nexthop via (Re: bonding nezvysi rychlost prenosu?)

[David Rohleder]

trcka na poda.cz (David Trcka) writes:

> On 28 Nov 2001, David Rohleder wrote:
> 
> > A navic network unreachable je podstatne lepsi hlaska nez RTNETLINK:
> > invalid argument. IMHO ten prikaz prosel syntaktickym analyzatorem ip
> > route, ale pak narazil na neexistenci routy na <gw1> nebo <gw2>
> > 
> Nenarazil:
> 
> [root na dave david]# /sbin/ip r
> 62.168.19.64/26 dev eth0  proto kernel  scope link  src 62.168.19.66 
> 127.0.0.0/8 dev lo  scope link 
> default via 62.168.19.65 dev eth0 
> [root na dave david]# /sbin/ip r a 10.0.0.0/24 nexthop via 62.168.19.67 
> nexthop via 62.168.19.68
> RTNETLINK answers: Network is unreachable
> 
> Musel narazit na neco jineho. Ale jak to zjistit? Z vypisu strace moc 
> moudry nejsem, protoze se pouziva komunikace pres netlink socket (jsou tam 
> same sendmsg,recvmsg) a konci to na
> 

Tak ted mi to opravdu nedalo, prelozil jsem si jadro s podporou
multipath a zkusil

ip route add 10.0.0.0/16 nexthop via a.b.c.d dev eth0 nexthop via
f.g.h.j dev eth1 

a v pohode to sezral:

# ip route ls

a.b.c.16/28 dev eth1  proto kernel  scope link  src a.b.c.e
f.g.h.0/24 dev eth0  proto kernel  scope link  src f.g.h.i
10.0.0.0/16 
	nexthop via a.b.c.d  dev eth1 weight 1 dead
	nexthop via f.g.h.j  dev eth0 weight 1

Jedna cesta je sice dead, ale je to jenom pokus, ktery jen tak nemuze
fungovat. Ale evidentne ten prikaz sezral.

Mozna nam tam predtim chybela definice dev ethX


-- 
-------------------------------------------------------------------------
David Rohleder						davro na ics.muni.cz
Institute of Computer Science, Masaryk University
Brno, Czech Republic
-------------------------------------------------------------------------