[lxc-users] Networking in Ubuntu with 2 ip failover in LXC

[m.byryn1u]

W dniu 2014-08-13 17:24, Fajar A. Nugraha pisze:
> On Wed, Aug 13, 2014 at 10:08 PM, m.byryn1u <m.bryn1u at gmail.com> wrote:
>> W dniu 2014-08-13 16:57, Fajar A. Nugraha pisze:
>>
>>> On Wed, Aug 13, 2014 at 7:34 PM, Tamas Papp <tompos at martos.bme.hu> wrote:
>>>> Resending..
>>>>
>>>>
>>>> On 08/13/2014 02:27 PM, Tamas Papp wrote:
>>>>
>>>> host machine:
>>>>
>>>> auto br0
>>>> iface br0 inet static
>>>>           bridge_ports eth0
>>>>           bridge_stp off
>>>>           bridge_maxwait 0
>>>>           bridge_fd 0
>>>>           address 94.23.237.216
>>>>           netmask 255.255.255.0
>>>>           network 94.23.237.0
>>>>           broadcast 94.23.237.255
>>>>           gateway 94.23.237.254
>>>>
>>>>
>>>> Don't setup br0:0
>>>>
>>>>
>>>> Remove the following entries from CONTAINER/config:
>>>>
>>>> lxc.network.ipv4 = 91.121.239.228/32
>>>> lxc.network.ipv4.gateway = 91.121.239.254
>>>>
>>>>
>>>>
>>>> guest machine:
>>>>
>>>> auto eth0
>>>> iface eth0 inet static
>>>>       address 91.121.239.228
>>>>       netmask 255.255.255.0
>>>>       gateway 94.23.237.254
>>>>
>>>>
>>>> This should work.
>>> No, that won't work.
>>>
>>> 94.23.237.254 is not part of 91.121.239.0/24.
>>>
>>> @Bryn, how is the ROUTER (i.e. 94.23.237.254) setup? Is it configured
>>> to route the additional IP (91.121.239.228) thru host's IP
>>> (94.23.237.216), the way some dedicated server provider does (e.g.
>>> serverloft)?
>>>
>>> If yes, then the EASY way would be to put 91.121.239.228 as an alias
>>> in host's interface (I'd just use eth0, no need to use a bridge there)
>>> and setup a static NAT to whatever IP the container has (e.g.
>>> 10.0.3.251, connected to host's lxcbr0 bridge)
>>>
>> Hey,
>>
>> It's server dedicated in OVH.
> Then ask OVH how to use that IP.
>
>> I have one ip like  94.23.237.216 and i bought one more called "ip failover"
>> like 91.121.239.228. You say as an alias and setup static NAT. But what
>> about 2 ip failover ?
>
> If my guess is right, it's similar to serverloft. They will say
> "simply put it as an IP alias/additional IP on your server".
>
> As in, the additional IP is routed to ONE of your server's IP.
> Permanently. Can't be used on other server. Thus, there can be NO
> failover.
>
> It's not a standard failover setup where two or more physical servers
> each have an IP in the same network segment, and you can have one or
> more virtual IP for your services that can fail over to any of the
> servers.
>
I had a server on FreeBSD-10 before and i had 2 jails with 1 failover ip 
per jail, work well.
OVH says :

post-up /sbin/ifconfig eth0:X IP.FAIL.OVER netmask 255.255.255.255 
broadcast IP.FAIL.OVER
post-down /sbin/ifconfig eth0:X down

Or as bridge:


/etc/network/interfaces
auto lo eth0
iface lo inet loopback
iface eth0 inet static
addressIP.FAIL.OVER
netmask 255.255.255.255
broadcastIP.FAIL.OVER
post-up route addIP..SERWERA.254dev eth0
post-up route add default gwIP.SERWERA.254
post-down route delIP..SERWERA.254dev eth0
post-down route del default gwIP.SERWERA.254

That's why i think that method 2 post befor should works.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxcontainers.org/pipermail/lxc-users/attachments/20140813/60120d82/attachment.html>