[Linux-cluster] Two node NFS cluster serving multiple networks

Thu Mar 13 16:23:34 UTC 2008

Is there a good document somewhere which explains in not too great technical 
terms how to use multiple nics on a system. I've been running bonded nics for 
many years but getting a machine to use two (or more networks) is still a 
mystery to me.

For example, I have a VoIP machine which has two nics which I have problems 
with because I don't understand the above yet.

This machine has a nic allows incoming VoIP/ZIP connections to it's public IP 
address on a T1. The router blocks everything but that traffic.

Then it has a second nic which has a private IP on it to allow for management 
of the machine. Yet recently, it lost it's DNS, it can't seem to get access to 
DNS on it's own. I can force it to use DNS by typing ping commands a couple of 
times but it cannot do it on it's own to get it's updates for example. 

Basically, I need the machine to see it's public gateway at xx.x.237.59 to 
route it's VoIP/SIP traffic but I also need it to see it's private gateway at 
192.168.1.0 so that it can use DNS and other internal services properly.

route -n
Kernel IP routing table
Destination   Gateway      Genmask            Flags Metric Ref    Use Iface
xx.x.237.56   0.0.0.0        255.255.255.248 U     0      0        0 eth0
192.168.1.0  0.0.0.0        255.255.255.0    U     0      0        0 eth1
169.254.0.0  0.0.0.0        255.255.0.0        U     0      0        0 eth1
0.0.0.0         69.2.237.57   0.0.0.0             UG    0      0        0 eth0

ifconfig
eth0      Link encap:Ethernet  HWaddr 00:90:27:DC:4B:E6
          inet addr:xx.x.237.59  Bcast:69.2.237.63  Mask:255.255.255.248
          inet6 addr: fe80::290:27ff:fedc:4be6/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:33910280 errors:16 dropped:0 overruns:0 frame:16
          TX packets:45988648 errors:0 dropped:0 overruns:0 carrier:0
          collisions:24746 txqueuelen:1000
          RX bytes:681966199 (650.3 MiB)  TX bytes:1657358619 (1.5 GiB)

eth1      Link encap:Ethernet  HWaddr 00:13:20:55:D7:CE
          inet addr:192.168.1.102  Bcast:192.168.1.255  Mask:255.255.255.0
          inet6 addr: fe80::213:20ff:fe55:d7ce/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:87417784 errors:0 dropped:0 overruns:0 frame:0
          TX packets:70881957 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:4171601084 (3.8 GiB)  TX bytes:1547562481 (1.4 GiB)

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:6501004 errors:0 dropped:0 overruns:0 frame:0
          TX packets:6501004 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:897257336 (855.6 MiB)  TX bytes:897257336 (855.6 MiB)

Mike

On Wed, 12 Mar 2008 10:39:50 -0700, Alex Kompel wrote:
> You will still need some way to tell the system through which
> 
> interface you want to route outgoing packets for each target.
> You can achieve the same with greater ease by splitting the network in
> 2 subnets and assigning each to a single interface.
> It all depends on the problem you are trying to solve. If you want
> redundancy - use active-passive bonding, you want throughput - use
> active-active bonding (if your switch supports link aggregation), if
> you want security and isolation - use separate subnets.
> 
> -Alex
> 
> 2008/3/12 Brian Kroth <bpkroth at wisc.edu>:
>> This is a hypothetical, but what if you have two interfaces on the same
>> network and want to force one service IP to one interface and the other
>> to a different interface?  I think what everyone is wondering is how
>> much control one has over the service IP placement.
>> 
>> Thanks,
>> Brian
>> 
>> Finnur Örn Guðmundsson - TM Software <fog at t.is> 2008-03-12 14:36:
>> 
>> 
>>> Hi,
>>> 
>>> I see no reason why you could not have 3 diffrent interfaces, each
>>> connected to the networks you are trying to serve the NFS requests
>>> to/from. RG Manager will add the floating interfaces to the "correct"
>>> interface, that is, if your floating ip is 1.2.3.4 and you have a
>>> interface with the IP address 1.2.3.3 he will add the IP to that
>>> interface.
>>> 
>>> 
>>> Bgrds,
>>> Finnur
>>> 
>>> -----Original Message-----
>>> From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-
>>> bounces at redhat.com] On Behalf Of gordan at bobich.net
>>> Sent: 12. mars 2008 14:10
>>> To: linux clustering
>>> Subject: Re: [Linux-cluster] Two node NFS cluster serving multiple
>>> networks
>>> 
>>> Sounds very similar to what I'm trying to achieve (see the other thread
>>> about binding failover resources to interfaces). I've not seen a
>>> response
>>> yet, so I'm most curious to see if you'll get any.
>>> 
>>> Gordan
>>> 
>>> On Wed, 12 Mar 2008, Randy Brown wrote:
>>> 
>>>> I am using a two node cluster with Centos 5 with up to date patches.  
>>>> We have
>>>> three different networks to which I would like to serve nfs mounts
>>>> from this
>>>> cluster.  Can this even be done?  I have interfaces available for each
>>>> network in each node?
>>>> 
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>> 
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>> 
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>> 
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster