Hi we have a production Environment that seem to have a memory leak.
THIS IS THE MEMORY CONFIGURATION
±-----------------------------------------±-------------------+
key_buffer_size | 8.000 MB |
query_cache_size | 1.000 MB |
innodb_buffer_pool_size | 24576.000 MB |
innodb_additional_mem_pool_size | 0.000 MB |
innodb_log_buffer_size | 16.000 MB |
±-----------------------------------------±-------------------+
BASE MEMORY | 24601.000 MB |
±-----------------------------------------±-------------------+
sort_buffer_size | 0.250 MB |
read_buffer_size | 0.125 MB |
read_rnd_buffer_size | 0.250 MB |
join_buffer_size | 0.250 MB |
thread_stack | 0.250 MB |
binlog_cache_size | 0.031 MB |
tmp_table_size | 50.000 MB |
±-----------------------------------------±-------------------+
MEMORY PER CONNECTION | 51.156 MB |
±-----------------------------------------±-------------------+
Max_used_connections | 804 |
max_connections | 1500 |
±-----------------------------------------±-------------------+
TOTAL (MIN) | 65730.625 MB |
TOTAL (MAX) | 101335.375 MB |
±-----------------------------------------±-------------------+
so it’s not supposed to go over 66 gigabytes
we are running mysql 5.7.32-35-log on linux 18.04.1-Ubuntu
SQL starts at about 27GIGA use after restart and climes to 130+ giga of memory use.
I enabled performance schema memory instruments and queried
as you can see connections from client takes about 120 giga of memory
95 percent of connections are idle/sleep and even if I kill connections that are idle for more then 600 seconds no memory is freed. also connections are killed after 1800 seconds of being idle.
did anyone encounter an issue like that?