[RFC] Add checkpoint support for streamlookup (#9299)

(cherry picked from commit c5d3c1a)
ydb-platform · Sep 24, 2024 · 39c1baa · 39c1baa
1 parent 1ec7315
commit 39c1baa
Show file tree

Hide file tree

Showing 5 changed files with 36 additions and 10 deletions.
diff --git a/ydb/library/yql/dq/actors/compute/dq_compute_actor_impl.h b/ydb/library/yql/dq/actors/compute/dq_compute_actor_impl.h
@@ -1369,8 +1369,20 @@ class TDqComputeActorBase : public NActors::TActorBootstrapped<TDerived>
     }
 
     void PollAsyncInput() {
+        if (!Running) {
+            CA_LOG_T("Skip polling inputs and sources because not running");
+            return;
+        }
+
+        CA_LOG_T("Poll inputs");
+        for (auto& [inputIndex, transform] : InputTransformsMap) {
+            if (auto resume = transform.PollAsyncInput(MetricsReporter, WatermarksTracker, RuntimeSettings.AsyncInputPushLimit)) {
+                ContinueExecute(*resume);
+            }
+        }
+
         // Don't produce any input from sources if we're about to save checkpoint.
-        if (!Running || (Checkpoints && Checkpoints->HasPendingCheckpoint() && !Checkpoints->ComputeActorStateSaved())) {
+        if ((Checkpoints && Checkpoints->HasPendingCheckpoint() && !Checkpoints->ComputeActorStateSaved())) {
             CA_LOG_T("Skip polling sources because of pending checkpoint");
             return;
         }
@@ -1381,13 +1393,6 @@ class TDqComputeActorBase : public NActors::TActorBootstrapped<TDerived>
                 ContinueExecute(*resume);
             }
         }
-
-        CA_LOG_T("Poll inputs");
-        for (auto& [inputIndex, transform] : InputTransformsMap) {
-            if (auto resume = transform.PollAsyncInput(MetricsReporter, WatermarksTracker, RuntimeSettings.AsyncInputPushLimit)) {
-                ContinueExecute(*resume);
-            }
-        }
     }
 
     void OnNewAsyncInputDataArrived(const IDqComputeActorAsyncInput::TEvNewAsyncInputDataArrived::TPtr& ev) {

diff --git a/ydb/library/yql/dq/actors/input_transforms/dq_input_transform_lookup.cpp b/ydb/library/yql/dq/actors/input_transforms/dq_input_transform_lookup.cpp
@@ -184,7 +184,7 @@ class TInputTransformStreamLookupBase
             }
         }
         finished = IsFinished();
-        return 0;
+        return AwaitingQueue.RowCount();
     }
 
     TMaybe<google::protobuf::Any> ExtraData() override {

diff --git a/ydb/library/yql/dq/actors/task_runner/task_runner_actor_local.cpp b/ydb/library/yql/dq/actors/task_runner/task_runner_actor_local.cpp
@@ -143,6 +143,18 @@ class TLocalTaskRunnerActor
                 return false;
             }
         }
+        for (const auto transformId: InputTransforms) {
+            const auto t = TaskRunner->GetInputTransform(transformId);
+            if (t) {
+                auto [_, transform] = *t;
+                if (!transform->Empty()) {
+                    return false;
+                }
+                if (transform->IsPending()) {
+                    return false;
+                }
+            }
+        }
         return true;
     }
 
@@ -436,6 +448,7 @@ class TLocalTaskRunnerActor
         for (auto i = 0; i != inputs.size(); ++i) {
             if (auto t = TaskRunner->GetInputTransform(i)) {
                 inputTransforms[i] = *t;
+                InputTransforms.emplace(i);
             }
         }
 
@@ -488,6 +501,7 @@ class TLocalTaskRunnerActor
     const TTxId TxId;
     const ui64 TaskId;
     THashSet<ui32> Inputs;
+    THashSet<ui32> InputTransforms;
     THashSet<ui32> Sources;
     TIntrusivePtr<NDq::IDqTaskRunner> TaskRunner;
     THashSet<ui32> InputChannelsWithDisabledCheckpoints;

diff --git a/ydb/library/yql/dq/runtime/dq_async_input.cpp b/ydb/library/yql/dq/runtime/dq_async_input.cpp
@@ -6,6 +6,7 @@ namespace NYql::NDq {
 class TDqAsyncInputBuffer : public TDqInputImpl<TDqAsyncInputBuffer, IDqAsyncInputBuffer> {
     using TBaseImpl = TDqInputImpl<TDqAsyncInputBuffer, IDqAsyncInputBuffer>;
     friend TBaseImpl;
+    bool Pending = false;
 public:
     TDqAsyncInputBufferStats PushStats;
     TDqInputStats PopStats;
@@ -32,7 +33,7 @@ class TDqAsyncInputBuffer : public TDqInputImpl<TDqAsyncInputBuffer, IDqAsyncInp
     }
 
     void Push(NKikimr::NMiniKQL::TUnboxedValueBatch&& batch, i64 space) override {
-        Y_ABORT_UNLESS(!batch.empty() || !space);
+        Pending = space != 0;
         if (!batch.empty()) {
             AddBatch(std::move(batch), space);
         }
@@ -41,6 +42,10 @@ class TDqAsyncInputBuffer : public TDqInputImpl<TDqAsyncInputBuffer, IDqAsyncInp
     virtual void Push(TDqSerializedBatch&&, i64) override {
         YQL_ENSURE(!"Unimplemented");
     }
+
+    bool IsPending() const override {
+        return Pending;
+    }
 };
 
 IDqAsyncInputBuffer::TPtr CreateDqAsyncInputBuffer(

diff --git a/ydb/library/yql/dq/runtime/dq_async_input.h b/ydb/library/yql/dq/runtime/dq_async_input.h
@@ -23,6 +23,8 @@ class IDqAsyncInputBuffer : public IDqInput {
     virtual void Push(TDqSerializedBatch&& batch, i64 space) = 0;
 
     virtual void Finish() = 0;
+
+    virtual bool IsPending() const { return false; };
 };
 
 IDqAsyncInputBuffer::TPtr CreateDqAsyncInputBuffer(ui64 inputIndex, const TString& type, NKikimr::NMiniKQL::TType* inputType, ui64 maxBufferBytes,