Odd autoregistration behavior with Zabbix 7.0

[jhboricua]

I'm upgrading to Zabbix 7 and have a prod and non-prod container deployment of Zabbix 7.0.2

Our current Zabbix implementation is 5.0 LTS (5.0.42 to be exact). We are using the Zabbix Agent2 for the monitored hosts.

I'm seeing an odd behavior when I added a few host currently registered on the 5.0 deployment to the 7.0 deployment. When the host autoregisters in 7.0, its hosts interface IP address field is being registered as 127.0.0.1, and with localhost in the 'DNS name' field. On the 5.0 deployment, when the hosts registers, their actual IP address is in the interface IP field with the 'DNS name' field left empty.

The only parameters I have modified in the agent configuration file are:

Servers: with dns names of zabbix server endpoints
ServersActive: same as above
HostName: FQDN name of the monitored server.

Here's what it looks in 5.0:


Same host on 7.0:



As a result I'm seeing these in the backend logs of the 7.0 zabbix server:

"cannot send list of active checks to "127.0.0.1": host [svw005syseng700.xxxxxxx] not found"

Updating the agent in the server from 5.0.42 to 7.0.2 doesn't change the registration issue. Trying to understand why these host are not registering correctly in 7.0.

[jhboricua]

For some reason the screenshots were showing for me but not when I looked at the thread from another machine. Re-uploaded screenshots.

[cyber]

I think there was somekind of DNS related change in 7. At least I have noticed some topics lately, which mention this... Maybe its somehow related, but I cannot really say, as I had no issues with autoreg on 7....
Maybe Markku has some comments?

[Markku]

No comments, I haven't tested autoregistration with Zabbix 7.

Markku

[jhboricua]

For now, I have restored the default registration behavior in Zabbix 7 by adding the HostInterface parameter with the IP address of the host as a value in the agent configuration file.