Merge branch 'master' into Implement-automatic-index-creation

config/flags.yaml

@@ -119,6 +119,14 @@ modifications:
     value: "false"
     override: true
 
+  - name: "storage_enable_automatic_label_index_creation"
+    value: "false"
+    override: true
+
+  - name: "storage_enable_automatic_edge_type_index_creation"
+    value: "false"
+    override: true
+
   - name: "query_callable_mappings_path"
     value: "/etc/memgraph/apoc_compatibility_mappings.json"
     override: true

src/flags/general.cpp

@@ -131,6 +131,14 @@ DEFINE_uint64(storage_recovery_thread_count,
 DEFINE_bool(storage_enable_schema_metadata, false,
             "Controls whether metadata should be collected about the resident labels and edge types.");
 
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DEFINE_bool(storage_enable_automatic_label_index_creation, false,
+            "Controls whether label indexes on vertices should be created automatically.");
+
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DEFINE_bool(storage_enable_automatic_edge_type_index_creation, false,
+            "Controls whether edge-type indexes on relationships should be created automatically.");
+
 // NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 DEFINE_bool(storage_delta_on_identical_property_update, true,
             "Controls whether updating a property with the same value should create a delta object.");

src/flags/general.hpp

@@ -85,6 +85,10 @@ DECLARE_uint64(storage_recovery_thread_count);
 // NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 DECLARE_bool(storage_enable_schema_metadata);
 // NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DECLARE_bool(storage_enable_automatic_label_index_creation);
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DECLARE_bool(storage_enable_automatic_edge_type_index_creation);
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 DECLARE_bool(storage_delta_on_identical_property_update);
 
 // NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)

src/memgraph.cpp

@@ -333,8 +333,18 @@ int main(int argc, char **argv) {
                .wal_directory = FLAGS_data_directory + "/rocksdb_wal"},
       .salient.items = {.properties_on_edges = FLAGS_storage_properties_on_edges,
                         .enable_schema_metadata = FLAGS_storage_enable_schema_metadata,
+                        .enable_label_index_auto_creation = FLAGS_storage_enable_automatic_label_index_creation,
+                        .enable_edge_type_index_auto_creation =
+                            FLAGS_storage_properties_on_edges ? FLAGS_storage_enable_automatic_edge_type_index_creation
+                                                              : false,
                         .delta_on_identical_property_update = FLAGS_storage_delta_on_identical_property_update},
       .salient.storage_mode = memgraph::flags::ParseStorageMode()};
+  if (db_config.salient.items.enable_edge_type_index_auto_creation && !db_config.salient.items.properties_on_edges) {
+    spdlog::warn(
+        "Automatic index creation on edge-types has been set but properties on edges are disabled. This will "
+        "implicitly disallow automatic edge-type index creation. If you wish to use automatic edge-type index "
+        "creation, enable properties on edges as well.");
+  }
   spdlog::info("config recover on startup {}, flags {} {}", db_config.durability.recover_on_startup,
                FLAGS_storage_recover_on_startup, FLAGS_data_recovery_on_startup);
   memgraph::utils::Scheduler jemalloc_purge_scheduler;

src/storage/v2/config.hpp

@@ -37,6 +37,8 @@ struct SalientConfig {
   struct Items {
     bool properties_on_edges{true};
     bool enable_schema_metadata{false};
+    bool enable_label_index_auto_creation{false};
+    bool enable_edge_type_index_auto_creation{false};
     bool delta_on_identical_property_update{true};
     friend bool operator==(const Items &lrh, const Items &rhs) = default;
   } items;
@@ -45,13 +47,19 @@ struct SalientConfig {
 };
 
 inline void to_json(nlohmann::json &data, SalientConfig::Items const &items) {
-  data = nlohmann::json{{"properties_on_edges", items.properties_on_edges},
-                        {"enable_schema_metadata", items.enable_schema_metadata}};
+  data = nlohmann::json{
+      {"properties_on_edges", items.properties_on_edges},
+      {"enable_schema_metadata", items.enable_schema_metadata},
+      {"enable_label_index_auto_creation", items.enable_label_index_auto_creation},
+      {"enable_edge_type_index_auto_creation", items.enable_edge_type_index_auto_creation},
+  };
 }
 
 inline void from_json(const nlohmann::json &data, SalientConfig::Items &items) {
   data.at("properties_on_edges").get_to(items.properties_on_edges);
   data.at("enable_schema_metadata").get_to(items.enable_schema_metadata);
+  data.at("enable_label_index_auto_creation").get_to(items.enable_label_index_auto_creation);
+  data.at("enable_edge_type_index_auto_creation").get_to(items.enable_edge_type_index_auto_creation);
 }
 
 inline void to_json(nlohmann::json &data, SalientConfig const &config) {

src/storage/v2/disk/storage.cpp

@@ -1927,8 +1927,11 @@ void DiskStorage::DiskAccessor::FinalizeTransaction() {
   }
 }
 
-utils::BasicResult<StorageIndexDefinitionError, void> DiskStorage::DiskAccessor::CreateIndex(LabelId label) {
+utils::BasicResult<StorageIndexDefinitionError, void> DiskStorage::DiskAccessor::CreateIndex(
+    LabelId label, bool unique_access_needed) {
+  if (unique_access_needed) {
     MG_ASSERT(unique_guard_.owns_lock(), "Create index requires unique access to the storage!");
+  }
   auto *on_disk = static_cast<DiskStorage *>(storage_);
   auto *disk_label_index = static_cast<DiskLabelIndex *>(on_disk->indices_.label_index_.get());
   if (!disk_label_index->CreateIndex(label, on_disk->SerializeVerticesForLabelIndex(label))) {
@@ -1956,7 +1959,8 @@ utils::BasicResult<StorageIndexDefinitionError, void> DiskStorage::DiskAccessor:
   return {};
 }
 
-utils::BasicResult<StorageIndexDefinitionError, void> DiskStorage::DiskAccessor::CreateIndex(EdgeTypeId /*edge_type*/) {
+utils::BasicResult<StorageIndexDefinitionError, void> DiskStorage::DiskAccessor::CreateIndex(
+    EdgeTypeId /*edge_type*/, bool /*unique_access_needed*/) {
   throw utils::NotYetImplemented(
       "Edge-type index related operations are not yet supported using on-disk storage mode.");
 }

src/storage/v2/disk/storage.hpp

@@ -160,11 +160,13 @@ class DiskStorage final : public Storage {
 
     void FinalizeTransaction() override;
 
-    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label) override;
+    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label,
+                                                                      bool unique_access_needed = true) override;
 
     utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label, PropertyId property) override;
 
-    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type) override;
+    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type,
+                                                                      bool unique_access_needed = true) override;
 
     utils::BasicResult<StorageIndexDefinitionError, void> DropIndex(LabelId label) override;
 

src/storage/v2/inmemory/storage.cpp

@@ -317,6 +317,11 @@ Result<EdgeAccessor> InMemoryStorage::InMemoryAccessor::CreateEdge(VertexAccesso
     guard_from.lock();
   }
 
+  if (storage_->config_.salient.items.enable_edge_type_index_auto_creation &&
+      !storage_->indices_.edge_type_index_->IndexExists(edge_type)) {
+    storage_->edge_types_to_auto_index_->emplace(edge_type);
+  }
+
   if (!PrepareForWrite(&transaction_, from_vertex)) return Error::SERIALIZATION_ERROR;
   if (from_vertex->deleted) return Error::DELETED_OBJECT;
 
@@ -807,6 +812,25 @@ utils::BasicResult<StorageManipulationError, void> InMemoryStorage::InMemoryAcce
 
     {
       std::unique_lock<utils::SpinLock> engine_guard(storage_->engine_lock_);
+
+      if (storage_->config_.salient.items.enable_label_index_auto_creation) {
+        storage_->labels_to_auto_index_.WithLock([&](auto &label_indices) {
+          for (const auto label : label_indices) {
+            CreateIndex(label, false);
+          }
+          label_indices.clear();
+        });
+      }
+
+      if (storage_->config_.salient.items.enable_edge_type_index_auto_creation) {
+        storage_->edge_types_to_auto_index_.WithLock([&](auto &edge_type_indices) {
+          for (const auto edge_type : edge_type_indices) {
+            CreateIndex(edge_type, false);
+          }
+          edge_type_indices.clear();
+        });
+      }
+
       auto *mem_unique_constraints =
           static_cast<InMemoryUniqueConstraints *>(storage_->constraints_.unique_constraints_.get());
       commit_timestamp_.emplace(mem_storage->CommitTimestamp(reparg.desired_commit_timestamp));
@@ -1262,8 +1286,11 @@ void InMemoryStorage::InMemoryAccessor::FinalizeTransaction() {
   }
 }
 
-utils::BasicResult<StorageIndexDefinitionError, void> InMemoryStorage::InMemoryAccessor::CreateIndex(LabelId label) {
+utils::BasicResult<StorageIndexDefinitionError, void> InMemoryStorage::InMemoryAccessor::CreateIndex(
+    LabelId label, bool unique_access_needed) {
+  if (unique_access_needed) {
     MG_ASSERT(unique_guard_.owns_lock(), "Creating label index requires a unique access to the storage!");
+  }
   auto *in_memory = static_cast<InMemoryStorage *>(storage_);
   auto *mem_label_index = static_cast<InMemoryLabelIndex *>(in_memory->indices_.label_index_.get());
   if (!mem_label_index->CreateIndex(label, in_memory->vertices_.access(), std::nullopt)) {
@@ -1291,8 +1318,10 @@ utils::BasicResult<StorageIndexDefinitionError, void> InMemoryStorage::InMemoryA
 }
 
 utils::BasicResult<StorageIndexDefinitionError, void> InMemoryStorage::InMemoryAccessor::CreateIndex(
-    EdgeTypeId edge_type) {
+    EdgeTypeId edge_type, bool unique_access_needed) {
+  if (unique_access_needed) {
     MG_ASSERT(unique_guard_.owns_lock(), "Create index requires a unique access to the storage!");
+  }
   auto *in_memory = static_cast<InMemoryStorage *>(storage_);
   auto *mem_edge_type_index = static_cast<InMemoryEdgeTypeIndex *>(in_memory->indices_.edge_type_index_.get());
   if (!mem_edge_type_index->CreateIndex(edge_type, in_memory->vertices_.access())) {

src/storage/v2/inmemory/storage.hpp

@@ -241,7 +241,8 @@ class InMemoryStorage final : public Storage {
     /// * `IndexDefinitionError`: the index already exists.
     /// * `ReplicationError`:  there is at least one SYNC replica that has not confirmed receiving the transaction.
     /// @throw std::bad_alloc
-    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label) override;
+    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label,
+                                                                      bool unique_access_needed = true) override;
 
     /// Create an index.
     /// Returns void if the index has been created.
@@ -257,7 +258,8 @@ class InMemoryStorage final : public Storage {
     /// * `ReplicationError`:  there is at least one SYNC replica that has not confirmed receiving the transaction.
     /// * `IndexDefinitionError`: the index already exists.
     /// @throw std::bad_alloc
-    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type) override;
+    utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type,
+                                                                      bool unique_access_needed = true) override;
 
     /// Drop an existing index.
     /// Returns void if the index has been dropped.

src/storage/v2/storage.hpp

@@ -299,11 +299,13 @@ class Storage {
 
     std::vector<EdgeTypeId> ListAllPossiblyPresentEdgeTypes() const;
 
-    virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label) = 0;
+    virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label,
+                                                                              bool unique_access_needed = true) = 0;
 
     virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(LabelId label, PropertyId property) = 0;
 
-    virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type) = 0;
+    virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateIndex(EdgeTypeId edge_type,
+                                                                              bool unique_access_needed = true) = 0;
 
     virtual utils::BasicResult<StorageIndexDefinitionError, void> DropIndex(LabelId label) = 0;
 
@@ -466,6 +468,12 @@ class Storage {
   utils::SynchronizedMetaDataStore<LabelId> stored_node_labels_;
   utils::SynchronizedMetaDataStore<EdgeTypeId> stored_edge_types_;
 
+  // Sets that hold onto labels and edge-types that have to be created.
+  // Used only if the label/edge-type autocreation flags are enabled,
+  // does not affect index creation otherwise.
+  utils::Synchronized<std::set<LabelId>, utils::SpinLock> labels_to_auto_index_;
+  utils::Synchronized<std::set<EdgeTypeId>, utils::SpinLock> edge_types_to_auto_index_;
+
   std::atomic<uint64_t> vertex_id_{0};
   std::atomic<uint64_t> edge_id_{0};
 };

src/storage/v2/vertex_accessor.cpp

@@ -118,6 +118,11 @@ Result<bool> VertexAccessor::AddLabel(LabelId label) {
     storage_->stored_node_labels_.try_insert(label);
   }
 
+  if (storage_->config_.salient.items.enable_label_index_auto_creation &&
+      !storage_->indices_.label_index_->IndexExists(label)) {
+    storage_->labels_to_auto_index_->emplace(label);
+  }
+
   /// TODO: some by pointers, some by reference => not good, make it better
   storage_->constraints_.unique_constraints_->UpdateOnAddLabel(label, *vertex_, transaction_->start_timestamp);
   if (transaction_->constraint_verification_info) transaction_->constraint_verification_info->AddedLabel(vertex_);

tests/e2e/CMakeLists.txt

@@ -78,6 +78,7 @@ add_subdirectory(query_planning)
 add_subdirectory(awesome_functions)
 add_subdirectory(high_availability)
 add_subdirectory(concurrency)
+add_subdirectory(index_auto_creation)
 
 add_subdirectory(replication_experimental)
 

tests/e2e/configuration/default_config.py

@@ -124,6 +124,16 @@ startup_config_dict = {
         "false",
         "Controls whether metadata should be collected about the resident labels and edge types.",
     ),
+    "storage_enable_automatic_label_index_creation": (
+        "false",
+        "false",
+        "Controls whether label indexes on vertices should be created automatically.",
+    ),
+    "storage_enable_automatic_edge_type_index_creation": (
+        "false",
+        "false",
+        "Controls whether edge-type indexes on relationships should be created automatically.",
+    ),
     "password_encryption_algorithm": ("bcrypt", "bcrypt", "The password encryption algorithm used for authentication."),
     "pulsar_service_url": ("", "", "Default URL used while connecting to Pulsar brokers."),
     "query_execution_timeout_sec": (

tests/e2e/index_auto_creation/CMakeLists.txt

@@ -0,0 +1,6 @@
+function(copy_auto_index_queries_e2e_python_files FILE_NAME)
+    copy_e2e_python_files(index_auto_creation ${FILE_NAME})
+endfunction()
+
+copy_auto_index_queries_e2e_python_files(common.py)
+copy_auto_index_queries_e2e_python_files(index_auto_creation.py)

tests/e2e/index_auto_creation/common.py

@@ -0,0 +1,27 @@
+# Copyright 2024 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import typing
+
+import mgclient
+import pytest
+
+
+@pytest.fixture(scope="module")
+def cursor(**kwargs) -> mgclient.Connection:
+    connection = mgclient.connect(host="localhost", port=7687, **kwargs)
+    connection.autocommit = True
+    return connection.cursor()
+
+
+def execute_and_fetch_all(cursor: mgclient.Cursor, query: str, params: dict = dict()) -> typing.List[tuple]:
+    cursor.execute(query, params)
+    return cursor.fetchall()

tests/e2e/index_auto_creation/index_auto_creation.py

@@ -0,0 +1,318 @@
+# Copyright 2024 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import sys
+
+import pytest
+from common import cursor, execute_and_fetch_all
+
+
+# Helper functions
+def get_index_stats(cursor):
+    return execute_and_fetch_all(cursor, "SHOW INDEX INFO")
+
+
+def index_exists(indices, index_name):
+    for index in indices:
+        if index[1] == index_name:
+            return True
+
+    return False
+
+
+def index_count_is(indices, index_name, count):
+    return count == index_count(indices, index_name)
+
+
+def number_of_index_structures_are(indices, predicted_size):
+    return len(indices) == predicted_size
+
+
+def index_count(indices, index_name):
+    for index in indices:
+        if index[1] == index_name:
+            return index[3]
+
+    return 0
+
+
+#####################
+# Label index tests #
+#####################
+
+
+def test_auto_create_single_label_index(cursor):
+    label = "SOMELABEL"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label})")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 1)
+    assert index_exists(index_stats, label)
+    assert index_count_is(index_stats, label, 1)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_multiple_label_index(cursor):
+    label1 = "SOMELABEL1"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label1})")
+
+    label2 = "SOMELABEL2"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label2})")
+
+    label3 = "SOMELABEL3"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label3})")
+
+    index_stats = get_index_stats(cursor)
+
+    assert number_of_index_structures_are(index_stats, 3)
+    assert index_exists(index_stats, label1)
+    assert index_exists(index_stats, label2)
+    assert index_exists(index_stats, label3)
+    assert index_count_is(index_stats, label1, 1)
+    assert index_count_is(index_stats, label2, 1)
+    assert index_count_is(index_stats, label3, 1)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label1}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 2)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label2}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 1)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label3}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_single_label_index_with_multiple_entries(cursor):
+    label = "SOMELABEL"
+    for _ in range(100):
+        execute_and_fetch_all(cursor, f"CREATE (n:{label})")
+    index_stats = get_index_stats(cursor)
+
+    print(len(index_stats[0]))
+
+    assert number_of_index_structures_are(index_stats, 1)
+    assert index_exists(index_stats, label)
+    assert index_count_is(index_stats, label, 100)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_multiple_label_index_with_multiple_entries(cursor):
+    label1 = "SOMELABEL1"
+    for _ in range(120):
+        execute_and_fetch_all(cursor, f"CREATE (n:{label1})")
+
+    label2 = "SOMELABEL2"
+    for _ in range(100):
+        execute_and_fetch_all(cursor, f"CREATE (n:{label2})")
+
+    label3 = "SOMELABEL3"
+    for _ in range(80):
+        execute_and_fetch_all(cursor, f"CREATE (n:{label3})")
+
+    index_stats = get_index_stats(cursor)
+
+    print(index_stats)
+
+    assert number_of_index_structures_are(index_stats, 3)
+    assert index_exists(index_stats, label1)
+    assert index_exists(index_stats, label2)
+    assert index_exists(index_stats, label3)
+    assert index_count_is(index_stats, label1, 120)
+    assert index_count_is(index_stats, label2, 100)
+    assert index_count_is(index_stats, label3, 80)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label1}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 2)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label2}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 1)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label3}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+#########################
+# Edge-type index tests #
+#########################
+
+
+def test_auto_create_single_edge_type_index(cursor):
+    label_from = "LABEL_FROM"
+    label_to = "LABEL_TO"
+    edge_type = "SOMEEDGETYPE"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_from})")
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_to})")
+    execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type}]->(m)")
+
+    index_stats = get_index_stats(cursor)
+    assert index_exists(index_stats, label_from)
+    assert index_exists(index_stats, label_to)
+    assert index_exists(index_stats, edge_type)
+    assert index_count_is(index_stats, label_from, 1)
+    assert index_count_is(index_stats, label_to, 1)
+    assert index_count_is(index_stats, edge_type, 1)
+    assert number_of_index_structures_are(index_stats, 3)  # 2 label + 1 edge-type
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_from}")
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_to}")
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_multiple_edge_type_index(cursor):
+    label_from = "LABEL_FROM"
+    label_to = "LABEL_TO"
+    edge_type1 = "SOMEEDGETYPE1"
+    edge_type2 = "SOMEEDGETYPE2"
+    edge_type3 = "SOMEEDGETYPE3"
+
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_from})")
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_to})")
+    execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type1}]->(m)")
+    execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type2}]->(m)")
+    execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type3}]->(m)")
+
+    index_stats = get_index_stats(cursor)
+
+    assert number_of_index_structures_are(index_stats, 5)  # 2 label + 3 edge-type
+    assert index_exists(index_stats, label_from)
+    assert index_exists(index_stats, label_to)
+    assert index_exists(index_stats, edge_type1)
+    assert index_exists(index_stats, edge_type2)
+    assert index_exists(index_stats, edge_type3)
+    assert index_count_is(index_stats, edge_type1, 1)
+    assert index_count_is(index_stats, edge_type2, 1)
+    assert index_count_is(index_stats, edge_type3, 1)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_from}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 4)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_to}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 3)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type1}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 2)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type2}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 1)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type3}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_single_edge_type_index_with_multiple_entries(cursor):
+    label_from = "LABEL_FROM"
+    label_to = "LABEL_TO"
+    edge_type = "SOMEEDGETYPE"
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_from})")
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_to})")
+    for _ in range(100):
+        execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type}]->(m)")
+
+    index_stats = get_index_stats(cursor)
+    assert index_exists(index_stats, label_from)
+    assert index_exists(index_stats, label_to)
+    assert index_exists(index_stats, edge_type)
+    assert number_of_index_structures_are(index_stats, 3)  # 2 label + 1 edge-type
+
+    assert index_count_is(index_stats, edge_type, 100)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_from}")
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_to}")
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+def test_auto_create_multiple_edge_type_index_with_multiple_entries(cursor):
+    label_from = "LABEL_FROM"
+    label_to = "LABEL_TO"
+    edge_type1 = "SOMEEDGETYPE1"
+    edge_type2 = "SOMEEDGETYPE2"
+    edge_type3 = "SOMEEDGETYPE3"
+
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_from})")
+    execute_and_fetch_all(cursor, f"CREATE (n:{label_to})")
+    for _ in range(120):
+        execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type1}]->(m)")
+    for _ in range(100):
+        execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type2}]->(m)")
+    for _ in range(80):
+        execute_and_fetch_all(cursor, f"MATCH (n:{label_from}), (m:{label_to}) CREATE (n)-[:{edge_type3}]->(m)")
+
+    index_stats = get_index_stats(cursor)
+
+    assert number_of_index_structures_are(index_stats, 5)  # 2 label + 3 edge-type
+    assert index_exists(index_stats, label_from)
+    assert index_exists(index_stats, label_to)
+    assert index_exists(index_stats, edge_type1)
+    assert index_exists(index_stats, edge_type2)
+    assert index_exists(index_stats, edge_type3)
+    assert index_count_is(index_stats, edge_type1, 120)
+    assert index_count_is(index_stats, edge_type2, 100)
+    assert index_count_is(index_stats, edge_type3, 80)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_from}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 4)
+
+    execute_and_fetch_all(cursor, f"DROP INDEX ON :{label_to}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 3)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type1}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 2)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type2}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 1)
+
+    execute_and_fetch_all(cursor, f"DROP EDGE INDEX ON :{edge_type3}")
+    index_stats = get_index_stats(cursor)
+    assert number_of_index_structures_are(index_stats, 0)
+
+    execute_and_fetch_all(cursor, f"MATCH (n) DETACH DELETE n")
+
+
+if __name__ == "__main__":
+    sys.exit(pytest.main([__file__, "-rA"]))

tests/e2e/index_auto_creation/workloads.yaml

@@ -0,0 +1,13 @@
+index_auto_creation: &index_auto_creation
+  cluster:
+    main:
+      args: ["--bolt-port", "7687", "--log-level=TRACE", "--also-log-to-stderr", "--storage-properties-on-edges=TRUE", "--storage-enable-automatic-label-index-creation=TRUE", "--storage-enable-automatic-edge-type-index-creation=TRUE"]
+      log_file: "index_auto_creation.log"
+      setup_queries: []
+      validation_queries: []
+
+workloads:
+  - name: "Label index auto creation"
+    binary: "tests/e2e/pytest_runner.sh"
+    args: ["index_auto_creation/index_auto_creation.py"]
+    <<: *index_auto_creation