[erlang-questions] node discovery (was nodes not visible until net_adm:world() )

Richard Andrews <>
Wed Oct 3 01:39:24 CEST 2007


--- Claes Wikström <> wrote:

> Node to node connections are on-demand, thus not until you
> actively try to spawn/send to a remote node is connection
> established. Hence originally nodes() == [], and then if you do
> 
> {foo, } ! whatever
> 
> will a connection to  be established and if all
> is well, nodes() == []

It seems the new node needs to be connected to a least one node in the global
nameserver network before any of the global registration information can be
used. Once connected to the first node, the rest become visible automatically.
However the new node should not know any other node names a-priori.

Is there a more appropriate mechanism to achieve a fully connected network than
net_adm:world()? I find that call distasteful as the connected network is a
side-effect of the call rather than the intent and as such might not always
work that way in the future. Is there any call which has the explicit intent of
creating a fully connected network?



      Sick of deleting your inbox? Yahoo!7 Mail has free unlimited storage.
http://au.docs.yahoo.com/mail/unlimitedstorage.html




More information about the erlang-questions mailing list