Hi,
I suggest looking at the Grafana logs. Also try accessing your data
source URL of the problematic Prometheus instance from your Grafana
server via curl. Maybe there is some firewall restriction in place (ping
is not sufficient, you will need tcp access on the relevant port [9090,
by default]).
Kind regards,
Christian