[erlang-questions] monitor and failure detectors

[Johan Montelius]

Agree, there is no point in delaying information but if I have a  
distributed system with plenty of nodes and I want five processes (on  
different nodes) to monitor each other more closely, then I could let them  
have a monitor timeout of two seconds while all other nodes monitor each  
other as usual with a 60s timeout.

If one had this approach one could of course also ask the question - if we  
don't have any nodes monitoring each other (or links between them) then  
why query the status of another node at all? Setting up heartbeats between  
nodes could be a on demand service.

   Johan


On Mon, 07 Jun 2010 07:23:15 +0200, Scott Lystig Fritchie  
<fritchie@REDACTED> wrote:

> Delaying the delivery of those {'DOWN', ...} events doesn't seem to me to
> have much useful value.


-- 
Associate Professor Johan Montelius
Royal Institute of Technology - KTH
School of Information and Communication Technology - ICT