если пробовать сделать повторную вставку этого же куска файла (после предварительной попытки вставки цельного куска в 18 Гб), то не долго думая выкидывает ошибку (секунда или две где-то):
Code: 210. DB::NetException: Connection reset by peer while writing to socket (127.0.0.1:9000) Итого:
1. если пробовать залить по одному файлу - то заходит не далее 14 файла (в среднем в файле 50-55 тыс. строк).
вливается не больше 900 000 строк
2. если вливать цельным куском в 18 Гб, то работает продолжительное время и падает где-то на 140 секунде.
вливается 15.3 млн строк.
PS вставка каждый раз производилась на пустой таблице.
Настройки сервера (из дефолтных менял только max_execution_time - для проверки, но не помогло) :
SELECT *
FROM system.settings
FORMAT CSV
"min_compress_block_size","65536",0
"max_compress_block_size","1048576",0
"max_block_size","65536",0
"max_insert_block_size","1048576",0
"min_insert_block_size_rows","1048576",0
"min_insert_block_size_bytes","268435456",0
"max_threads","8",0
"max_read_buffer_size","1048576",0
"max_distributed_connections","1024",0
"max_query_size","262144",0
"interactive_delay","100000",0
"connect_timeout","10",0
"connect_timeout_with_failover_ms","50",0
"receive_timeout","300",0
"send_timeout","300",0
"queue_max_wait_ms","5000",0
"poll_interval","10",0
"distributed_connections_pool_size","1024",0
"connections_with_failover_max_tries","3",0
"extremes","0",0
"use_uncompressed_cache","0",1
"replace_running_query","0",0
"background_pool_size","6",0
"distributed_directory_monitor_sleep_time_ms","100",0
"optimize_move_to_prewhere","1",0
"replication_alter_partitions_sync","1",0
"replication_alter_columns_timeout","60",0
"load_balancing","random",1
"totals_mode","after_having_exclusive",0
"totals_auto_threshold","0.5",0
"compile","0",0
"min_count_to_compile","3",0
"group_by_two_level_threshold","100000",0
"group_by_two_level_threshold_bytes","100000000",0
"distributed_aggregation_memory_efficient","0",0
"aggregation_memory_efficient_merge_threads","1",0
"max_parallel_replicas","1",0
"parallel_replicas_count","0",0
"parallel_replica_offset","0",0
"skip_unavailable_shards","0",0
"distributed_group_by_no_merge","0",0
"merge_tree_min_rows_for_concurrent_read","163840",0
"merge_tree_min_rows_for_seek","0",0
"merge_tree_coarse_index_granularity","8",0
"merge_tree_max_rows_to_use_cache","1048576",0
"merge_tree_uniform_read_distribution","1",0
"optimize_min_equality_disjunction_chain_length","3",0
"min_bytes_to_use_direct_io","0",0
"force_index_by_date","0",0
"force_primary_key","0",0
"strict_insert_defaults","0",0
"mark_cache_min_lifetime","10000",0
"max_streams_to_max_threads_ratio","1",0
"network_compression_method","lz4",0
"priority","0",0
"log_queries","0",0
"log_queries_cut_to_length","100000",0
"distributed_product_mode","deny",0
"global_subqueries_method","push",0
"max_concurrent_queries_for_user","0",0
"insert_quorum","0",0
"insert_quorum_timeout","600000",0
"select_sequential_consistency","0",0
"table_function_remote_max_addresses","1000",0
"max_distributed_processing_threads","8",0
"read_backoff_min_latency_ms","1000",0
"read_backoff_max_throughput","1048576",0
"read_backoff_min_interval_between_events_ms","1000",0
"read_backoff_min_events","2",0
"memory_tracker_fault_probability","0",0
"enable_http_compression","0",0
"http_zlib_compression_level","3",0
"http_native_compression_disable_checksumming_on_decompress","0",0
"resharding_barrier_timeout","300",0
"count_distinct_implementation","uniq",0
"max_rows_to_read","0",0
"max_bytes_to_read","0",0
"read_overflow_mode","throw",0
"max_rows_to_group_by","0",0
"group_by_overflow_mode","throw",0
"max_bytes_before_external_group_by","0",0
"max_rows_to_sort","0",0
"max_bytes_to_sort","0",0
"sort_overflow_mode","throw",0
"max_bytes_before_external_sort","0",0
"max_result_rows","0",0
"max_result_bytes","0",0
"result_overflow_mode","throw",0
"max_execution_time","86400",1
"timeout_overflow_mode","throw",0
"min_execution_speed","0",0
"timeout_before_checking_execution_speed","0",0
"max_columns_to_read","0",0
"max_temporary_columns","0",0
"max_temporary_non_const_columns","0",0
"max_subquery_depth","100",0
"max_pipeline_depth","1000",0
"max_ast_depth","1000",0
"max_ast_elements","50000",0
"readonly","0",0
"max_rows_in_set","0",0
"max_bytes_in_set","0",0
"set_overflow_mode","throw",0
"max_rows_in_join","0",0
"max_bytes_in_join","0",0
"join_overflow_mode","throw",0
"max_rows_to_transfer","0",0
"max_bytes_to_transfer","0",0
"transfer_overflow_mode","throw",0
"max_rows_in_distinct","0",0
"max_bytes_in_distinct","0",0
"distinct_overflow_mode","throw",0
"max_memory_usage","10000000000",1
"max_memory_usage_for_user","0",0
"max_memory_usage_for_all_queries","0",0
"max_network_bandwidth","0",0
"max_network_bytes","0",0
пятница, 19 августа 2016 г., 21:30:28 UTC+3 пользователь Юрий Кудряшов написал: