Hi Folks, There's a general issue that's been brewing with core and client access for the last few stable releases, which has gotten quite a bit worse with 1.19 that i wanted to bring up. Namely its how does an api client access the api.
Ignoring history, at the moment with a 'jenv' file is basically required for a client to be able to construct a secure, authenticated connection to the end point as certs, passwords and endpoint url are in jenv file. Unfortunately the jenv file format is an evolved one rather than a design one, more to the point, its been changing without notice and breaking clients. The most recent breaking change from 1.18 to 1.19 is the endpoint addresses have gone from being a list of public api endpoint per server, to a list of all addresses (including lots of unreachable ones) per server with no way of distinguishing which addresses correspond to which server or type/name of network. At the moment the juju cli just initiates connections en masse to all the addresses and stops when it gets a working connection, this is a heavy burden for other clients though. Ideally the cli client would distinguish which address in the cache its actually using to connect as opposed to just recording a bunch of addresses that don't work. Also it would ideally record network names/types for those addresses and distinguish between multiple servers. There's a few bugs entries related to this. https://bugs.launchpad.net/juju-core/+bug/1308491 https://bugs.launchpad.net/juju-core/+bug/1311227 cheers, Kapil
-- Juju-dev mailing list [email protected] Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju-dev
