SERVER-60672 Simpler pushdown when timeseries collection has no mixed-schema buckets

4 files changed, 56 insertions, 4 deletions

diff --git a/src/mongo/db/views/SConscript b/src/mongo/db/views/SConscript
index 5ca16d561ed..38d2204fce9 100644
--- a/src/mongo/db/views/SConscript
+++ b/src/mongo/db/views/SConscript
@@ -27,6 +27,7 @@ env.Library(
     ],
     LIBDEPS=[
         '$BUILD_DIR/mongo/base',
+        '$BUILD_DIR/mongo/db/catalog/collection',
         '$BUILD_DIR/mongo/db/pipeline/aggregation',
         '$BUILD_DIR/mongo/db/query/collation/collator_factory_interface',
         '$BUILD_DIR/mongo/db/repl/repl_coordinator_interface',
diff --git a/src/mongo/db/views/resolved_view.cpp b/src/mongo/db/views/resolved_view.cpp
index 9c754c559e7..c0435b8e924 100644
--- a/src/mongo/db/views/resolved_view.cpp
+++ b/src/mongo/db/views/resolved_view.cpp
@@ -72,15 +72,30 @@ ResolvedView ResolvedView::fromBSON(const BSONObj& commandResponseObj) {
         collationSpec = collationElt.embeddedObject().getOwned();
     }
 
+
+    boost::optional<bool> mixedSchema = boost::none;
+    if (auto mixedSchemaElem = viewDef[kTimeseriesMayContainMixedData]) {
+        uassert(6067204,
+                str::stream() << "view definition must have " << kTimeseriesMayContainMixedData
+                              << " of type bool or no such field",
+                mixedSchemaElem.type() == BSONType::Bool);
+
+        mixedSchema = boost::optional<bool>(mixedSchemaElem.boolean());
+    }
+
     return {NamespaceString(viewDef["ns"].valueStringData()),
             std::move(pipeline),
-            std::move(collationSpec)};
+            std::move(collationSpec),
+            std::move(mixedSchema)};
 }
 
 void ResolvedView::serialize(BSONObjBuilder* builder) const {
     BSONObjBuilder subObj(builder->subobjStart("resolvedView"));
     subObj.append("ns", _namespace.ns());
     subObj.append("pipeline", _pipeline);
+    // Only serialize if it doesn't contain mixed data.
+    if ((_timeseriesMayContainMixedData && !(*_timeseriesMayContainMixedData)))
+        subObj.append(kTimeseriesMayContainMixedData, *_timeseriesMayContainMixedData);
     if (!_defaultCollation.isEmpty()) {
         subObj.append("collation", _defaultCollation);
     }
@@ -125,6 +140,21 @@ AggregateCommandRequest ResolvedView::asExpandedViewAggregation(
         }
         resolvedPipeline[1] =
             BSON(DocumentSourceInternalConvertBucketIndexStats::kStageName << builder.obj());
+    } else if (resolvedPipeline.size() >= 1 &&
+               resolvedPipeline[0][DocumentSourceInternalUnpackBucket::kStageNameInternal] &&
+               serverGlobalParams.featureCompatibility.isGreaterThanOrEqualTo(
+                   multiversion::FeatureCompatibilityVersion::kVersion_5_2)) {
+        auto unpackStage = resolvedPipeline[0];
+
+        BSONObjBuilder builder;
+        for (const auto& elem :
+             unpackStage[DocumentSourceInternalUnpackBucket::kStageNameInternal].Obj()) {
+            builder.append(elem);
+        }
+        builder.append(DocumentSourceInternalUnpackBucket::kAssumeNoMixedSchemaData,
+                       ((_timeseriesMayContainMixedData && !(*_timeseriesMayContainMixedData))));
+        resolvedPipeline[0] =
+            BSON(DocumentSourceInternalUnpackBucket::kStageNameInternal << builder.obj());
     }
 
     AggregateCommandRequest expandedRequest{_namespace, resolvedPipeline};
diff --git a/src/mongo/db/views/resolved_view.h b/src/mongo/db/views/resolved_view.h
index 19cfefe82f9..d19642c8dd5 100644
--- a/src/mongo/db/views/resolved_view.h
+++ b/src/mongo/db/views/resolved_view.h
@@ -46,10 +46,12 @@ class ResolvedView final : public ErrorExtraInfo {
 public:
     ResolvedView(const NamespaceString& collectionNs,
                  std::vector<BSONObj> pipeline,
-                 BSONObj defaultCollation)
+                 BSONObj defaultCollation,
+                 boost::optional<bool> timeseriesMayContainMixedData = boost::none)
         : _namespace(collectionNs),
           _pipeline(std::move(pipeline)),
-          _defaultCollation(std::move(defaultCollation)) {}
+          _defaultCollation(std::move(defaultCollation)),
+          _timeseriesMayContainMixedData(timeseriesMayContainMixedData) {}
 
     static ResolvedView fromBSON(const BSONObj& commandResponseObj);
 
@@ -74,6 +76,7 @@ public:
 
     // ErrorExtraInfo API
     static constexpr auto code = ErrorCodes::CommandOnShardedViewNotSupportedOnMongod;
+    static constexpr StringData kTimeseriesMayContainMixedData = "timeseriesMayContainMixedData"_sd;
     void serialize(BSONObjBuilder* bob) const final;
     static std::shared_ptr<const ErrorExtraInfo> parse(const BSONObj&);
 
@@ -88,6 +91,8 @@ private:
     // that operations on the view which do not specify a collation inherit the default. Operations
     // on the view which specify any other collation fail with a user error.
     BSONObj _defaultCollation;
+
+    boost::optional<bool> _timeseriesMayContainMixedData;
 };
 
 }  // namespace mongo
diff --git a/src/mongo/db/views/view_catalog.cpp b/src/mongo/db/views/view_catalog.cpp
index 84e0f7b46a8..b043eb5a03d 100644
--- a/src/mongo/db/views/view_catalog.cpp
+++ b/src/mongo/db/views/view_catalog.cpp
@@ -740,6 +740,7 @@ StatusWith<ResolvedView> ViewCatalog::resolveView(
         std::vector<NamespaceString> dependencyChain{nss};
 
         int depth = 0;
+        boost::optional<bool> mixedData = boost::none;
         for (; depth < ViewGraph::kMaxViewDepth; depth++) {
             auto view =
                 _lookup(opCtx, *resolvedNss, ViewCatalogLookupBehavior::kValidateDurableViews);
@@ -761,10 +762,25 @@ StatusWith<ResolvedView> ViewCatalog::resolveView(
                 return StatusWith<ResolvedView>(
                     {*resolvedNss,
                      std::move(resolvedPipeline),
-                     collation ? std::move(collation.get()) : CollationSpec::kSimpleSpec});
+                     collation ? std::move(collation.get()) : CollationSpec::kSimpleSpec,
+                     mixedData});
             }
 
             resolvedNss = &view->viewOn();
+
+            if (view->timeseries()) {
+                auto tsCollection =
+                    CollectionCatalog::get(opCtx)->lookupCollectionByNamespace(opCtx, *resolvedNss);
+                tassert(6067201,
+                        str::stream() << "expected time-series buckets collection " << *resolvedNss
+                                      << " to exist",
+
+                        tsCollection);
+                mixedData = tsCollection
+                    ? tsCollection->getTimeseriesBucketsMayHaveMixedSchemaData()
+                    : false;
+            }
+
             dependencyChain.push_back(*resolvedNss);
             if (!collation) {
                 if (timeSeriesCollator) {