This changes the overall approach for grabbing downtime information from Topology. Previously, we generated the list of downed servers by getting a diff of the servers returned in the namespaces JSON when querying that endpoint with ?include_downed=<0/1>. The issue with this approach is that Topology services marked as Pelican services are explicitly excluded from the returned JSON.
However, Pelican services that are recorded in Topology are still included in the downtime URL, so this new approach grabs the list of resource names from that instead. That was probably the right approach all along, but topology endpoints remain inscrutable ¯_(ツ)_/¯
This also corrects a documentation issue that claimed filtered servers could be added in the Director using their hostname. In fact, they need to be filtered using their site name!
This changes the overall approach for grabbing downtime information from Topology. Previously, we generated the list of downed servers by getting a diff of the servers returned in the namespaces JSON when querying that endpoint with
?include_downed=<0/1>. The issue with this approach is that Topology services marked as Pelican services are explicitly excluded from the returned JSON.However, Pelican services that are recorded in Topology are still included in the downtime URL, so this new approach grabs the list of resource names from that instead. That was probably the right approach all along, but topology endpoints remain inscrutable ¯_(ツ)_/¯
This also corrects a documentation issue that claimed filtered servers could be added in the Director using their hostname. In fact, they need to be filtered using their site name!
Closes #1578 and partially closes #1258