I've recently found that one of my SSD's has high wear level count. A bit of diags shows its related to my zabbix VM and specifically the db side (mariadb). Mariadb is in the same VM as zabbix. I'm seeing 350GB/24hours writes.
The process that appears to be the issue is 'mysqld ib_pg_flush_co' according to iotop and it's taking a consistent 2-3MB/s write with frequent peaks higher. If i stop zabbix-server (or the VM entirely) then MB/s will drop to <150KB/s consistent.
Since this is for monitoring at home, it's a very small install so surprised it even has 2-3MB data in 60secs, let alone per second.
22 hosts, 1266 items, values/sec 11.09, zabbix v5.4.7, mariadb 10.3.31.
currently mysql.cnf has just the following confgured and it hasnt made much of an improvement
what can i do to improve this and stop it eating through the SSD?
The process that appears to be the issue is 'mysqld ib_pg_flush_co' according to iotop and it's taking a consistent 2-3MB/s write with frequent peaks higher. If i stop zabbix-server (or the VM entirely) then MB/s will drop to <150KB/s consistent.
Since this is for monitoring at home, it's a very small install so surprised it even has 2-3MB data in 60secs, let alone per second.
22 hosts, 1266 items, values/sec 11.09, zabbix v5.4.7, mariadb 10.3.31.
currently mysql.cnf has just the following confgured and it hasnt made much of an improvement
Code:
innodb_buffer_pool_size=4G innodb_log_buffer_size=64M innodb_log_file_size=192M