can't set default route for virtual interface

can't set default route for virtual interface

[Lloyd Standish]

Hi All,

I'm setting up multiple virtual interfaces to allow load-balancing over several uplinks using a single NIC, on Debian Squeeze.  Packets will be marked by iptables, and sent to the proper routing tables via "ip rule add fwmark <n>".

I'm having trouble setting routes for the VLANs.

root@debiandesk2:/etc/network# ifconfig

192.168.100.255  Mask:255.255.255.0
            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
            RX packets:49654 errors:0 dropped:0 overruns:0 frame:0
            TX packets:54712 errors:0 dropped:0 overruns:0 carrier:0
            collisions:0 txqueuelen:1000
            RX bytes:35989170 (34.3 MiB)  TX bytes:8795113 (8.3 MiB)
            Interrupt:27 Base address:0xc000

eth1      Link encap:Ethernet  HWaddr 00:08:54:dc:1b:79
            inet addr:192.168.10.1  Bcast:192.168.10.255  Mask:255.255.255.0
            inet6 addr: fe80::208:54ff:fedc:1b79/64 Scope:Link
            UP BROADCAST MULTICAST  MTU:1500  Metric:1
            RX packets:6077 errors:0 dropped:0 overruns:0 frame:0
            TX packets:9451 errors:0 dropped:0 overruns:0 carrier:73
            collisions:0 txqueuelen:1000
            RX bytes:821439 (802.1 KiB)  TX bytes:9756873 (9.3 MiB)
            Interrupt:19 Base address:0xdc00

eth0.2    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
            inet addr:192.168.200.10  Bcast:192.168.200.255  Mask:255.255.255.0
            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
            RX packets:17776 errors:0 dropped:0 overruns:0 frame:0
            TX packets:15037 errors:0 dropped:1 overruns:0 carrier:0
            collisions:0 txqueuelen:0
            RX bytes:16258598 (15.5 MiB)  TX bytes:2075870 (1.9 MiB)

eth0.3    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
            inet addr:192.168.2.10  Bcast:192.168.2.255  Mask:255.255.255.0
            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
            RX packets:1083 errors:0 dropped:0 overruns:0 frame:0
            TX packets:1320 errors:0 dropped:0 overruns:0 carrier:0
            collisions:0 txqueuelen:0
            RX bytes:113499 (110.8 KiB)  TX bytes:273602 (267.1 KiB)

My firewall/load-balancing script is able to set the routes for the first virtual interface (eth0.2) OK:

+ ip route flush table T0
+ ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10 table T0
+ ip rule add from 192.168.200.10 table T0
+ ip rule add iif lo to 201.200.105.6 table T0
+ ip route add 192.168.10.0/24 dev eth1 table T0
+ ip route add 127.0.0.0/8 dev lo table T0
+ ip rule del from all fwmark 0
+ ip rule add fwmark 0 table T0
+ ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10
+ ip route add default via 192.168.200.1 table T0

However, it fails as follows when adding the default route for the second virtual interface (eth0.3):

+ ip route flush table T1
+ ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10 table T1
+ ip rule add from 192.168.2.10 table T1
+ ip rule add iif lo to 8.8.4.4 table T1
+ ip route add 192.168.10.0/24 dev eth1 table T1
+ ip route add 127.0.0.0/8 dev lo table T1
+ ip rule del from all fwmark 1
+ ip rule add fwmark 1 table T1
+ ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10
+ ip route add default via 192.168.2.1 table T1
RTNETLINK answers: No such process

Similarly, I was unable to use "ip" to set a "nexthop via" default route for virtual interfaces.

I'd greatly appreciate any help.

Regards,
Lloyd

Re: can't set default route for virtual interface

[Andrew Beverley]

On Tue, 2011-12-27 at 17:32 -0600, Lloyd Standish wrote:
> + ip route add default via 192.168.2.1 table T1
> RTNETLINK answers: No such process

This normally means that the kernel doesn't think that 192.168.2.1 is a
valid address. Is that IP address definitely set up by that point in the
script?

Have you tried *not* using ifconfig to set up your interfaces? You
should only be using iproute2.

> Similarly, I was unable to use "ip" to set a "nexthop via" default route for virtual interfaces.
> 

Same probably applies to this.

Andy

Re: can't set default route for virtual interface

[Brian Austin - Standard Universal]

table T1 needs this first?

ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10 table T1




On 28/12/2011 10:32 AM, Lloyd Standish wrote:
> Hi All,
>
> I'm setting up multiple virtual interfaces to allow load-balancing 
> over several uplinks using a single NIC, on Debian Squeeze.  Packets 
> will be marked by iptables, and sent to the proper routing tables via 
> "ip rule add fwmark <n>".
>
> I'm having trouble setting routes for the VLANs.
>
> root@debiandesk2:/etc/network# ifconfig
>
> 192.168.100.255  Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:49654 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:54712 errors:0 dropped:0 overruns:0 carrier:0
>            collisions:0 txqueuelen:1000
>            RX bytes:35989170 (34.3 MiB)  TX bytes:8795113 (8.3 MiB)
>            Interrupt:27 Base address:0xc000
>
> eth1      Link encap:Ethernet  HWaddr 00:08:54:dc:1b:79
>            inet addr:192.168.10.1  Bcast:192.168.10.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::208:54ff:fedc:1b79/64 Scope:Link
>            UP BROADCAST MULTICAST  MTU:1500  Metric:1
>            RX packets:6077 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:9451 errors:0 dropped:0 overruns:0 carrier:73
>            collisions:0 txqueuelen:1000
>            RX bytes:821439 (802.1 KiB)  TX bytes:9756873 (9.3 MiB)
>            Interrupt:19 Base address:0xdc00
>
> eth0.2    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
>            inet addr:192.168.200.10  Bcast:192.168.200.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:17776 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:15037 errors:0 dropped:1 overruns:0 carrier:0
>            collisions:0 txqueuelen:0
>            RX bytes:16258598 (15.5 MiB)  TX bytes:2075870 (1.9 MiB)
>
> eth0.3    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
>            inet addr:192.168.2.10  Bcast:192.168.2.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:1083 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:1320 errors:0 dropped:0 overruns:0 carrier:0
>            collisions:0 txqueuelen:0
>            RX bytes:113499 (110.8 KiB)  TX bytes:273602 (267.1 KiB)
>
> My firewall/load-balancing script is able to set the routes for the 
> first virtual interface (eth0.2) OK:
>
> + ip route flush table T0
> + ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10 table T0
> + ip rule add from 192.168.200.10 table T0
> + ip rule add iif lo to 201.200.105.6 table T0
> + ip route add 192.168.10.0/24 dev eth1 table T0
> + ip route add 127.0.0.0/8 dev lo table T0
> + ip rule del from all fwmark 0
> + ip rule add fwmark 0 table T0
> + ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10
> + ip route add default via 192.168.200.1 table T0
>
> However, it fails as follows when adding the default route for the 
> second virtual interface (eth0.3):
>
> + ip route flush table T1
> + ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10 table T1
> + ip rule add from 192.168.2.10 table T1
> + ip rule add iif lo to 8.8.4.4 table T1
> + ip route add 192.168.10.0/24 dev eth1 table T1
> + ip route add 127.0.0.0/8 dev lo table T1
> + ip rule del from all fwmark 1
> + ip rule add fwmark 1 table T1
> + ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10
> + ip route add default via 192.168.2.1 table T1
> RTNETLINK answers: No such process
>
> Similarly, I was unable to use "ip" to set a "nexthop via" default 
> route for virtual interfaces.
>
> I'd greatly appreciate any help.
>
> Regards,
> Lloyd
> -- 
> To unsubscribe from this list: send the line "unsubscribe netfilter" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: can't set default route for virtual interface

[Brian Austin - Standard Universal]

or add the dev to the failed command? (yes I saw you already had that 
other command doh)

ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10 table T1



On 28/12/2011 10:32 AM, Lloyd Standish wrote:
> Hi All,
>
> I'm setting up multiple virtual interfaces to allow load-balancing 
> over several uplinks using a single NIC, on Debian Squeeze.  Packets 
> will be marked by iptables, and sent to the proper routing tables via 
> "ip rule add fwmark <n>".
>
> I'm having trouble setting routes for the VLANs.
>
> root@debiandesk2:/etc/network# ifconfig
>
> 192.168.100.255  Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:49654 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:54712 errors:0 dropped:0 overruns:0 carrier:0
>            collisions:0 txqueuelen:1000
>            RX bytes:35989170 (34.3 MiB)  TX bytes:8795113 (8.3 MiB)
>            Interrupt:27 Base address:0xc000
>
> eth1      Link encap:Ethernet  HWaddr 00:08:54:dc:1b:79
>            inet addr:192.168.10.1  Bcast:192.168.10.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::208:54ff:fedc:1b79/64 Scope:Link
>            UP BROADCAST MULTICAST  MTU:1500  Metric:1
>            RX packets:6077 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:9451 errors:0 dropped:0 overruns:0 carrier:73
>            collisions:0 txqueuelen:1000
>            RX bytes:821439 (802.1 KiB)  TX bytes:9756873 (9.3 MiB)
>            Interrupt:19 Base address:0xdc00
>
> eth0.2    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
>            inet addr:192.168.200.10  Bcast:192.168.200.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:17776 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:15037 errors:0 dropped:1 overruns:0 carrier:0
>            collisions:0 txqueuelen:0
>            RX bytes:16258598 (15.5 MiB)  TX bytes:2075870 (1.9 MiB)
>
> eth0.3    Link encap:Ethernet  HWaddr 20:cf:30:d8:6b:ea
>            inet addr:192.168.2.10  Bcast:192.168.2.255  
> Mask:255.255.255.0
>            inet6 addr: fe80::22cf:30ff:fed8:6bea/64 Scope:Link
>            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>            RX packets:1083 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:1320 errors:0 dropped:0 overruns:0 carrier:0
>            collisions:0 txqueuelen:0
>            RX bytes:113499 (110.8 KiB)  TX bytes:273602 (267.1 KiB)
>
> My firewall/load-balancing script is able to set the routes for the 
> first virtual interface (eth0.2) OK:
>
> + ip route flush table T0
> + ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10 table T0
> + ip rule add from 192.168.200.10 table T0
> + ip rule add iif lo to 201.200.105.6 table T0
> + ip route add 192.168.10.0/24 dev eth1 table T0
> + ip route add 127.0.0.0/8 dev lo table T0
> + ip rule del from all fwmark 0
> + ip rule add fwmark 0 table T0
> + ip route add 192.168.200.0/24 dev eth0.2 src 192.168.200.10
> + ip route add default via 192.168.200.1 table T0
>
> However, it fails as follows when adding the default route for the 
> second virtual interface (eth0.3):
>
> + ip route flush table T1
> + ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10 table T1
> + ip rule add from 192.168.2.10 table T1
> + ip rule add iif lo to 8.8.4.4 table T1
> + ip route add 192.168.10.0/24 dev eth1 table T1
> + ip route add 127.0.0.0/8 dev lo table T1
> + ip rule del from all fwmark 1
> + ip rule add fwmark 1 table T1
> + ip route add 192.168.2.0/24 dev eth0.3 src 192.168.2.10
> + ip route add default via 192.168.2.1 table T1
> RTNETLINK answers: No such process
>
> Similarly, I was unable to use "ip" to set a "nexthop via" default 
> route for virtual interfaces.
>
> I'd greatly appreciate any help.
>
> Regards,
> Lloyd
> -- 
> To unsubscribe from this list: send the line "unsubscribe netfilter" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: can't set default route for virtual interface

[Lloyd Standish]

On Tue, 27 Dec 2011 18:42:02 -0600, Brian Austin - Standard Universal <brian@standarduniversal.com.au> wrote:

> or add the dev to the failed command? (yes I saw you already had that other command doh)

Thank you - I added that and the route was added successfully!

--
Lloyd

Re: can't set default route for virtual interface

[Lloyd Standish]

On Tue, 27 Dec 2011 17:43:01 -0600, Andrew Beverley <andy@andybev.com> wrote:

> On Tue, 2011-12-27 at 17:32 -0600, Lloyd Standish wrote:
>> + ip route add default via 192.168.2.1 table T1
>> RTNETLINK answers: No such process
>This normally means that the kernel doesn't think that 192.168.2.1 is a
> valid address. Is that IP address definitely set up by that point in the
> script?

Thanks for the heads-up - yes, 192.168.2.1 was set-up.  After adding the "dev" as Brian Austin suggested, the route was added. :)

>Have you tried *not* using ifconfig to set up your interfaces? You
> should only be using iproute2.

I've seen that mentioned before, and I'm confused.  iproute2 means, for firewall code, "ip" and "tc" - is that right?  I usually bring up interfaces with "ifup" when I can, since that sets the interface address specified in /etc/network/interfaces, whereas "ifconfig <if> up" doesn't.

I notice that after adding the Debian "vlan" package, ifup and ifdown appear to bring virtual interfaces up and down OK.  But ifup/down are not part of iproute2, right?  Do you know of any vlan problem with  ifup/down?  Sometimes interfaces can't be taken down with "ifdown," and I have to use ifconfig.

--
Lloyd

Re: can't set default route for virtual interface

[Andrew Beverley]

On Tue, 2011-12-27 at 23:36 -0600, Lloyd Standish wrote:
> >Have you tried *not* using ifconfig to set up your interfaces? You
> > should only be using iproute2.
> 
> I've seen that mentioned before, and I'm confused.  iproute2 means, for
>  firewall code, "ip" and "tc" - is that right?

Yes, and "ip" can do everything that all the various net-tools commands
do.

>   I usually bring up interfaces with "ifup" when I can

Ah, therein lies an interesting situation. ifup is pretty old too, and
makes calls to the ifconfig command, which is a real pain. However,
unless someone tells me otherwise, I'm not aware of a replacement for
it. There was a project a while ago called netconf, but this seems to
have died:

http://wiki.debian.org/netconf

> , since that sets the interface address specified in /etc/network/interfaces, whereas "ifconfig <if> up" doesn't.

That's the best way to do it with Debian.

> I notice that after adding the Debian "vlan" package, ifup and ifdown
>  appear to bring virtual interfaces up and down OK.  But ifup/down are
>  not part of iproute2, right?

Yeah, correct. The most accepted way of doing it seems to be to use ifup
and ifdown to bring the physical interface up and down, and to do
everything else by adding "ip" commands into
your /etc/network/interfaces file using the
up/down/pre-up/pre-down/post-up/post-down options.

>   Do you know of any vlan problem with 
>  ifup/down?  Sometimes interfaces can't be taken down with "ifdown,"
>  and I have to use ifconfig.

I get similar problems when adding IPV6 addresses, although I use "ip"
to sort it out manually ;-)

Anybody know what happened to netconf?

Andy

Re: can't set default route for virtual interface

[Amos Jeffries]

On 28/12/2011 11:21 p.m., Andrew Beverley wrote:
> On Tue, 2011-12-27 at 23:36 -0600, Lloyd Standish wrote:
>>> Have you tried *not* using ifconfig to set up your interfaces? You
>>> should only be using iproute2.
>> I've seen that mentioned before, and I'm confused.  iproute2 means, for
>>   firewall code, "ip" and "tc" - is that right?
> Yes, and "ip" can do everything that all the various net-tools commands
> do.
>
>>    I usually bring up interfaces with "ifup" when I can
> Ah, therein lies an interesting situation. ifup is pretty old too, and
> makes calls to the ifconfig command, which is a real pain. However,
> unless someone tells me otherwise, I'm not aware of a replacement for
> it.

That would be ...

   ip link set dev eth0 up
   ip link set dev eth0 down

   :)

AYJ

Re: can't set default route for virtual interface

[Andrew Beverley]

On Fri, 2011-12-30 at 00:27 +1300, Amos Jeffries wrote:
> On 28/12/2011 11:21 p.m., Andrew Beverley wrote:
> > On Tue, 2011-12-27 at 23:36 -0600, Lloyd Standish wrote:
> >>    I usually bring up interfaces with "ifup" when I can
> > Ah, therein lies an interesting situation. ifup is pretty old too, and
> > makes calls to the ifconfig command, which is a real pain. However,
> > unless someone tells me otherwise, I'm not aware of a replacement for
> > it.
> 
> That would be ...
> 
>    ip link set dev eth0 up
>    ip link set dev eth0 down

Yes :-)

Although that only really replaces the ifconfig part. It would be nice
to have a proper replacement for ifup/ifdown that does the rest of the
networking stuff, such as setting the ip address and calling scripts, as
is contained in /etc/network/interfaces. It'll also call pppd and so on
as well.

Andy

Re: can't set default route for virtual interface

[Jan Engelhardt]

On Wednesday 2011-12-28 11:21, Andrew Beverley wrote:
>On Tue, 2011-12-27 at 23:36 -0600, Lloyd Standish wrote:
>> >Have you tried *not* using ifconfig to set up your interfaces? You
>> > should only be using iproute2.
>> 
>> I've seen that mentioned before, and I'm confused.  iproute2 means, for
>>  firewall code, "ip" and "tc" - is that right?
>
>Yes, and "ip" can do everything that all the various net-tools commands
>do.

...and more!

>>   I usually bring up interfaces with "ifup" when I can
>
>Ah, therein lies an interesting situation. ifup is pretty old too, and
>makes calls to the ifconfig command, which is a real pain.

More precisely: ifup is a distribution-specific script, not to be
confused with net-tools/iproute2 (different layer). In that regard,
its age may not mean anything.

ifup also refers to _multiple_ scripts; for example, openSUSE's ifup
uses iproute - since a looong time (therefore it is also old, but at
least it is current).