cache problem on controllers

asked 2019-06-01 04:46:12 -0600

mehrdad gravatar image

I have 3 controller, that they cluster to gether by reddis and keepalived. their total ram it's about 32 GB , used 19 GB and buff/cache 6.5~7 GB my OS:centos 7.5 my OpenStack version: QUEEN my sql version: mysql Ver 14.14 I configured 3GB for reddis database usage of ram. i didn't grow up and i don't have user's anymore. but everyday cache memory is grow up till i free dirty_cache htop command show's me a lot of /usr/sbin/mysqld jobs. my database configuration is : default_storage_engine = InnoDB innodb_buffer_pool_instances = 2 innodb_buffer_pool_size = 2G innodb_file_per_table = 1 innodb_flush_log_at_trx_commit = 0 innodb_flush_method = O_DIRECT innodb_log_buffer_size = 16M innodb_log_file_size = 512M innodb_stats_on_metadata = 0 innodb_read_io_threads = 64 innodb_write_io_threads = 64 query_cache_limit = 4M # UPD query_cache_size = 48M # UPD query_cache_type = 1 key_buffer_size = 48M low_priority_updates = 1 concurrent_insert = 2 max_connections = 1000 back_log = 512 thread_cache_size = 100 thread_stack = 192K interactive_timeout = 180 wait_timeout = 180 join_buffer_size = 4M read_buffer_size = 3M read_rnd_buffer_size = 4M sort_buffer_size = 4M table_definition_cache = 10000 table_open_cache = 10000 open_files_limit = 60000 max_heap_table_size = 128M tmp_table_size = 128M ft_min_word_len = 3 log_error = /var/log/mysql_error.log log_queries_not_using_indexes = 1 long_query_time = 5 slow_query_log = 0 slow_query_log_file = /var/log/mysql_slow.log wsrep_slave_threads = 32 wsrep_provider_options = "gcs.fc_limit= 160; gcs.fc_factor=0.8" wsrep_notify_cmd = notify innodb_autoinc_lock_mode = 2 max_user_connections = 4096 thread_cache_size = 4096 binlog_format = ROW wsrep_sst_method=rsync max_allowed_packet = 256M

**as I search for that problem I found database service doesn't close open page files!?

can anyone help me to solve this problem?**

edit retag flag offensive close merge delete