Is this a proofed concept? Because if I’m not mistaken now, the sattelite hosts only act as a relay host and carry out the checks, or transmit them to the clients. The result of the check is then transmitted back to the master, who writes the check result to the database.
Icingaweb2 then takes the information from the database and displays it. Furthermore, commands such as rechecks, acknowlegements and downtimes are accepted via the icinga2 api on the master.
The problem would be now, if the connection to the master host is interrupted, the satellite host can no longer accept commands from the master and also report no check results. Icingaweb2 on the satellite would also not be able to transmit commands via api.
I also do not think that it is possible to operate separate databases for master and satellite hosts. Even if you replicate the database to the second site, no new check results would get into it, as this was only done by the master.
However, it might be possible to operate a second master at the second location. You would also have to run two databases for each side and icingaweb2 would be installed on both master instances.
You could run the icinga director on the master instance on your side and this would then replicate the hosts and checks to the other side.
I am not sure what will happen if the connection is lost. In this case, both databases would have different entries and the primary master on your side would not be up to date until after having run all checks again.
Please correct me if i’m wrong, i’m always happy to learn new stuff.