On Sunday, September 08, 2013 12:50:23 PM Paul Colomiets wrote: > Hi Martin, > > On Sun, Sep 8, 2013 at 10:07 AM, Martin Sustrik <sustrik@xxxxxxxxxx> wrote: > > Hi Alex, Paul, > > > >>> Mm, good point. The core of it, as I said above, is that these decisions > >>> seem > >>> like they are best made at *topology* deployment time rather than > >>> *participant* deployment, or even more dynamically than that. > >> > >> Participant deployment = topology change, AFAIU. E.g. if one of the > >> nodes failed you may need to turn nn_connect socket to nn_bind. > > > > This is an interesting disucssion and is directly related to different > > admin roles that Alex referred to in the original email. > > > > So what roles do we have? My feeling was that there's a programmer who > > writes a distributed application without committing to any particular > > topology setup and, at each client site, there's an app admin (or > > deployer, > > if you will) that maps abstract topologies defined by the programmer to > > actual network topology on the site. And that's it. > > Yes. > > > So, am I missing something? Is there anyone else involved? What's > > "participant deployment"? Etc. > > It's a task not a role, AFAIU. But that "AFAIU" means I'm barely > speculating. You're correct. When I say "topology deployment" I'm referring more to the initial rollout, where the topology is first getting set up; a participant would be an endpoint - part of the application itself; participant deployment covers things like adding a new node to a REP cluster.