Linux Netfilter discussions
 help / color / mirror / Atom feed
From: Curt Brune <curt@acm.org>
To: netfilter@vger.kernel.org
Subject: ebtables and iptables and NAT questions
Date: Sun, 2 Mar 2008 18:23:57 -0800	[thread overview]
Message-ID: <20080303022357.GA21681@curtisb.com> (raw)

Hi -

First off -- my sincere thanks to all the folks who work on these
tools.  I been using ipchains/iptables for nearly 10 years now.

Scenario: I have your basic dual NIC Linux box that is my home
firewall box.  One NIC connected to the DSL modem, one NIC for the
interal 10.0.0.0/24 private network.  Using iptables and NAT with no
problems for years.

I get the wild hair up my butt that I want to play with xen and
virtual machines -- with not too much space in my place I figure I
just do it on my firewall box (please no hissing from the audience).

BTW -- I have already read the xen networking guide
http://wiki.xensource.com/xenwiki/XenNetworking and the
"ebtables/iptables interaction on a Linux based bridge"
http://ebtables.sourceforge.net/br_fw_ia/br_fw_ia.html

Very interesting, but I'm still stuck.

As you may know xen uses bridging to connect the VMs to the physical
network.  I soon discovered that creating the bridge interface kills
my NAT.

Setup:

eth0 -- DSL modem
eth1 -- internal 10.0.0.0/24 network.  Also particpates in the xen
bridge, xenbr0.

My test is to try to ping my ISP's gateway from a second host, host2.
ASCII art:

+-----------+    +---------------------------+      +-------------+
| host2     |    |      Firewall Box         |      | ISP         |
| 10.0.0.22 |    |                           |      | 64.81.XX.XX |
|           +--->|eth1                   eth0+----->|             |
|           |    |                           |      |             |
+-----------+    +---------------------------+      +-------------+

The xen bridge script creates a bridge, xenbr0, and renames physical
eth1 to peth1 -- eth1 becomes a virtual interface that is connected to
dom0's vif0.0 interface.  The upshot is peth1, vif0.0 and any domU
virtual interfaces belong to the bridge, xenbr0.

I added some iptable/ebtable LOG statements to see how the traffic
goes when I ping the ISP gateway from host2.  What I saw was (slightly
editted for easy reading):

ebtable: broute BROUTING   :IN=peth1 SRC=10.0.0.22 IP DST=64.81.XX.XX
ebtable: nat  PREROUTING   :IN=peth1 SRC=10.0.0.22 IP DST=64.81.XX.XX
iptable: mangle PREROUTING :IN=xenbr0 PHYSIN=peth1 SRC=10.0.0.22 DST=64.81.XX.XX 
iptable: nat PREROUTING    :IN=xenbr0 PHYSIN=peth1 SRC=10.0.0.22 DST=64.81.XX.XX 
ebtable: filter FORWARD    :IN=peth1 OUT=vif0.0 SRC=10.0.0.22 IP DST=64.81.XX.XX
iptable: mangle FORWARD    :IN=xenbr0 OUT=xenbr0 PHYSIN=peth1 PHYSOUT=vif0.0 SRC=10.0.0.22 DST=64.81.XX.XX 
iptable: filter FORWARD    :IN=xenbr0 OUT=xenbr0 PHYSIN=peth1 PHYSOUT=vif0.0 SRC=10.0.0.22 DST=64.81.XX.XX 
ebtable: nat POSTROUTING   :OUT=vif0.0 SRC=10.0.0.22 IP DST=64.81.XX.XX
iptable: mangle POSTROUTING:OUT=xenbr0 PHYSIN=peth1 PHYSOUT=vif0.0 SRC=10.0.0.22 DST=64.81.XX.XX
iptable: nat POSTROUTING   :OUT=xenbr0 PHYSIN=peth1 PHYSOUT=vif0.0 SRC=10.0.0.22 DST=64.81.XX.XX
iptable: mangle PREROUTING :IN=eth1 SRC=10.0.0.22 DST=64.81.XX.XX
iptable: mangle FORWARD    :IN=eth1 OUT=eth0 SRC=10.0.0.22 DST=64.81.XX.XX
iptable: filter FORWARD    :IN=eth1 OUT=eth0 SRC=10.0.0.22 DST=64.81.XX.XX
iptable: mangle POSTROUTING:OUT=eth0 SRC=10.0.0.22 DST=64.81.XX.XX


I all looked OK to me until the last four lines.  The packet/frame
comes in on peth1, enters the bridge, goes out vif0.0, re-enters on
eth1 and gets forwarded to eth0.

What troubles me is that I expected to see "nat PREROUTING" and "nat
POSTROUTING" lines after the packet re-entered eth1.  The missing "nat
POSTROUTING" line is where the MASQUERADE should have happened.

It seems like the iptables "nat" table was only consulted once, but I
needed it to be hit twice.

Any ideas?

Cheers,
Curt


                 reply	other threads:[~2008-03-03  2:23 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080303022357.GA21681@curtisb.com \
    --to=curt@acm.org \
    --cc=netfilter@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox