Fedora 7

Olivier Galibert galibert at pobox.com
Fri Jan 5 16:00:37 UTC 2007


On Fri, Jan 05, 2007 at 10:46:49AM -0500, Dan Williams wrote:
> On Fri, 2007-01-05 at 15:54 +0100, Olivier Galibert wrote:
> > - eth0 has a static IP in 192.44.78/24, DNS, gw, etc, and also has a
> >   static route allowing to use the interface for 192.168.87.0/24 using
> >   its static IP which is not in 192.168.*
> 
> Can you explain the configuration for these in a bit more detail?  So in
> this example, eth0 has a static IP and DNS, gw, etc are _also_ in the
> same 192.44.78/24 network.  But you've added a route that pushes traffic
> for the 192.168.87.0/24 network over eth0 (which has no IP address in
> that network range), correct?

Yep.  These are the desktop boxes, which want to be able to talk to
the boxes on the internal network.

2: eth0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:08:74:a8:7c:f8 brd ff:ff:ff:ff:ff:ff
    inet 192.44.78.23/24 brd 192.44.78.255 scope global eth0
    inet6 fe80::208:74ff:fea8:7cf8/64 scope link 
       valid_lft forever preferred_lft forever

192.168.87.0/24 dev eth0  scope link 
192.44.78.0/24 dev eth0  proto kernel  scope link  src 192.44.78.23 
169.254.0.0/16 dev eth0  scope link 
default via 192.44.78.22 dev eth0 


> > - eth0 has a static IP in 192.44.78/24, DNS, gw. eth1 has an IP in
> >   192.168.87/24.
> 
> I think the config would cover that, no?  Each interface can obviously
> have different IP settings (addr, mask, bcast).  DNS is a bit different
> because the order of resolvers in resolv.conf determines some of the
> behavior, even if the routing rules determine which interface the
> requests against a particular server actually go out on.

Yeah, it's globally more standard.  It was to show that it's not a
given every interface has dns/nis/etc.  These are the servers which
are connected to both networks equally.

Note btw that the internal network has a (friggin' fast) secondary NIS
server, and the configuration is broadcast.  So most of the times
these boxes use the NIS server on eth1 rather than eth0.

2: eth0: <BROADCAST,MULTICAST,UP,10000> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:30:48:5a:c3:ee brd ff:ff:ff:ff:ff:ff
    inet 192.44.78.93/24 brd 192.44.78.255 scope global eth0
    inet6 fe80::230:48ff:fe5a:c3ee/64 scope link 
       valid_lft forever preferred_lft forever
3: eth1: <BROADCAST,MULTICAST,UP,10000> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:30:48:5a:c3:ef brd ff:ff:ff:ff:ff:ff
    inet 192.168.87.89/24 brd 192.168.87.255 scope global eth1
    inet6 fe80::230:48ff:fe5a:c3ef/64 scope link 
       valid_lft forever preferred_lft forever

192.168.87.0/24 dev eth1  proto kernel  scope link  src 192.168.87.89 
192.44.78.0/24 dev eth0  proto kernel  scope link  src 192.44.78.93 
169.254.0.0/16 dev eth1  scope link 
224.0.0.0/4 dev eth1  scope link 
default via 192.44.78.22 dev eth0 


> What does the output of /sbin/route -n look like for this case?

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.87.0    0.0.0.0         255.255.255.0   U     0      0        0 eth1
192.44.78.0     0.0.0.0         255.255.255.0   U     0      0        0 eth0
169.254.0.0     0.0.0.0         255.255.0.0     U     0      0        0 eth1
224.0.0.0       0.0.0.0         240.0.0.0       U     0      0        0 eth1
0.0.0.0         192.44.78.22    0.0.0.0         UG    0      0        0 eth0


> > - eth0 and eth1 have the same static IP in 192.168.87/24, eth1 has the
> >   route for 192.168.87/24, eth0 has only a route for 192.44.78/24.
> 
> I'm not sure the configuration mentioned above excludes this
> possibility, can you be more specific on how you couldn't configure your
> interfaces this way in the proposed config structure?  (keeping in mind
> that gateway should be handled differently than the config structure
> says)

I'm not sure it doesn't, I'm not sure it does, but the orthogonality
between address and route is interesting.  These are the internal
boxes, with no gw access (they're not routable anyway).

2: eth0: <BROADCAST,MULTICAST,UP,10000> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:14:5e:86:81:58 brd ff:ff:ff:ff:ff:ff
    inet 192.168.87.171/24 brd 192.168.87.255 scope global eth0
    inet6 fe80::214:5eff:fe86:8158/64 scope link 
       valid_lft forever preferred_lft forever
3: eth1: <BROADCAST,MULTICAST,UP,10000> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:14:5e:86:81:59 brd ff:ff:ff:ff:ff:ff
    inet 192.168.87.171/24 scope global eth1
    inet6 fe80::214:5eff:fe86:8159/64 scope link 
       valid_lft forever preferred_lft forever

192.168.87.0/24 dev eth1  proto kernel  scope link  src 192.168.87.171 
192.44.78.0/24 dev eth0  scope link 
169.254.0.0/16 dev eth1  scope link 
224.0.0.0/4 dev eth1  scope link 

  OG.




More information about the fedora-devel-list mailing list