Does the trustdomain that is used to create the root CA have any relationship with the DNS domain of the controllers hosting the network?
If not, are there any constraints on the trustdomain? For example, can it be a single-part name like myzitinetwork? Or does it have to look like a domainname myzitinetwork.com?
What are the constraints on the controller name in the spiffe id? Do they have to match the any part of the DNS names assigned to the controller?
Questions around new authentication API (edge-oidc):
Is this new API going to be used for both edge-clients and edge-management auth?
If we bind the edge-client and edge-management-api to different addresses, can we include edge-oidc in both the bindings?
No. Easy to think they do because there is a protocol and hostname, but they are unrelated. Trust domains are used to verify that a SPIFFE ID in a verifiable document (cert, jwt) is "trusted"/"related"/"applicable" to the receiving application.
In OpenZiti, we use the trust domain as an opaque string. Since it is a URI, it must conform to URI hostname standards. Top-level domains (.com, .net etc.) are not required. Trust domains are never resolved through DNS for any reason.
Controller SPIFFE IDs on server and client certificates should be in the format of spiffe://<trust domain>/controller/<id>