[lug] routing

[Hugh Brown]

I have determined that it really was a routing issue by putting each
crossover-paired set of cards on their own subnet (everything works fine
now).  This is a test bed for what will later be a 64 node cluster. 
We're trying to get around buying a gigabit switch, but crossovering
(creative english :)  64 nodes means 64! cables and 63 cards in each
machine which will never happen (only have ~6 pci slots).

I may try and go back to the all on the same subnet and carefully set up
the route instead of letting redhat do it for me.

Thanks for the help.

Hugh


On Thu, 2002-04-25 at 05:04, Charles Morrison wrote:
> D. Stimits wrote:
> 
> > Don't forget, netmask is also very important. Perhaps the results of
> > ifconfig from each machine needs to be listed.
> > 
> > D. Stimits, stimits at idcomm.com
> > 
> 
> 
> Be my guest. :o)
> 
> Not to mention the hosts file if you don't want to reference by IP 
> number only.
> 
> 
> 
> 
> > Charles Morrison wrote:
> > 
> >>Charles Morrison wrote:
> >>
> >>Don't you love people who reply to their own emails ?
> >>
> >>
> >>>If what you're trying to do is have one card dedicated to communicate
> >>>soley to one other machine, use different class C networks for each
> >>>destination and it's associated card and have them be members of one
> >>>network. This way you don't have to muck with routing tables.
> >>>
> >>OK, that's confusing. maybe some examples.
> >>
> >>network
> >>        host IP(node1)  dest node        dest IP
> >>10.1.12.x
> >>  10.1.12.1      node2          10.1.12.2
> >>10.1.13.x
> >>  10.1.13.1      node3          10.1.13.3
> >>10.1.14.x
> >>  10.1.14.1      node4          10.1.14.4
> >>
> >>Keeping the node numbers as the last numbers of the IP should help keep
> >>things straight. so the other nodes would be set up ...
> >>
> >>network
> >>        host IP(node2)  dest node        dest IP
> >>10.1.12.x
> >>10.1.12.2
> >>node1
> >>        10.1.12.1
> >>10.1.23.x
> >>10.1.23.2
> >>  node3         10.1.23.3
> >>10.1.24.x
> >>10.1.24.2
> >>node4
> >>        10.1.24.4
> >>
> >>network
> >>        host IP(node3)  dest node        dest IP
> >>10.1.13.x
> >>  10.1.13.3      node1          10.1.13.1
> >>10.1.23.x
> >>  10.1.23.3      node2          10.1.23.2
> >>10.1.34.x
> >>  10.1.34.3      node4          10.1.34.4
> >>
> >>network
> >>        host IP(node4)  dest node        dest IP
> >>10.1.14.x
> >>  10.1.14.4      node1          10.1.14.1
> >>10.1.24.x
> >>  10.1.24.4      node2          10.1.24.2
> >>10.1.34.x
> >>  10.1.34.4      node3          10.1.34.3
> >>
> >>The third set of numbers in the IP is the node to node number. Maybe
> >>this is more confusing...
> >>
> >>
> >>>
> >>>>What's the next step?
> >>>>
> >>>>Is there a way for a machine to ping itself from eth0 to eth1 (e.g. the
> >>>>onboard nic to a gig nic)?
> >>>>
> >>>>The nodes right now are vanilla redhat with the updates applied.
> >>>>
> >>>>Hugh
> >>>>
> > _______________________________________________
> > Web Page:  http://lug.boulder.co.us
> > Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> > Join us on IRC: lug.boulder.co.us port=6667 channel=#colug
> > 
> > 
> > 
> 
> 
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> Join us on IRC: lug.boulder.co.us port=6667 channel=#colug