Cannot get agent to show as available after moving machine to another DC/proxy

[Gadg3ts]

So we have two data centres.
Zabbix is monitoring about 160 hosts, with a zabbix-proxy in each data centre, pushing out to a zabbix-server that lives in AWS.
One particular machine has been moved from data centre 'A' to data centre 'B' and had the OS reinstalled (old version of CentOS to Proxmox).
The agents are deployed/configured via ansible and also set to auto-register.
The newly re-installed agent and proxy 'B' are definitely communicating as I can see that in the appropriate logs (dns is fine on both directions).
When I delete the host it registers again and gives itself the correct proxy in the main zabbix gui.
But the agent won't show up as green in the gui, nor is it showing anything in 'latest data'.
I first tried purging the history of the existing data before I removed the host entry and set this particular machine to auto-register, in case that was the problem, but it didn't help.

I'm thinking there's a blockage somewhere in that zabbix somehow hasn't accepted the machine has moved to the other proxy.
Anyone seen anything similar and managed to fix it?


Logs on the main server are showing this after removing the host and letting it add itself again (and then clearing the templates and waiting a few minutes before adding the 'linux agent active' templates):

21204:20210222:141447.245 In evaluate_function() function:'vs34.example.net:agent.ping.nodata({$AGE NT.NODATA_TIMEOUT})'
21204:20210222:141447.267 End of evaluate() error:'Cannot evaluate expression: "Cannot evaluate function "vs34.example.net:agent.ping.nodata({$AGENT.NO DATA _TIMEOUT})": item does not have enough data after serve
r start or item creation.".'
21204:20210222:141447.267 evaluate_expressions():expression [(ZBX_UNKNOWN56)=1] cannot be evaluated: Cannot evaluate expression: "Cannot evaluate function "vs34.example.net:agent.ping.nodata({$AGENT.NO DATA _TIME
OUT})": item does not have enough data after server start or item creation.".
21203:20210222:141517.012 In evaluate_function() function:'vs34.example.net:agent.ping.nodata({$AGE NT.NODATA_TIMEOUT})'
21203:20210222:141517.016 End of evaluate() error:'Cannot evaluate expression: "Cannot evaluate function "vs34.example.net:agent.ping.nodata({$AGENT.NO DATA _TIMEOUT})": item does not have enough data after serve
r start or item creation.".'
21203:20210222:141517.016 evaluate_expressions():expression [(ZBX_UNKNOWN0)=1] cannot be evaluated: Cannot evaluate expression: "Cannot evaluate function "vs34.example.net:agent.ping.nodata({$AGENT.NO DATA _TIMEOUT})": item does not have enough data after server start or item creation.".

Does the "ZBX_UNKNOWN" mean there's an orphan expression/trigger that is causing this?..


Thanks,
Sean

[Gadg3ts]

"There are no stupid questions"?..
The registration (now) adds "Template OS Linux by Zabbix Agent Active" although I only added that this morning to see if it would make a difference. 'Normally' this is added via the ansible module on agent deployment.
The ports are open both directions between the proxy and machine (it's a very flat network), so that all seems ok. Plus the time is correct within 3 seconds between everything (I sorted ntp a few weeks ago).
It's only this one machine that is misbehaving.
Proxmox is Debian underneath, so there's no opening ports like on CentOS/RHEL.

[Gadg3ts]

Incidentally, the agent data is now showing up, but the agent isn't showing as available, probably because of this (from the zabbix server log):
21201:20210222:152047.033 In evaluate_function() function:'vs34.example.net:agent.ping.nodata({$AGE NT.NODATA_TIMEOUT})'
which is weird, because that works if I do it manually from the proxy to the machine.

[cyber]

Maybe a stupid question, but is your autoregistration also assigning some templates to that host? .. "The Green" will appear only in case of successful passive checks.
If your passive checks are not getting any data, there might be some issues with connection towards the host. some local fw or selinux etc...