difference between "ping -I INTERFACE" and "ping -I IPADDRESS" ?

[Frantisek Hanzlik]

Kernel Guardian wrote:
> On 29 September 2011 09:56, Frantisek Hanzlik <franta at hanzlici.cz> wrote:
> 
> How are you create rules? (maybe there is a problem?)
> what do you want to achieve with wifi* adsl table, load balancing or
> splitting access?

I tried create rules with Fedora convention - via appropriate files in
/etc/sysconfig/network-scripts/rule-NETDEV and
/etc/sysconfig/network-scripts/route-NETDEV

But, I not know how "in Fedora manner" set some more complicated things
as load balancing:

ip route add default scope global nexthop via 10.128.254.2 dev eth1 weight 1 \
 nexthop via 10.3.60.1 dev eth2 weight 1 nexthop dev ppp0 weight 1

etc. Reading /etc/sysconfig/network-scripts/ scripts it see as there is
possible custom post-initialization things with created /sbin/ifup-local
program/script, but it is called at end of device init - and there are
things which is suitable/necessary do after all interface starts, at end
of whole network initialization. Which probably isn't possible solve in
existing Fedora framework.

Regarding what I want achieve with networking at this Linux machine:
no great matter, DSL link is backup, and two wireless are main internet
connection. There are some light-loaded services accessed via different
links by their different FQDN (e.g. ftp.mydomain is one link,
mail.mydomain second). eth0 is LAN interface with several clients, maybe
sometimes there will be outcomming traffic splitted with some netfilter
packet marking.

All things on Linux machine works well, thus there is no need this case.
All is about as I until now was thinking, that "-I netdev" == "-I IP",
nothing else. And it isn't true.
(not sure when i clearly explain all, excuse my terrible english grammar)

Franta