Merge pull request #27451 from vespa-engine/toregge/handle-removal-of-docid-with-empty-tensor-after-restart

Handle removal of docid with empty tensor in HNSW index after restart.

2 files changed, 17 insertions, 2 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 f59c16c76f9..b238044a67b 100644
--- a/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
+++ b/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
@@ -855,6 +855,17 @@ TEST_F(HnswMultiIndexTest, docid_with_empty_tensor_can_be_removed)
     this->remove_document(1);
 }
 
+TEST_F(HnswMultiIndexTest, docid_with_empty_tensor_can_be_removed_after_restart)
+{
+    this->init(false);
+    this->vectors.set(1, {});
+    this->add_document(1);
+    auto data = this->save_index();
+    this->init(false);
+    this->load_index(data);
+    this->remove_document(1);
+}
+
 TEST(LevelGeneratorTest, gives_various_levels)
 {
     InvLogLevelGenerator generator(4);
diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_nodeid_mapping.cpp b/searchlib/src/vespa/searchlib/tensor/hnsw_nodeid_mapping.cpp
index 8c169e63f9a..a908ebd7210 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_nodeid_mapping.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_nodeid_mapping.cpp
@@ -82,14 +82,18 @@ HnswNodeidMapping::allocate_ids(uint32_t docid, uint32_t subspaces)
 vespalib::ConstArrayRef<uint32_t>
 HnswNodeidMapping::get_ids(uint32_t docid) const
 {
-    assert(docid < _refs.size());
+    if (docid >= _refs.size()) {
+        return {};
+    }
     return _nodeids.get(_refs[docid]);
 }
 
 void
 HnswNodeidMapping::free_ids(uint32_t docid)
 {
-    assert(docid < _refs.size());
+    if (docid >= _refs.size()) {
+        return;
+    }
     EntryRef ref = _refs[docid];
     if (!ref.valid()) {
         return;