[Spread-users] retrans problem: UDP tests

Jonathan Stanton jonathan at cnds.jhu.edu
Wed Oct 9 17:32:12 EDT 2002


The short answer to question 2 is that no, that isn't possible the way
you present it. Spread will register both segments, but a particular
host can only be in one of them, so which ever address occurs first in
the file will be used. 

The closest thing I know of is a 'multipath' patch developed for Spread
that sends all of the messages accross two parallel networks (so two
broadcast/multicast addresses I think) for fault-tolerance reasons. The
patch can be found on the devel page of www.spread.org.

Can you say why you are trying to accomplish this? Maybe your end-goal
could be done a different way that would work. 

Something similar that is commonly done with Spread is to run two
daemons on the same machine and have each send/listen on different
multicast addresses and ports and thus communicate with different
subsets of the machines on the network. Especially with logging
services, this approach can work as the log records are rarely
inter-related and so ordering guaranteees are not needed for log
messages coming from different hosts.


On Tue, Oct 08, 2002 at 11:18:15AM +0200, Martin Decker wrote:
> Hello,
> two questions:
> 1) thanks for the hint to tst lowlevel udp with ./r and ./s. I found out that 
> one host has a miss rate of 15%, whereas the other has only 2%. I will have a 
> look at that machine now.
> 2) is it possible to configure one spread daemon to work with 2 mulicast 
> adresses?
> I want one host to be the sender to two multicast addresses, with 2 receivers 
> in each addresss.
> Spread_Segment {
>     dev907-green
>     dev908-green
>     dev912  
> }
> Spread_Segment {
>     dev911  
>     dev912  
> }
> is that possible?
> It seems that in the log file, only the first segment is listed, and the 
> second multicast address is listed without members.
> Martin
> -- 
> // Martin Decker, Systems Engineer, Data Center Operations     //
> // AOL Germany, Dingolfinger-Str. 2-4, D-81673 Munich, Germany //
> // AIM: mdecker77 mail: mdecker77 at aol.com ICQ: 15620821        //
> // Phone:+49.8966572214                                        //
> _______________________________________________
> Spread-users mailing list
> Spread-users at lists.spread.org
> http://lists.spread.org/mailman/listinfo/spread-users

Jonathan R. Stanton         jonathan at cs.jhu.edu
Dept. of Computer Science   
Johns Hopkins University    

More information about the Spread-users mailing list