Zabbix continuously consumes more and more RAM

[Mateusz]

Greetings Everyone

My problem is with Zabbix 6.0.0 with MySQL 8.0.29 on Ubuntu 20.04 (Zabbix and DB on the same virtual machine). After the reboot and during operation server continues to consume more and more RAM. I have two instances of Zabbix 6.0.0 which have similar issue. RAM Memory for the machine has been recently extended from 8GB to 16GB and it just continues to consume more. When I look at the memory graph, I see that since last reboot, every hour between xx:15 and xx:20, available memory drops down by around 200MB. Looking at the Linux server I see that it's /usr/bin/mysqld process that is taking the memory. Could someone give me a hint on troubleshooting this?

[cyber]

First of all, try to update to latest (6.0.6). 6.0.0 as initial release is always buggy.. No matter how much betas or RC-s have been produced..
Maybe someone else can say something about debugging mysql...

[Mateusz]

Thanks, upgrading to latest is always a good advice. I am starting to wonder if this is even a problem. I see that overnight memory usage has lowered but in long-term perspective it's still growing.

[Mateusz]

Finally memory usage is stable at around 10GB.. I'll be still watching this topic if someone wants to share their experience.

[Colttt]

mysql uses only as much RAM as you configured it to use ;-)
so it hardly depends on you mysql-configuration

[Ismael Cruz]

I have the same problem, the Zabbix database is consuming a lot of memory, it grows more and more on the graph.

free -h
total used free shared buf/cache available
Mem: 19Gi 17Gi 291Mi 1.0Mi 1.9Gi 1.8Gi

TOP
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1437512 mysql 20 0 18.2g 16.9g 15264 S 27.6 86.7 9572:42 mariadbd


df -h
500G 296G 205G 60% /database


zabbix version: 6.0
version bank mariadb 10.6



my.cnf

[mysqld]
max_allowed_packet = 128M
open_files_limit = 4096
max_connections = 250

sync_binlog = 0

performance_schema = ON
query_cache_size = 0
query_cache_type = 0
innodb_buffer_pool_instances = 8
innodb_buffer_pool_size = 11G
innodb_file_per_table = ON
innodb_log_file_size = 1G
innodb_stats_on_metadata = OFF
innodb_flush_method = O_DIRECT
innodb_old_blocks_time = 1000
innodb_flush_log_at_trx_commit = 0

sort_buffer_size = 1M
join_buffer_size = 512K
read_buffer_size = 512K
read_rnd_buffer_size = 1M
tmp_table_size = 32M
max_heap_table_size = 32M

[Ismael Cruz]

The problem has been solved.

I had a routine in the database that performed an insert, line by line; it was a file with many inserts, and that was the problem.

I changed the routine, setting it to create inserts in blocks, resulting in only one or two inserts. This solved the memory consumption problem.