Antalya 26.1 - Forward port of list objects cache #1040

programs/server/Server.cpp

@@ -87,6 +87,7 @@
 #include <Storages/System/attachSystemTables.h>
 #include <Storages/System/attachInformationSchemaTables.h>
 #include <Storages/Cache/registerRemoteFileMetadatas.h>
+#include <Storages/Cache/ObjectStorageListObjectsCache.h>
 #include <AggregateFunctions/registerAggregateFunctions.h>
 #include <Functions/UserDefined/IUserDefinedSQLObjectsStorage.h>
 #include <Functions/registerFunctions.h>
@@ -415,6 +416,9 @@ namespace ServerSetting
     extern const ServerSettingsUInt64 keeper_server_socket_send_timeout_sec;
     extern const ServerSettingsString hdfs_libhdfs3_conf;
     extern const ServerSettingsString config_file;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_ttl;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_size;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_max_entries;
 }
 
 namespace ErrorCodes
@@ -425,6 +429,9 @@ namespace ErrorCodes
 namespace FileCacheSetting
 {
     extern const FileCacheSettingsBool load_metadata_asynchronously;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_size;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_max_entries;
+    extern const ServerSettingsUInt64 object_storage_list_objects_cache_ttl;
 }
 
 }
@@ -2737,6 +2744,10 @@ try
     /// try set up encryption. There are some errors in config, error will be printed and server wouldn't start.
     CompressionCodecEncrypted::Configuration::instance().load(config(), "encryption_codecs");
 
+    ObjectStorageListObjectsCache::instance().setMaxSizeInBytes(server_settings[ServerSetting::object_storage_list_objects_cache_size]);
+    ObjectStorageListObjectsCache::instance().setMaxCount(server_settings[ServerSetting::object_storage_list_objects_cache_max_entries]);
+    ObjectStorageListObjectsCache::instance().setTTL(server_settings[ServerSetting::object_storage_list_objects_cache_ttl]);
+
     auto replicas_reconnector = ReplicasReconnector::init(global_context);
 
     /// Set current database name before loading tables and databases because

src/Access/Common/AccessType.h

@@ -334,6 +334,7 @@ enum class AccessType : uint8_t
     M(SYSTEM_DROP_SCHEMA_CACHE, "SYSTEM CLEAR SCHEMA CACHE, SYSTEM DROP SCHEMA CACHE, DROP SCHEMA CACHE", GLOBAL, SYSTEM_DROP_CACHE) \
     M(SYSTEM_DROP_FORMAT_SCHEMA_CACHE, "SYSTEM CLEAR FORMAT SCHEMA CACHE, SYSTEM DROP FORMAT SCHEMA CACHE, DROP FORMAT SCHEMA CACHE", GLOBAL, SYSTEM_DROP_CACHE) \
     M(SYSTEM_DROP_S3_CLIENT_CACHE, "SYSTEM CLEAR S3 CLIENT CACHE, SYSTEM DROP S3 CLIENT, DROP S3 CLIENT CACHE", GLOBAL, SYSTEM_DROP_CACHE) \
+    M(SYSTEM_DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE, "SYSTEM DROP OBJECT STORAGE LIST OBJECTS CACHE", GLOBAL, SYSTEM_DROP_CACHE) \
     M(SYSTEM_DROP_CACHE, "DROP CACHE", GROUP, SYSTEM) \
     M(SYSTEM_RELOAD_CONFIG, "RELOAD CONFIG", GLOBAL, SYSTEM_RELOAD) \
     M(SYSTEM_RELOAD_USERS, "RELOAD USERS", GLOBAL, SYSTEM_RELOAD) \

src/Common/ProfileEvents.cpp

@@ -1298,6 +1298,10 @@ The server successfully detected this situation and will download merged part fr
     M(RuntimeFilterRowsChecked, "Number of rows checked by JOIN Runtime Filters", ValueType::Number) \
     M(RuntimeFilterRowsPassed, "Number of rows that passed (not filtered out by) JOIN Runtime Filters", ValueType::Number) \
     M(RuntimeFilterRowsSkipped, "Number of rows in blocks that were skipped by JOIN Runtime Filters", ValueType::Number) \
+    M(ObjectStorageListObjectsCacheHits, "Number of times object storage list objects operation hit the cache.", ValueType::Number) \
+    M(ObjectStorageListObjectsCacheMisses, "Number of times object storage list objects operation miss the cache.", ValueType::Number)                                                                                                                                                                                                                                  \
+    M(ObjectStorageListObjectsCacheExactMatchHits, "Number of times object storage list objects operation hit the cache with an exact match.", ValueType::Number) \
+    M(ObjectStorageListObjectsCachePrefixMatchHits, "Number of times object storage list objects operation miss the cache using prefix matching.", ValueType::Number) \
 
 
 #ifdef APPLY_FOR_EXTERNAL_EVENTS

src/Common/TTLCachePolicy.h

@@ -273,10 +273,10 @@ class TTLCachePolicy : public ICachePolicy<Key, Mapped, HashFunction, WeightFunc
         return res;
     }
 
-private:
+protected:
     using Cache = std::unordered_map<Key, MappedPtr, HashFunction>;
     Cache cache;
-
+private:
     /// TODO To speed up removal of stale entries, we could also add another container sorted on expiry times which maps keys to iterators
     /// into the cache. To insert an entry, add it to the cache + add the iterator to the sorted container. To remove stale entries, do a
     /// binary search on the sorted container and erase all left of the found key.

src/Core/ServerSettings.cpp

@@ -1469,7 +1469,10 @@ The policy on how to perform a scheduling of CPU slots specified by `concurrent_
     ```xml
     <skip_check_for_incorrect_settings>1</skip_check_for_incorrect_settings>
     ```
-    )", 0)
+    )", 0) \
+    DECLARE(UInt64, object_storage_list_objects_cache_size, 500000000, "Maximum size of ObjectStorage list objects cache in bytes. Zero means disabled.", 0)                            \
+    DECLARE(UInt64, object_storage_list_objects_cache_max_entries, 1000, "Maximum size of ObjectStorage list objects cache in entries. Zero means disabled.", 0)                      \
+    DECLARE(UInt64, object_storage_list_objects_cache_ttl, 3600, "Time to live of records in ObjectStorage list objects cache in seconds. Zero means unlimited", 0)
 
 /// Settings with a path are server settings with at least one layer of nesting that have a fixed structure (no lists, lists, enumerations, repetitions, ...).
 #define LIST_OF_SERVER_SETTINGS_WITH_PATH(DECLARE, ALIAS) \

src/Core/Settings.cpp

@@ -7456,6 +7456,9 @@ Write full paths (including s3://) into iceberg metadata files.
 )", EXPERIMENTAL) \
     DECLARE(String, iceberg_metadata_compression_method, "", R"(
 Method to compress `.metadata.json` file.
+)", EXPERIMENTAL) \
+    DECLARE(Bool, use_object_storage_list_objects_cache, false, R"(
+Cache the list of objects returned by list objects calls in object storage
 )", EXPERIMENTAL) \
     DECLARE(Bool, make_distributed_plan, false, R"(
 Make distributed query plan.

src/Core/SettingsChangesHistory.cpp

@@ -289,6 +289,8 @@ const VersionToSettingsChangesMap & getSettingsChangesHistory()
             {"allow_experimental_lightweight_update", false, true, "Lightweight updates were moved to Beta."},
             {"s3_slow_all_threads_after_retryable_error", false, false, "Added an alias for setting `backup_slow_all_threads_after_retryable_s3_error`"},
             {"serialize_string_in_memory_with_zero_byte", true, true, "New setting"},
+            {"iceberg_metadata_log_level", "none", "none", "New setting."},
+            {"use_object_storage_list_objects_cache", false, false, "New setting."},
         });
         addSettingsChanges(settings_changes_history, "25.7",
         {

src/Disks/DiskObjectStorage/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h

@@ -35,6 +35,8 @@ class AzureObjectStorage : public IObjectStorage
         const String & description_,
         const String & common_key_prefix_);
 
+    bool supportsListObjectsCache() override { return true; }
+
     void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const override;
 
     /// Sanitizer build may crash with max_keys=1; this looks like a false positive.

src/Disks/DiskObjectStorage/ObjectStorages/IObjectStorage.h

@@ -329,6 +329,8 @@ class IObjectStorage
         throw Exception(ErrorCodes::NOT_IMPLEMENTED, "The method 'tagObjects' is only implemented for S3 and Azure storages");
     }
 #endif
+
+    virtual bool supportsListObjectsCache() { return false; }
 };
 
 using ObjectStoragePtr = std::shared_ptr<IObjectStorage>;

src/Disks/DiskObjectStorage/ObjectStorages/S3/S3ObjectStorage.h

@@ -63,6 +63,8 @@ class S3ObjectStorage : public IObjectStorage
 
     ObjectStorageType getType() const override { return ObjectStorageType::S3; }
 
+    bool supportsListObjectsCache() override { return true; }
+
     bool exists(const StoredObject & object) const override;
 
     std::unique_ptr<ReadBufferFromFileBase> readObject( /// NOLINT

src/Interpreters/InterpreterSystemQuery.cpp

@@ -46,6 +46,7 @@
 #include <Processors/Executors/PullingPipelineExecutor.h>
 #include <Processors/Sources/SourceFromSingleChunk.h>
 #include <QueryPipeline/QueryPipeline.h>
+#include <Storages/Cache/ObjectStorageListObjectsCache.h>
 #include <Storages/Freeze.h>
 #include <Storages/MaterializedView/RefreshTask.h>
 #include <Storages/ObjectStorage/Azure/Configuration.h>
@@ -470,7 +471,12 @@ BlockIO InterpreterSystemQuery::execute()
 #else
             throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, "The server was compiled without the support for AWS S3");
 #endif
-
+        case Type::DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE:
+        {
+            getContext()->checkAccess(AccessType::SYSTEM_DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE);
+            ObjectStorageListObjectsCache::instance().clear();
+            break;
+        }
         case Type::CLEAR_FILESYSTEM_CACHE:
         {
             getContext()->checkAccess(AccessType::SYSTEM_DROP_FILESYSTEM_CACHE);
@@ -1998,6 +2004,7 @@ AccessRightsElements InterpreterSystemQuery::getRequiredAccessForDDLOnCluster()
         case Type::CLEAR_SCHEMA_CACHE:
         case Type::CLEAR_FORMAT_SCHEMA_CACHE:
         case Type::CLEAR_S3_CLIENT_CACHE:
+        case Type::DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE:
         {
             required_access.emplace_back(AccessType::SYSTEM_DROP_CACHE);
             break;

src/Parsers/ASTSystemQuery.cpp

@@ -576,6 +576,7 @@ void ASTSystemQuery::formatImpl(WriteBuffer & ostr, const FormatSettings & setti
         case Type::CLEAR_COMPILED_EXPRESSION_CACHE:
         case Type::CLEAR_S3_CLIENT_CACHE:
         case Type::CLEAR_ICEBERG_METADATA_CACHE:
+        case Type::DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE:
         case Type::RESET_COVERAGE:
         case Type::RESTART_REPLICAS:
         case Type::JEMALLOC_PURGE:

src/Parsers/ASTSystemQuery.h

@@ -51,6 +51,7 @@ class ASTSystemQuery : public IAST, public ASTQueryWithOnCluster
         CLEAR_SCHEMA_CACHE,
         CLEAR_FORMAT_SCHEMA_CACHE,
         CLEAR_S3_CLIENT_CACHE,
+        DROP_OBJECT_STORAGE_LIST_OBJECTS_CACHE,
         STOP_LISTEN,
         START_LISTEN,
         RESTART_REPLICAS,

src/Storages/Cache/ObjectStorageListObjectsCache.cpp

@@ -0,0 +1,210 @@
+#include <Storages/Cache/ObjectStorageListObjectsCache.h>
+#include <Common/TTLCachePolicy.h>
+#include <Common/ProfileEvents.h>
+#include <boost/functional/hash.hpp>
+
+namespace ProfileEvents
+{
+extern const Event ObjectStorageListObjectsCacheHits;
+extern const Event ObjectStorageListObjectsCacheMisses;
+extern const Event ObjectStorageListObjectsCacheExactMatchHits;
+extern const Event ObjectStorageListObjectsCachePrefixMatchHits;
+}
+
+namespace DB
+{
+
+template <typename Key, typename Mapped, typename HashFunction, typename WeightFunction, typename IsStaleFunction>
+class ObjectStorageListObjectsCachePolicy : public TTLCachePolicy<Key, Mapped, HashFunction, WeightFunction, IsStaleFunction>
+{
+public:
+    using BasePolicy = TTLCachePolicy<Key, Mapped, HashFunction, WeightFunction, IsStaleFunction>;
+    using typename BasePolicy::MappedPtr;
+    using typename BasePolicy::KeyMapped;
+    using BasePolicy::cache;
+
+    ObjectStorageListObjectsCachePolicy()
+        : BasePolicy(CurrentMetrics::end(), CurrentMetrics::end(), std::make_unique<NoCachePolicyUserQuota>())
+    {
+    }
+
+    std::optional<KeyMapped> getWithKey(const Key & key) override
+    {
+        if (const auto it = cache.find(key); it != cache.end())
+        {
+            if (!IsStaleFunction()(it->first))
+            {
+                return std::make_optional<KeyMapped>({it->first, it->second});
+            }
+            // found a stale entry, remove it but don't return. We still want to perform the prefix matching search
+            BasePolicy::remove(it->first);
+        }
+
+        if (const auto it = findBestMatchingPrefixAndRemoveExpiredEntries(key); it != cache.end())
+        {
+            return std::make_optional<KeyMapped>({it->first, it->second});
+        }
+
+        return std::nullopt;
+    }
+
+private:
+    auto findBestMatchingPrefixAndRemoveExpiredEntries(Key key)
+    {
+        while (!key.prefix.empty())
+        {
+            if (const auto it = cache.find(key); it != cache.end())
+            {
+                if (IsStaleFunction()(it->first))
+                {
+                    BasePolicy::remove(it->first);
+                }
+                else
+                {
+                    return it;
+                }
+            }
+
+            key.prefix.pop_back();
+        }
+
+        return cache.end();
+    }
+};
+
+ObjectStorageListObjectsCache::Key::Key(
+    const String & storage_description_,
+    const String & bucket_,
+    const String & prefix_,
+    const std::chrono::steady_clock::time_point & expires_at_,
+    std::optional<UUID> user_id_)
+    : storage_description(storage_description_), bucket(bucket_), prefix(prefix_), expires_at(expires_at_), user_id(user_id_) {}
+
+bool ObjectStorageListObjectsCache::Key::operator==(const Key & other) const
+{
+    return storage_description == other.storage_description && bucket == other.bucket && prefix == other.prefix;
+}
+
+size_t ObjectStorageListObjectsCache::KeyHasher::operator()(const Key & key) const
+{
+    std::size_t seed = 0;
+
+    boost::hash_combine(seed, key.storage_description);
+    boost::hash_combine(seed, key.bucket);
+    boost::hash_combine(seed, key.prefix);
+
+    return seed;
+}
+
+bool ObjectStorageListObjectsCache::IsStale::operator()(const Key & key) const
+{
+    return key.expires_at < std::chrono::steady_clock::now();
+}
+
+size_t ObjectStorageListObjectsCache::WeightFunction::operator()(const Value & value) const
+{
+    std::size_t weight = 0;
+
+    for (const auto & object : value)
+    {
+        const auto object_metadata = object->metadata;
+        weight += object->relative_path.capacity() + sizeof(object_metadata);
+
+        // variable size
+        if (object_metadata)
+        {
+            weight += object_metadata->etag.capacity();
+            weight += object_metadata->attributes.size() * (sizeof(std::string) * 2);
+
+            for (const auto & [k, v] : object_metadata->attributes)
+            {
+                weight += k.capacity() + v.capacity();
+            }
+        }
+    }
+
+    return weight;
+}
+
+ObjectStorageListObjectsCache::ObjectStorageListObjectsCache()
+    : cache(std::make_unique<ObjectStorageListObjectsCachePolicy<Key, Value, KeyHasher, WeightFunction, IsStale>>())
+{
+}
+
+void ObjectStorageListObjectsCache::set(
+    const Key & key,
+    const std::shared_ptr<Value> & value)
+{
+    auto key_with_ttl = key;
+    key_with_ttl.expires_at = std::chrono::steady_clock::now() + std::chrono::seconds(ttl_in_seconds);
+
+    cache.set(key_with_ttl, value);
+}
+
+void ObjectStorageListObjectsCache::clear()
+{
+    cache.clear();
+}
+
+std::optional<ObjectStorageListObjectsCache::Value> ObjectStorageListObjectsCache::get(const Key & key, bool filter_by_prefix)
+{
+    const auto pair = cache.getWithKey(key);
+
+    if (!pair)
+    {
+        ProfileEvents::increment(ProfileEvents::ObjectStorageListObjectsCacheMisses);
+        return {};
+    }
+
+    ProfileEvents::increment(ProfileEvents::ObjectStorageListObjectsCacheHits);
+
+    if (pair->key == key)
+    {
+        ProfileEvents::increment(ProfileEvents::ObjectStorageListObjectsCacheExactMatchHits);
+        return *pair->mapped;
+    }
+
+    ProfileEvents::increment(ProfileEvents::ObjectStorageListObjectsCachePrefixMatchHits);
+
+    if (!filter_by_prefix)
+    {
+        return *pair->mapped;
+    }
+
+    Value filtered_objects;
+
+    filtered_objects.reserve(pair->mapped->size());
+
+    for (const auto & object : *pair->mapped)
+    {
+        if (object->relative_path.starts_with(key.prefix))
+        {
+            filtered_objects.push_back(object);
+        }
+    }
+
+    return filtered_objects;
+}
+
+void ObjectStorageListObjectsCache::setMaxSizeInBytes(std::size_t size_in_bytes_)
+{
+    cache.setMaxSizeInBytes(size_in_bytes_);
+}
+
+void ObjectStorageListObjectsCache::setMaxCount(std::size_t count)
+{
+    cache.setMaxCount(count);
+}
+
+void ObjectStorageListObjectsCache::setTTL(std::size_t ttl_in_seconds_)
+{
+    ttl_in_seconds = ttl_in_seconds_;
+}
+
+ObjectStorageListObjectsCache & ObjectStorageListObjectsCache::instance()
+{
+    static ObjectStorageListObjectsCache instance;
+    return instance;
+}
+
+}