @@ message StreamInstanceHistoryRequest {
 message HistoryChunk {
     repeated HistoryEvent events = 1;
 }
+
+message InstanceBatch {
+    // A maximum of 500 instance IDs can be provided in this list.
+    repeated string instanceIds = 1;
+}