Re: [vserver] source IP messed up on multihomed guest with 2.6.38.6-vs2.3.0.37-rc15

Thomas Weber wrote:
> Am Samstag, den 28.05.2011, 10:17 +0200 schrieb Daniel Hokka Zakrisson:
>
>> echo 255 > /proc/sys/vserver/debug_net
>> and rerun your tests, and paste the kernel output.
>
> I had to redo the tests on a different box (original one is on site
> now).
> Anyway, the config:
>
> host:
> # ip addr ls eth0
> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
> state UP qlen 1000
> link/ether 00:25:90:34:4f:6a brd ff:ff:ff:ff:ff:ff
> inet 192.168.55.200/24 brd 192.168.55.255 scope global eth0
> inet 192.168.1.51/24 brd 192.168.1.255 scope global eth0:aby
>
> guest: (dev, ip, prefix)
> # cat interfaces/01/*
> eth0
> 192.168.55.68
> 24
>
> # cat interfaces/10/*
> eth0
> 192.168.1.68
> 24
>
> with guest running:
> # ip addr ls eth0
> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
> state UP qlen 1000
> link/ether 00:25:90:34:4f:6a brd ff:ff:ff:ff:ff:ff
> inet 192.168.55.200/24 brd 192.168.55.255 scope global eth0
> inet 192.168.1.51/24 brd 192.168.1.255 scope global eth0:aby
> inet 192.168.55.68/24 brd 192.168.55.255 scope global secondary eth0
> inet 192.168.1.68/24 brd 192.168.1.255 scope global secondary eth0
>
>
> what i did inside the guest:
> # ping -n 192.168.1.10
>
> Attached the logs of both 2 kernel versions, the 2.6.37* one is the
> working one.
> I'd say the netmasks in v4_addr_match / ip_v4_find_src are incorrect for
> the 2.6.38* one.

Yeah, exactly. http://people.linux-vserver.org/~dhozac/p/k/delta-ipv4src-fix02.diff
should fix your issue.

> I'd guess this info is enough but let me know if you need anything
> further.

-- 
Daniel Hokka Zakrisson