Currently the Zabbix Agent item "net.tcp.service" results in a value of "Down (0)" when using a FQDN longer than say 64 characters.

We're in the situation where our FQDN of the host that we'd like to check is more than 64 characters, because Amazon Aurora's cluster endpoint is constructed like: "<Our custom Aurora cluster ID>.cluster-cl51l5wnyxzg.us-east-1.rds.amazonaws.com"

We can easily reproduce the issue by simply running the following from commandline:

zabbix_agentd -c /etc/zabbix/zabbix_agentd.conf -t 'net.tcp.service[tcp,our-longish-cluster-id.cluster-cl51l5wnyxzg.us-east1.rds.amazonaws.com,3306]'

which outputs

net.tcp.service[tcp,our-longish-cluster-id.cluster-cl51l5wnyxzg.us-east-1.rds.amazonaws.com,3306] [u|0]

The closest ticket that I got to this issue was ZBXNEXT-1520, but its a database schema issue.

As an interim solution we've simply gone and created a CNAME DNS record under our own domain just for the use with net.tcp.service.