I’ve got a master-master HA icinga setup, with separate icingaweb+director node.
It pushes config and commands into one of the masters, but if that one goes down, how can I make changes?
I can define a second command transport in icinga, but it won’t let me deploy any changes if one of the transport endpoints is unreachable.
Or should I be running icingaweb+director directly on the master icinga servers?
You can very easily switch the endpoint for director deployment to the second node, but if the first one comes back again you will have two authoritative versions of your configuration which you need to clean up. This is why there is no automatic fail over and fallback and the limiting factor is Icinga 2 in this case.