I have a distributed Icinga 2 setup where the master and satellite zones contain two endpoints each. This is meant to support high availability (HA). What I notice is that checks are scheduled “at random” between each pair of endpoints. This gives unwanted results e.g. in the case of check apt, because the metric may from a different host than the one that is associated with the service. How is one supposed to handle such a situation?
In your Service set command_endpoint = "<DESIRED ENDPOINT>".
It apparently gets fairly complicated in the case if a HA setup, but yes, this indeed proved to be the right starting point. Thanks.