Zabbix 3.4 and MySQL partitioning

[scout]

Hello guys,

is there still an option and recommendation to use MySQL table partitioning for big Zabbix database (i.e. 100G+)?
If no, is there a normal way to disable MySQL table partitioning for Zabbix database?
Right now we are using MySQL partitioning and would like to avoid feature implementation of Zabbix system and MySQL table partionining for Zabbix system deprecation.

Thanks

[tmroberts]

As a former DBA and now a Zabbix admin over a system that currently is running about 800GB and growing, you WANT to use partitioning. Not only will it make pruning data more efficient than the house keeper processes, but it will also make your queries run significantly faster. The house keeper function does delete operations on the tables that get partitioned (history and trend tables) and when you have significant data in those tables, deleting from the table just makes things a mess. Partitioning allows for the removal of data quickly and with out table locks.

[kloczek]

Hello guys,

is there still an option and recommendation to use MySQL table partitioning for big Zabbix database (i.e. 100G+)?

What you mean "normal"?

If no, is there a normal way to disable MySQL table partitioning for Zabbix database?

Depends how you are maintaining partitions just disable SQL event or external cron job.

Right now we are using MySQL partitioning and would like to avoid feature implementation of Zabbix system and MySQL table partionining for Zabbix system deprecation.

Zabbix code has no any collisions with use partitioning so by definition it will be no deprecation.
Only feature which may collide would be possible adding partitioning naively introduced by server and/or proxy which may require disable all in-house partitions management to allow zabbix server/proxy maintain straight by those processes.

Zabbix sees/interacts with database over result of the queries like select, insert, update and results all of those queries is exactly the same in case use partitioned and non-partitioned tables. In other words use of partitioning or not is 100% transparent from point of the queries used by zabbix server, proxy and web frontend.