[H-GEN] IP Aliasing and traffic shapers

[David Jericho]

-----BEGIN PGP SIGNED MESSAGE-----


Thinking about the amount of bandwidth we horrible quakers consume at
meetings, would it be possible to put a traffic shaper on a device that is
simply an alias? 

For example, meesha is on 192.168.105.69, using eth1. It then has .68
aliased to device eth1:0. Now I apply a traffic shaper to eth1:0 for a
low bps rate. While I don't see why this wouldn't work (as well as
save everybody else getting annoyed at us :), are there any hitches that
people see? I can force the quake server to respond to only one ip.

Also, has anyone with a home network and a masq connection happening
played with name lookup masquerading? We had it working briefly but for
some reason for another it screwed up not being able to do a name lookup 
on proxy.uq.edu.au from Byron's machine (I think my machine was
having a fit).  The benefit with name lookup masq'ing is that people who
don't use UQNet when at home (i.e. not using krefti and cuscus as name
servers) can do name lookups without any changes to config files. 

To expand a bit more, I had the transparent proxy rules for
192.168.105.0/24 and 192.168.105.107/24 to 0.0.0.0/0 on ports 42/udp,
53/tcp and 53/udp redirected to the localhost. I had a caching name server
on the localhost forwarding to krefti and cuscus.

Is hydra going to be back at the next meeting? With three network cards?
Are there any objections from prentice if we use both humbug ips at the
same time? If not, I propose a setup such as below.

                         ---------
  --- 130.102.165.43 --- | hydra | --- 192.168.105.0/24 --- [ router ] --
	ftp/telnet	 ---------                                      |
			     |                                         ??? 
	                 ----------                                     |
  --- 130.102.165.21 --- | daffy? | --- machines running squid ----------
           http          ---------- 

I propose this as James said at the last meeting that he wanted to setup
a cache as well. If we have a group of machines present that are running
caches there is no point wasting them. And it lets us take caches and so
on up and down freely so those people can do other things (i.e. prevent
namely zer0 from crashing the router). At least we'd have to try hard to
exceed host/cable or cable length restrictions.

\/   David Jericho - Still dreaming about ATM switches
 \/   Finger s343974 at student.uq.edu.au for PGP key
  \/   Happiness isn't having what you want, it's wanting what you have. 


-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
Charset: noconv

iQCVAwUBM7mo7r4XfRuE8i8NAQGtawQAmUEgqaXhZhpoKiAipsv/C8zzdJJNI07Z
91x6s98u3wadJfz5qUEd7NybktObkHw2mHuGpYiBRzRJFSC4fwAsmbZgIrXWIo6o
GsT84bADCed0QnDDxDWfi3dkJc5w9BXavH5UyvZbdO+sPGb0n/SdsYJXhNdrpw/4
Z1fYQGbowGY=
=JhF8
-----END PGP SIGNATURE-----

----------------------- HUMBUG General List --------------------------------
echo "unsubscribe general" | mail majordomo at humbug.org.au # To Unsubscribe