[Spread-users] New to spread, some questions
Ryan Caudy
rcaudy at gmail.com
Sat Feb 12 12:10:16 EST 2005
First, I want to make sure that we're talking about the same things.
I think when you said "mailbox" you meant "group" in the terminology
Spread uses. A "mailbox" usually refers to the handle/socket/fd
belonging to a single connected process.
To answer your question, if A sends to a group that only has members
connected to daemon-A and daemon-B, daemon-C *will* receive the
message, but won't deliver it to any processes. The reason for this
is Spread's ordered-delivery protocol. Better-engineered schemes
exist, but aren't implemented in the publicly available version of
Spread.
Cheers,
Ryan
On Fri, 11 Feb 2005 00:57:23 -0800, J C Lawrence <claw at kanga.nu> wrote:
> On Fri, 4 Feb 2005 19:02:31 -0500
> Ryan Caudy <rcaudy at gmail.com> wrote:
>
> > For many basic applications, the recommendation is one Spread daemon
> > per machine that you'll have clients on. This way, all the clients
> > can be local to their daemons.
>
> Will a spread message to a given mailbox ONLY be delivered to the spread
> daemons which have processes listening on that mailbox, or will it be
> delivered to all spread daemons in the spread fabric?
>
> Three systems: A, B and C.
>
> Each system is running its own spread daemon which are mutually
> connected.
>
> Similarly, on each system there's a process which we'll respectively
> call A', B', and C', which are connected to that spread bus.
>
> A' & B' share a mailbox called "AB" (ie both are connected to it)
>
> B' & C' share a mailbox called "BC".
>
> A' & C' share a mailbox called "AC".
>
> If A' sends a message to the AB mailbox, will the spread daemon
> running on C receive it for a client process that doesn't exist?
>
> --
> J C Lawrence
> ---------(*) Satan, oscillate my metallic sonatas.
> claw at kanga.nu He lived as a devil, eh?
> http://www.kanga.nu/~claw/ Evil is a name of a foeman, as I live.
>
More information about the Spread-users
mailing list