Heartbeat: broadcast or point to point

[Alan Robertson]

Tom Vogt wrote:
> 
> Alan Robertson wrote:
> > I suppose you could piggyback acknowledgements on the back of your next
> > broadcast packet, since you know you'll be doing one in a second or so.
> >
> > Thoughts?
> 
> I think udp broadcasts are sufficient for my implementation because I
> only act if five in sequence go missing, so it's not critical if a
> packet or two get lost.

That may be sufficient for your particular application, but for some
applications, 5 seconds downtime without knowing it is a long time.  It
would be nice if we could do something which could be tuned to other
applications.

> I'm also now implementing control messages and I've been thinking about
> using tcp connections for those or in general thinking about the
> acknowledgement problem. if anyone has a good solution, I'd surely be
> interested.

As a note, the heartbeat code I have up on the net will send arbitrary
control messages over the same (heartbeat) channel.  Whatever you write
to the FIFO gets sent to all the members of the cluster.  That's how it
implements IP address takeover, etc.  I think Stephen's code uses the
same channel for everything.

	-- Alan Robertson
	   alanr@bell-labs.com