Background

To make a critical Zabbix instance more redundant, a second esxi host was made in another datacenter. The MySQL database from the original instance was transferred via a sqldump to a centralized DB cluster, and both hosts connect to this DB cluster to get their data.

At random intervals, the frontends of Zabbix report that there is no response from the Zabbix server that is main. After a while (anywhere between 5 and 20 minutes), it works as normal again for multiple hours.{}

Configuration has been checked several times compared to a cluster we have, it's identical and the problem doesnt appear there.

Logs don't reveal any kind of problem, neither in the server log nor the Nginx log. All polling and data ingestion in the database works fine, but the system information dies from time to time, and is quite often paired with the trapper processes going to 100%

Steps to reproduce:

1. Zabbix changed from single instance to HA setup
2. moved database to mysql innodb cluster after sql dump
3. configured frontend and backend zabbix server to connect via mysqlrouter
4. at random intervals, the webgui cannot read the system information (on either instance)

Result:
See screenshot screenshots attached

Expected:

System information should always load and work normally, this is the case in all our other instances of Zabbix