|
Hello Jean Paul,
> I'm setting up a DR_LVS with a director and two servers...
> I've to handle the ARP problem so I've put two NIC on the two
> realservers...
Ok, this is one way to do it.
> Now I've a big trouble with one of them... The problem appear also
> before I get up all the stuff around the LVS... Simply I power-on
> the real server, I make linux start and I set up the two eth from
> command line.... I give them 192.168.17.10 and 192.168.10.30 as IP
> address. (.30 is the address for the VIP).
Which one is which?
> The eth1 is leaved floating and is not really connected to the
> switch/hub of the net...
Ok, so I assume, 192.168.10.30 is the VIP with netmask /32? Please, use
netmasks when referring to IP addresses. Better even to help debugging
is if you show an output of 'ip addr show'. I trust tools better then
people typing ;)
> The problem is: If I ping from another PC (chosen randomly) to
> 192.168.17.30 (eth1 floating) I receive a reply from 192.168.17.10
> (eth0 as ifconfig shows me!!!).
Why do you ping to the 192.168.17.30 interface, when your VIP is
192.168.10.30/32?
> This make no sense for me....
Neither does it to me right now. What happens if you disable ip_forward?
I'm just asking out of curiosity.
> ifconfig return that 192.168.17.10 is eth0 and 192.168.17.30 is eth1
> (floating)....
Well, depending on the netmask you might have the same C class and then
it is definitely routed through eth0. If you could provide us with some
nice dump of 'ip addr show' and 'ip ro show table main' I would be very
pleased.
> It seems that eth0 reply to the server instead of eth1!!!
I need to know the netmasks and you routing setup.
> If I hot swap the UTP between the 2 NIC I receive no reply at all...
;) Do you mean if you connect your network cable to eth1?
> but making ifconfig eth0 down and ping'ing 192.168.17.30 or
> 192.168.17.10 will cause eth1 to reply!!!
Yes, because eth0 is not possible anymore, link state down or route
state dead. If you have a netmask problem, which is what I think you
have, then as soon as eth0 is down, the route for eth0 is killed too and
eth1 has free way to get packets.
> Excuse me in the case this is not pertinet to the mailing list...
E la prima volta, non ti affochiamo! Salve,
Roberto Nibali, ratz
|
