New Zabbix items do not gather data unless Zabbix server is restarted

[andriusmad]

Hello everyone,

This week I've noticed issues with our Zabbix monitoring.

First of all, this week we migrated from the Percona cluster backend to AWS RDS MySQL backend for the Zabbix server and most likely issues are related to this event.

Info:
Zabbix server 6.4.2 running native HA
AWS RDS MySQL 8.0.32 as the backend (via HAProxy)
Zabbix proxies 6.4.1 - 6.4.3, active mode
Zabbix agents 5.4, 6.0, 6.4

The issue is strange but simple. When new items are added to hosts (for example new host with autoregistration, a new item created on the host, or a new template added to the host) they do not gather any data. The item exists, but no data in the history seems that "nodata" triggers also do not react to the items.
However, if I restart the Zabbix server service, those items start filling up and continue doing so.

I've tried raising the Zabbix server log global level to 4 to check what could be wrong with the DB backend or the server, but nothing related. Checked Zabbix proxy logs and Zabbix agent logs as well and nothing related to this.

Maybe someone has encountered such an issue? I think it could be related to the RDS backend, because our stage setup is pretty much the same, but with Percona cluster, not RDS.

[andriusmad]

I've managed to find the issue. Apparently, migration to RDS wasn't completely done, and DB was missing deterministic triggers. After triggers were added, the issue was solved.

[andriusmad]

I've managed to find the issue. Apparently, migration to RDS wasn't completely done, and DB was missing deterministic triggers. After triggers were added, the issue was solved.

[dom_b]

I've managed to find the issue. Apparently, migration to RDS wasn't completely done, and DB was missing deterministic triggers. After triggers were added, the issue was solved.

I've had this exact same issue after migrating from mysql to postgres. Did you have to create the table? How did you import / create the correct table?

[9a8]

To fix the issue I had to execute all statements related to triggers from /usr/share/zabbix-sql-scripts/postgresql/server.sql.gz (lines 2279 to 2844 for version 6.4) against zabbix database.

[andriusmad]

I've had this exact same issue after migrating from mysql to postgres. Did you have to create the table? How did you import / create the correct table?

I've left that to the DB team, but as far as I know, they created mysqldump of the previous cluster, and the dump is formated as a script. From that, they got the commands for the triggers. Keep in mind, that these MySQL triggers, not Zabbix so its not a regular table, but a part of MySQL information schema.