Merge pull request #13684 from vespa-engine/arnej/first-1000-no-twophase

the very first documents added to HNSW index should use single-phase …

5 files changed, 18 insertions, 3 deletions

diff --git a/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp b/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
index 7dc0efc106d..cd989c03b4e 100644
--- a/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
+++ b/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
@@ -82,7 +82,7 @@ public:
         level_generator = generator.get();
         index = std::make_unique<HnswIndex>(vectors, std::make_unique<FloatSqEuclideanDistance>(),
                                             std::move(generator),
-                                            HnswIndex::Config(5, 2, 10, heuristic_select_neighbors));
+                                            HnswIndex::Config(5, 2, 10, 0, heuristic_select_neighbors));
     }
     void add_document(uint32_t docid, uint32_t max_level = 0) {
         level_generator->level = max_level;
diff --git a/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp b/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp
index 4dec9550f6f..1e10d94bb18 100644
--- a/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp
+++ b/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp
@@ -234,7 +234,7 @@ public:
         uint32_t m = 16;
         index = std::make_unique<HnswIndex>(vectors, std::make_unique<FloatSqEuclideanDistance>(),
                                             std::make_unique<InvLogLevelGenerator>(m),
-                                            HnswIndex::Config(2*m, m, 200, true));
+                                            HnswIndex::Config(2*m, m, 200, 10, true));
     }
     size_t get_rnd(size_t size) {
         return rng.nextUniform() * size;
diff --git a/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp b/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp
index 067280e9a23..0bb6f339455 100644
--- a/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp
@@ -36,6 +36,7 @@ DefaultNearestNeighborIndexFactory::make(const DocVectorAccess& vectors,
     HnswIndex::Config cfg(m * 2,
                           m,
                           params.neighbors_to_explore_at_insert(),
+                          10000,
                           true);
     return std::make_unique<HnswIndex>(vectors,
                                        make_distance_function(params.distance_metric(), cell_type),
diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp b/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
index 540a7a86104..5c36f7bcb37 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
@@ -373,6 +373,12 @@ HnswIndex::prepare_add_document(uint32_t docid,
             TypedCells vector,
             vespalib::GenerationHandler::Guard read_guard) const
 {
+    uint32_t max_nodes = _graph.node_refs.size();
+    if (max_nodes < _cfg.min_size_before_two_phase()) {
+        // the first documents added will do all work in write thread
+        // to ensure they are linked together:
+        return std::unique_ptr<PrepareResult>();
+    }
     PreparedAddDoc op = internal_prepare_add(docid, vector);
     (void) read_guard; // must keep guard until this point
     return std::make_unique<PreparedAddDoc>(std::move(op));
@@ -385,7 +391,11 @@ HnswIndex::complete_add_document(uint32_t docid, std::unique_ptr<PrepareResult>
     if (prepared && (prepared->docid == docid)) {
         internal_complete_add(docid, *prepared);
     } else {
-        LOG(warning, "complete_add_document called with invalid prepare_result");
+        // we expect this for the first documents added, so no warning for them
+        if (_graph.node_refs.size() > 1.25 * _cfg.min_size_before_two_phase()) {
+            LOG(warning, "complete_add_document(%u) called with invalid prepare_result %s/%u",
+                docid, (prepared ? "valid ptr" : "nullptr"), (prepared ? prepared->docid : 0u));
+        }
         // fallback to normal add
         add_document(docid);
     }
diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index.h b/searchlib/src/vespa/searchlib/tensor/hnsw_index.h
index ab3eced8fdc..c237e3b8fcf 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_index.h
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_index.h
@@ -39,21 +39,25 @@ public:
         uint32_t _max_links_at_level_0;
         uint32_t _max_links_on_inserts;
         uint32_t _neighbors_to_explore_at_construction;
+        uint32_t _min_size_before_two_phase;
         bool _heuristic_select_neighbors;
 
     public:
         Config(uint32_t max_links_at_level_0_in,
                uint32_t max_links_on_inserts_in,
                uint32_t neighbors_to_explore_at_construction_in,
+               uint32_t min_size_before_two_phase_in,
                bool heuristic_select_neighbors_in)
             : _max_links_at_level_0(max_links_at_level_0_in),
               _max_links_on_inserts(max_links_on_inserts_in),
               _neighbors_to_explore_at_construction(neighbors_to_explore_at_construction_in),
+              _min_size_before_two_phase(min_size_before_two_phase_in),
               _heuristic_select_neighbors(heuristic_select_neighbors_in)
         {}
         uint32_t max_links_at_level_0() const { return _max_links_at_level_0; }
         uint32_t max_links_on_inserts() const { return _max_links_on_inserts; }
         uint32_t neighbors_to_explore_at_construction() const { return _neighbors_to_explore_at_construction; }
+        uint32_t min_size_before_two_phase() const { return _min_size_before_two_phase; }
         bool heuristic_select_neighbors() const { return _heuristic_select_neighbors; }
     };