Recursive alerts of “Utilization of housekeeper processes over 75%"

[Barneppe]

Good morning,

I would need your help to better optimize my Zabbix infrastructure. it has been many days that the system has been showing the recursive alert of “ Utilization of housekeeper processes over 75%”.

The instance consists of three machines in HA cluster:

• Zabbix primary server - active (vCPU: 16 / 64 GB RAM)
• Zabbix secondary server - standby (vCPU: 16 / 64 GB RAM)
• MySQL database (vCPU: 16 / 64 GB RAM).

These are the items and alerts configured on the infrastructure:

• Number of hosts: 11406
• Number of items: 241908
• Number of triggers: 495302
• Required server performance, new values per second: 3048.52

Regarding the tuning of the MySQL 8.0 database, the innodb_dedicated_server parameter has been added in the “my.cnf” file so that the system independently configures some values of the innodb.

This is the graph of the internal processes:


Could you please help me?

Thank you in advance for the support.

[guille.rodriguez]

Housekeeper is the process that Zabbix launch to delete older data of items... but if you have a lot of host and a lot of items, every clean will load a lot of consumption on Database and resources. Instead let Zabbix do this clean, you can make that MySQL do this clean using MySQL Partitioning.

https://blog.zabbix.com/partitioning...cedures/13531/

https://github.com/OpensourceICTSolutions/zabbix-mysql-partitioning-perl

Read carefully

Video about https://www.youtube.com/watch?v=ANQ3DHTr6eo

[Vinodh4181]

Mysql Partitioning and cleaning is the best practice, but not that database person!! so a small workaround i did is created UserParameter that will execute housekeeper runtime command on zabbix server and executing it every 20mins that's solved my problem. as we couldn't configure Zabbix server conf to execute less than 1hr interval I used this approach.