SERVER-59044 Bailout early if a time series collection is empty when building a TrialStage

(cherry picked from commit bb6ec30bcbbb7b16f9f1c80cc3c9a99efcd5687c)

1 files changed, 10 insertions, 0 deletions

diff --git a/src/mongo/db/pipeline/pipeline_d.cpp b/src/mongo/db/pipeline/pipeline_d.cpp
index dca31cd55b0..28c870d99d8 100644
--- a/src/mongo/db/pipeline/pipeline_d.cpp
+++ b/src/mongo/db/pipeline/pipeline_d.cpp
@@ -185,6 +185,16 @@ createRandomCursorExecutor(const CollectionPtr& coll,
                                             minAdvancedToWorkRatio);
         trialStage = static_cast<TrialStage*>(root.get());
     } else if (expCtx->ns.isTimeseriesBucketsCollection()) {
+        // We can't take ARHASH optimization path for a direct $sample on the system.buckets
+        // collection because data is in compressed form. If we did have a direct $sample on the
+        // system.buckets collection, then the 'bucketUnpacker' would not be set up properly. We
+        // also should bail out early if a $sample is made against a time series collection that is
+        // empty. If we don't the 'minAdvancedToWorkRatio' can be nan/-nan depending on the
+        // architecture.
+        if (!(bucketUnpacker && numRecords)) {
+            return std::pair{nullptr, false};
+        }
+
         // Use a 'TrialStage' to run a trial between 'SampleFromTimeseriesBucket' and
         // 'UnpackTimeseriesBucket' with $sample left in the pipeline in-place. If the buckets are
         // not sufficiently full, or the 'SampleFromTimeseriesBucket' plan draws too many