SERVER-73110 Each shard should check extended-range for timeseries sorting

6 files changed, 46 insertions, 2 deletions

diff --git a/src/mongo/db/catalog/virtual_collection_impl.h b/src/mongo/db/catalog/virtual_collection_impl.h
index 055c61283a2..4366d86ccd5 100644
--- a/src/mongo/db/catalog/virtual_collection_impl.h
+++ b/src/mongo/db/catalog/virtual_collection_impl.h
@@ -242,7 +242,8 @@ public:
     }
 
     bool getRequiresTimeseriesExtendedRangeSupport() const final {
-        unimplementedTasserted();
+        // A virtual collection is never a time-series collection, so it never requires
+        // extended-range support.
         return false;
     }
 
diff --git a/src/mongo/db/commands/run_aggregate.cpp b/src/mongo/db/commands/run_aggregate.cpp
index 8db9c787f7f..8868ce91a8b 100644
--- a/src/mongo/db/commands/run_aggregate.cpp
+++ b/src/mongo/db/commands/run_aggregate.cpp
@@ -974,6 +974,13 @@ Status runAggregate(OperationContext* opCtx,
                                            ? collections.getMainCollection()->getCollectionOptions()
                                            : boost::optional<CollectionOptions>(boost::none));
 
+        // If any involved collection contains extended-range data, set a flag which individual
+        // DocumentSource parsers can check.
+        collections.forEach([&](const CollectionPtr& coll) {
+            if (coll->getRequiresTimeseriesExtendedRangeSupport())
+                expCtx->setRequiresTimeseriesExtendedRangeSupport(true);
+        });
+
         expCtx->startExpressionCounters();
         auto pipeline = Pipeline::parse(request.getPipeline(), expCtx);
         curOp->beginQueryPlanningTimer();
diff --git a/src/mongo/db/pipeline/document_source_internal_unpack_bucket.cpp b/src/mongo/db/pipeline/document_source_internal_unpack_bucket.cpp
index bc4d86b2744..256804b978f 100644
--- a/src/mongo/db/pipeline/document_source_internal_unpack_bucket.cpp
+++ b/src/mongo/db/pipeline/document_source_internal_unpack_bucket.cpp
@@ -291,6 +291,11 @@ boost::intrusive_ptr<DocumentSource> DocumentSourceInternalUnpackBucket::createF
     // If neither "include" nor "exclude" is specified, the default is "exclude": [] and
     // if that's the case, no field will be added to 'bucketSpec.fieldSet' in the for-loop below.
     BucketSpec bucketSpec;
+    // Use extended-range support if any individual collection requires it, even if 'specElem'
+    // doesn't mention this flag.
+    if (expCtx->getRequiresTimeseriesExtendedRangeSupport()) {
+        bucketSpec.setUsesExtendedRange(true);
+    }
     auto hasIncludeExclude = false;
     auto hasTimeField = false;
     auto hasBucketMaxSpanSeconds = false;
@@ -504,6 +509,14 @@ void DocumentSourceInternalUnpackBucket::serializeToArray(
     if (_assumeNoMixedSchemaData)
         out.addField(kAssumeNoMixedSchemaData, Value(_assumeNoMixedSchemaData));
 
+    if (spec.usesExtendedRange()) {
+        // Include this flag so that 'explain' is more helpful.
+        // But this is not so useful for communicating from one process to another,
+        // because mongos and/or the primary shard don't know whether any other shard
+        // has extended-range data.
+        out.addField(kUsesExtendedRange, Value{true});
+    }
+
     if (!spec.computedMetaProjFields().empty())
         out.addField("computedMetaProjFields", Value{[&] {
                          std::vector<Value> compFields;
diff --git a/src/mongo/db/pipeline/expression_context.h b/src/mongo/db/pipeline/expression_context.h
index 3ec3554a336..a0f04003479 100644
--- a/src/mongo/db/pipeline/expression_context.h
+++ b/src/mongo/db/pipeline/expression_context.h
@@ -511,6 +511,15 @@ public:
         _gotTemporarilyUnavailableException = v;
     }
 
+    // Sets or clears a flag which tells DocumentSource parsers whether any involved Collection
+    // may contain extended-range dates.
+    void setRequiresTimeseriesExtendedRangeSupport(bool v) {
+        _requiresTimeseriesExtendedRangeSupport = v;
+    }
+    bool getRequiresTimeseriesExtendedRangeSupport() const {
+        return _requiresTimeseriesExtendedRangeSupport;
+    }
+
 protected:
     static const int kInterruptCheckPeriod = 128;
 
@@ -535,6 +544,8 @@ protected:
 
     bool _isCappedDelete = false;
 
+    bool _requiresTimeseriesExtendedRangeSupport = false;
+
 private:
     boost::optional<ExpressionCounters> _expressionCounters = boost::none;
     bool _gotTemporarilyUnavailableException = false;
diff --git a/src/mongo/db/query/multiple_collection_accessor.h b/src/mongo/db/query/multiple_collection_accessor.h
index bf8c0ad5fe9..04c766770bb 100644
--- a/src/mongo/db/query/multiple_collection_accessor.h
+++ b/src/mongo/db/query/multiple_collection_accessor.h
@@ -101,6 +101,17 @@ public:
         _secondaryColls.clear();
     }
 
+    void forEach(std::function<void(const CollectionPtr&)> func) const {
+        if (hasMainCollection()) {
+            func(getMainCollection());
+        }
+        for (const auto& [name, coll] : getSecondaryCollections()) {
+            if (coll) {
+                func(coll);
+            }
+        }
+    }
+
 private:
     const CollectionPtr* _mainColl{&CollectionPtr::null};
 
diff --git a/src/mongo/db/sorter/sorter.cpp b/src/mongo/db/sorter/sorter.cpp
index 411c85ffbdb..92680a15971 100644
--- a/src/mongo/db/sorter/sorter.cpp
+++ b/src/mongo/db/sorter/sorter.cpp
@@ -1468,7 +1468,8 @@ void BoundedSorter<Key, Value, Comparator, BoundMaker>::add(Key key, Value value
     invariant(!_done);
     // If a new value violates what we thought was our min bound, something has gone wrong.
     uassert(6369910,
-            "BoundedSorter input is too out-of-order.",
+            str::stream() << "BoundedSorter input is too out-of-order: with bound "
+                          << _min->toString() << ", did not expect input " << key.toString(),
             !_checkInput || !_min || compare(*_min, key) <= 0);
 
     // Each new item can potentially give us a tighter bound (a higher min).