Copyright © 2009 Thomas M. Eastep
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, with no Front-Cover, and with no Back-Cover Texts. A copy of the license is included in the section entitled “GNU Free Documentation License”.
2010/06/27
Table of Contents
Open Virtuoso (OpenVZ) is an open source kernel-based virtualization solution from Parallels™ (formerly SWSoft™). Virtual servers take the form of containers (the OpenVZ documentation calls these Virtual Environments or VEs) which are created via templates. Templates are available for a wide variety of distributions and architectures.
OpenVZ requires a patched kernel. Beginning with Lenny, Debian™ supplies OpenVZ kernels through the standard stable repository.
As with any Shorewall installation involving other software, we suggest that you first install OpenVZ and get it working before attempting to add Shorewall. Alternatively, execute shorewall clear while installing and configuring OpenVZ.
The default OpenVZ networking configuration uses Proxy ARP. You
assign containers IP addresses in the IP network from one of your
interfaces and you are expected to set the proxy_arp flag on that
interface
(/proc/sys/net/ipv4/conf/
).interface
/proxy_arp
OpenVZ creates a point-to-point virtual interface in the host with a rather odd configuration.
Example (Single VE with IP address 206.124.146.178):
gateway:~# ip addr ls dev venet0 10: venet0: <BROADCAST,POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN link/void gateway:~# ip route ls dev venet0 206.124.146.178 scope link gateway:~#
The interface has no IP configuration yet it has a route to 206.124.146.178!
From within the VE with IP address 206.124.146.178, we have the following:
server:~ # ip addr ls 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 brd 127.255.255.255 scope host lo inet 127.0.0.2/8 brd 127.255.255.255 scope host secondary lo inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: venet0: <BROADCAST,POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN link/void inet 127.0.0.1/32 scope host venet0 inet 206.124.146.178/32 scope global venet0:0 server:~ # ip route ls 192.0.2.0/24 dev venet0 scope link 127.0.0.0/8 dev lo scope link default via 192.0.2.1 dev venet0 server:~ #
There are a couple of unique features of this configuration:
127.0.0.1/32 is configured on venet0 although the main routing
table routes loopback traffic through the lo
interface as normal.
There is a route to 192.0.2.0/24 through venet0 even though the interface has no IP address in that network. Note: 192.0.2.0/24 is reserved for use in documentation and for testing.
The default route is via 192.0.2.1 yet there is no interface on the host with that IP address.
All of this doesn't really affect the Shorewall configuration but it is interesting none the less.
We recommend handling the strange OpenVZ configuration in Shorewall as follows:
/etc/shorewall/zones
:
############################################################################### #ZONE TYPE OPTIONS IN OUT # OPTIONS OPTIONS net ipv4 vz ipv4
/etc/shorewall/interfaces
:
###############################################################################
#ZONE INTERFACE BROADCAST OPTIONS
net eth0 - proxyarp=1
vz venet0 - routeback,rp_filter=0
If you run Shorewall Multi-ISP support on the host, you should arrange for traffic to your containers to use the main routing table. In the configuration shown here, this entry in /etc/shorewall/route_rules is appropriate:
#SOURCE DEST PROVIDER PRIORITY - 206.124.146.178 main 1000
You can assign an RFC 1918 address to a VE and use masquerade/SNAT
to provide Internet access to the container. This is just a normal
simple Shorewall configuration as shown in the Two-interface Quick Start Guide. In this
configuration the firewall's internal interface is venet0
. Be sure to include the options
shown above.
If you have obtained an OpenVZ VE from a hosting service provider, you may find it difficult to configure any type of firewall within your VE. There are two VE parameters that control iptables behavior within the container:
name
Restrict access to iptables modules inside a container (The OpenVZ claims that by default all iptables modules that are loaded in the host system are accessible inside a container; I haven't tried that).
You can use the following values for
name
: iptable_filter
,
iptable_mangle
, ipt_limit
,
ipt_multiport
, ipt_tos
,
ipt_TOS
, ipt_REJECT
,
ipt_TCPMSS
, ipt_tcpmss
,
ipt_ttl
, ipt_LOG
,
ipt_length
, ip_conntrack
,
ip_conntrack_ftp
,
ip_conntrack_irc
, ipt_conntrack
,
ipt_state
, ipt_helper
,
iptable_nat
, ip_nat_ftp
,
ip_nat_irc
, ipt_REDIRECT
,
xt_mac
, ipt_owner
.
If your provider is using this option, you may be in deep trouble trying to use Shorewall in your container. Look at the output of shorewall show capabilities and weep. Then try to get your provider to remove this restriction on your container.
num
This parameter limits the number of iptables rules that are allowed within the container. The default is 100 which is too small for a Shorewall configuration. We recommend setting this to at least 200.
if you see annoying error messages as shown below during start/restart, remove the module-init-tools package from the VE.
server:/etc/shorewall # shorewall restart
Compiling...
Compiling /etc/shorewall/zones...
Compiling /etc/shorewall/interfaces...
Determining Hosts in Zones...
Preprocessing Action Files...
Pre-processing /usr/share/shorewall/action.Drop...
Pre-processing /usr/share/shorewall/action.Reject...
Compiling /etc/shorewall/policy...
Adding Anti-smurf Rules
Adding rules for DHCP
Compiling TCP Flags filtering...
Compiling Kernel Route Filtering...
Compiling Martian Logging...
Compiling MAC Filtration -- Phase 1...
Compiling /etc/shorewall/rules...
Generating Transitive Closure of Used-action List...
Processing /usr/share/shorewall/action.Reject for chain Reject...
Processing /usr/share/shorewall/action.Drop for chain Drop...
Compiling MAC Filtration -- Phase 2...
Applying Policies...
Generating Rule Matrix...
Creating iptables-restore input...
Compiling iptables-restore input for chain mangle:...
Compiling /etc/shorewall/routestopped...
Shorewall configuration compiled to /var/lib/shorewall/.restart
Restarting Shorewall....
Initializing...
Processing /etc/shorewall/init ...
Processing /etc/shorewall/tcclear ...
Setting up Route Filtering...
Setting up Martian Logging...
Setting up Proxy ARP...
Setting up Traffic Control...
Preparing iptables-restore input...
Running /usr/sbin/iptables-restore...
FATAL: Could not load /lib/modules/2.6.26-2-openvz-amd64/modules.dep: No such file or directory
FATAL: Could not load /lib/modules/2.6.26-2-openvz-amd64/modules.dep: No such file or directory
FATAL: Could not load /lib/modules/2.6.26-2-openvz-amd64/modules.dep: No such file or directory
FATAL: Could not load /lib/modules/2.6.26-2-openvz-amd64/modules.dep: No such file or directory
IPv4 Forwarding Enabled
Processing /etc/shorewall/start ...
Processing /etc/shorewall/started ...
done.
This section presents a working example. This is the configuration at shorewall.net during the summer of 2009.
The network diagram is shown below.
The two systems shown in the green box are OpenVZ Virtual Environments (containers).
In the files below, items in bold font are relevant to the networking/Shorewall configuration.
/etc/vz/conf
(long lines folded for
clarity).
## Global parameters VIRTUOZZO=yes LOCKDIR=/var/lib/vz/lock DUMPDIR=/var/lib/vz/dump VE0CPUUNITS=1000 ## Logging parameters LOGGING=yes LOGFILE=/var/log/vzctl.log LOG_LEVEL=0 VERBOSE=0 ## Disk quota parameters DISK_QUOTA=no VZFASTBOOT=no # The name of the device whose ip address will be used as source ip for VE. # By default automatically assigned. VE_ROUTE_SRC_DEV="eth3" # Controls which interfaces to send ARP requests and modify APR tables on. NEIGHBOUR_DEVS=detect ## Template parameters TEMPLATE=/var/lib/vz/template ## Defaults for VEs VE_ROOT=/home/vz/root/$VEID VE_PRIVATE=/home/vz/private/$VEID CONFIGFILE="vps.basic" #DEF_OSTEMPLATE="fedora-core-4" DEF_OSTEMPLATE="debian" ## Load vzwdog module VZWDOG="no" ## IPv4 iptables kernel modules IPTABLES="iptable_filter iptable_mangle ipt_limit ipt_multiport ipt_tos ipt_TOS ipt_REJECT ipt_TCPMSS ipt_tcpmss ipt_ttl ipt_LOG ipt_length ip_conntrack ip_conntrack_ftp ip_conntrack_irc ipt_conntrack ipt_state ipt_helper iptable_nat ip_nat_ftp ip_nat_irc ipt_REDIRECT xt_mac ipt_owner" ## Enable IPv6 IPV6="no"
/etc/vz/conf/101.conf
:
ONBOOT="yes" # UBC parameters (in form of barrier:limit) KMEMSIZE="574890800:589781600" LOCKEDPAGES="256:256" PRIVVMPAGES="1073741824:2137483648" SHMPAGES="21504:21504" NUMPROC="240:240" PHYSPAGES="0:9223372036854775807" VMGUARPAGES="262144:9223372036854775807" OOMGUARPAGES="26112:9223372036854775807" NUMTCPSOCK="360:360" NUMFLOCK="188:206" NUMPTY="16:16" NUMSIGINFO="256:256" TCPSNDBUF="1720320:2703360" TCPRCVBUF="1720320:2703360" OTHERSOCKBUF="1126080:2097152" DGRAMRCVBUF="262144:262144" NUMOTHERSOCK="360:360" DCACHESIZE="3409920:3624960" NUMFILE="9312:9312" AVNUMPROC="180:180" NUMIPTENT="200:200" # Disk quota parameters (in form of softlimit:hardlimit) DISKSPACE="1048576:1153024" DISKINODES="200000:220000" QUOTATIME="0" # CPU fair sheduler parameter CPUUNITS="1000" VE_ROOT="/home/vz/root/$VEID" VE_PRIVATE="/home/vz/private/$VEID" OSTEMPLATE="suse-11.1-x86_64" ORIGIN_SAMPLE="vps.basic" HOSTNAME="lists.shorewall.net" IP_ADDRESS="206.124.146.177" NAMESERVER="127.0.0.1" NAME="lists" SEARCHDOMAIN="shorewall.net"
This VE is the main server at shorewall.net. Note that some of the memory parameters are set ridiculously large -- I got tired of out-of-memory issues.
/etc/vz/conf/102.conf
(nearly default
configuration on Debian):
ONBOOT="yes" # UBC parameters (in form of barrier:limit) KMEMSIZE="14372700:14790164" LOCKEDPAGES="256:256" PRIVVMPAGES="65536:69632" SHMPAGES="21504:21504" NUMPROC="240:240" PHYSPAGES="0:9223372036854775807" VMGUARPAGES="33792:9223372036854775807" OOMGUARPAGES="26112:9223372036854775807" NUMTCPSOCK="360:360" NUMFLOCK="188:206" NUMPTY="16:16" NUMSIGINFO="256:256" TCPSNDBUF="1720320:2703360" TCPRCVBUF="1720320:2703360" OTHERSOCKBUF="1126080:2097152" DGRAMRCVBUF="262144:262144" NUMOTHERSOCK="360:360" DCACHESIZE="3409920:3624960" NUMFILE="9312:9312" AVNUMPROC="180:180" NUMIPTENT="200:200" # Disk quota parameters (in form of softlimit:hardlimit) DISKSPACE="1048576:1153024" DISKINODES="200000:220000" QUOTATIME="0" # CPU fair sheduler parameter CPUUNITS="1000" VE_ROOT="/home/vz/root/$VEID" VE_PRIVATE="/home/vz/private/$VEID" OSTEMPLATE="debian-5.0-amd64-minimal" ORIGIN_SAMPLE="vps.basic" HOSTNAME="server.shorewall.net" IP_ADDRESS="206.124.146.178" NAMESERVER="206.124.146.177" NAME="server"
I really don't use this server for anything currently but I'm planning to eventually splt the services between the two VEs.
Below are exerpts from the configuration files as they pertain to the OpenVZ environment.
/etc/shorewall/zones
:
#ZONE TYPE OPTIONS IN OUT
# OPTIONS OPTIONS
fw firewall
net ipv4 #Internet
loc ipv4 #Local wired Zone
dmz ipv4 #DMZ
...
/etc/shorewall/params
:
NET_IF=eth3
INT_IF=eth1
VPS_IF=venet0
...
/etc/shorewall/interfaces
:
#ZONE INTERFACE BROADCAST OPTIONS net $NET_IF detect dhcp,blacklist,tcpflags,optional,routefilter=0,nosmurfs,logmartions=0,proxyarp=1 loc $INT_IF detect dhcp,logmartians=1,routefilter=1,nets=(172.20.1.0/24),tcpflags dmz $VPS_IF detect logmartians=0,routefilter=0,nets=(206.124.146.177,206.124.146.178),routeback ...
This is a multi-ISP configuration so entries are required
in /etc/shorewall/route_rules
:
#SOURCE DEST PROVIDER PRIORITY
- 172.20.0.0/24 main 1000
- 206.124.146.177 main 1001
- 206.124.146.178 main 1001
I have set up Shorewall on Server (206.124.146.178) just to have an environment to test with. It is a quite vanilla one-interface configuration.
/etc/shorewall/zones:
#ZONE TYPE OPTIONS IN OUT # OPTIONS OPTIONS fw firewall net ipv4
/etc/shorewall/interfaces:
#ZONE INTERFACE BROADCAST OPTIONS
net venet0 detect dhcp,tcpflags,logmartians,nosmurfs
This is the configuration at shorewall.net during the spring of
2010. Rather than using the venet0 configuration shown above, this
configuration uses a bridge in preparation for adding IPv6 support in the
DMZ. The eth0 interface in each of the containers is statically configured
using the distributions' configuration tools
(/etc/network/interfaces
on Debian and Yast on
OpenSuSE).
The network diagram is shown below.
The two systems shown in the green box are OpenVZ Virtual Environments (containers).
The following stanza in /etc/network/interfaces on the host configures the bridge.
auto vzbr0 iface vzbr0 inet static pre-up /usr/sbin/brctl addbr vzbr0 address 206.124.146.176 network 206.124.146.176 broadcast 206.124.146.176 netmask 255.255.255.255 post-down /usr/sbin/brctl delbr br0
In the files below, items in bold font show the changes from the preceeding example.
/etc/vz/conf
(long lines folded for
clarity).
## Global parameters VIRTUOZZO=yes LOCKDIR=/var/lib/vz/lock DUMPDIR=/var/lib/vz/dump VE0CPUUNITS=1000 ## Logging parameters LOGGING=yes LOGFILE=/var/log/vzctl.log LOG_LEVEL=0 VERBOSE=0 ## Disk quota parameters DISK_QUOTA=no VZFASTBOOT=no # The name of the device whose ip address will be used as source ip for VE. # By default automatically assigned. VE_ROUTE_SRC_DEV="eth3" # Controls which interfaces to send ARP requests and modify APR tables on. NEIGHBOUR_DEVS=detect ## Template parameters TEMPLATE=/var/lib/vz/template ## Defaults for VEs VE_ROOT=/home/vz/root/$VEID VE_PRIVATE=/home/vz/private/$VEID CONFIGFILE="vps.basic" #DEF_OSTEMPLATE="fedora-core-4" DEF_OSTEMPLATE="debian" ## Load vzwdog module VZWDOG="no" ## IPv4 iptables kernel modules IPTABLES="iptable_filter iptable_mangle ipt_limit ipt_multiport ipt_tos ipt_TOS ipt_REJECT ipt_TCPMSS ipt_tcpmss ipt_ttl ipt_LOG ipt_length ip_conntrack ip_conntrack_ftp ip_conntrack_irc ipt_conntrack ipt_state ipt_helper iptable_nat ip_nat_ftp ip_nat_irc ipt_REDIRECT xt_mac ipt_owner" ## Enable IPv6 IPV6="no"
/etc/vz/conf/101.conf
:
ONBOOT="yes"
# UBC parameters (in form of barrier:limit)
KMEMSIZE="574890800:589781600"
LOCKEDPAGES="256:256"
PRIVVMPAGES="1073741824:2137483648"
SHMPAGES="21504:21504"
NUMPROC="240:240"
PHYSPAGES="0:9223372036854775807"
VMGUARPAGES="262144:9223372036854775807"
OOMGUARPAGES="26112:9223372036854775807"
NUMTCPSOCK="360:360"
NUMFLOCK="188:206"
NUMPTY="16:16"
NUMSIGINFO="256:256"
TCPSNDBUF="1720320:2703360"
TCPRCVBUF="1720320:2703360"
OTHERSOCKBUF="1126080:2097152"
DGRAMRCVBUF="262144:262144"
NUMOTHERSOCK="360:360"
DCACHESIZE="3409920:3624960"
NUMFILE="9312:9312"
AVNUMPROC="180:180"
NUMIPTENT="200:200"
# Disk quota parameters (in form of softlimit:hardlimit)
DISKSPACE="1048576:1153024"
DISKINODES="200000:220000"
QUOTATIME="0"
# CPU fair sheduler parameter
CPUUNITS="1000"
VE_ROOT="/home/vz/root/$VEID"
VE_PRIVATE="/home/vz/private/$VEID"
OSTEMPLATE="suse-11.1-x86_64"
ORIGIN_SAMPLE="vps.basic"
HOSTNAME="lists.shorewall.net"
NAMESERVER="127.0.0.1"
NAME="lists"
SEARCHDOMAIN="shorewall.net"
NETIF="ifname=eth0,mac=00:18:51:22:24:81,host_ifname=veth101.0,host_mac=00:18:51:B6:1A:F1"
This VE is the mail server at shorewall.net (MX and IMAP). Note that some of the memory parameters are set ridiculously large -- I got tired of out-of-memory issues.
/etc/vz/conf/102.conf
(nearly default
configuration on Debian):
ONBOOT="yes"
# UBC parameters (in form of barrier:limit)
KMEMSIZE="14372700:14790164"
LOCKEDPAGES="256:256"
PRIVVMPAGES="65536:69632"
SHMPAGES="21504:21504"
NUMPROC="240:240"
PHYSPAGES="0:9223372036854775807"
VMGUARPAGES="33792:9223372036854775807"
OOMGUARPAGES="26112:9223372036854775807"
NUMTCPSOCK="360:360"
NUMFLOCK="188:206"
NUMPTY="16:16"
NUMSIGINFO="256:256"
TCPSNDBUF="1720320:2703360"
TCPRCVBUF="1720320:2703360"
OTHERSOCKBUF="1126080:2097152"
DGRAMRCVBUF="262144:262144"
NUMOTHERSOCK="360:360"
DCACHESIZE="3409920:3624960"
NUMFILE="9312:9312"
AVNUMPROC="180:180"
NUMIPTENT="200:200"
# Disk quota parameters (in form of softlimit:hardlimit)
DISKSPACE="1048576:1153024"
DISKINODES="200000:220000"
QUOTATIME="0"
# CPU fair sheduler parameter
CPUUNITS="1000"
VE_ROOT="/home/vz/root/$VEID"
VE_PRIVATE="/home/vz/private/$VEID"
OSTEMPLATE="debian-5.0-amd64-minimal"
ORIGIN_SAMPLE="vps.basic"
HOSTNAME="server.shorewall.net"
NAMESERVER="206.124.146.177"
NAME="server"
NETIF="ifname=eth0,mac=00:18:51:22:24:80,host_ifname=veth102.0,host_mac=00:18:51:B6:1A:F0"
This server runs the rest of the services for shorewall.net (web server, ftp server, rsyncd, etc.).
With a bridged configuration, the VIF for a VE must be added to the bridge when the VE starts. That is accomplished using mount files.
/etc/vz/conf/101.mount:
#!/bin/bash
# This script source VPS configuration files in the same order as vzctl does
# if one of these files does not exist then something is really broken
[ -f /etc/vz/vz.conf ] || exit 1
[ -f $VE_CONFFILE ] || exit 1
# source both files. Note the order, it is important
. /etc/vz/vz.conf
. $VE_CONFFILE
# Add the VIF to the bridge after VE has started
{
BRIDGE=vzbr0
DEV=veth101.0
while sleep 1; do
/sbin/ifconfig $DEV 0 >/dev/null 2>&1
if [ $? -eq 0 ]; then
/usr/sbin/brctl addif $BRIDGE $DEV
break
fi
done
} &
/etc/vz/conf/102.mount:
#!/bin/bash
# This script source VPS configuration files in the same order as vzctl does
# if one of these files does not exist then something is really broken
[ -f /etc/vz/vz.conf ] || exit 1
[ -f $VE_CONFFILE ] || exit 1
# source both files. Note the order, it is important
. /etc/vz/vz.conf
. $VE_CONFFILE
# Add VIF to bridge after VE has started
{
BRIDGE=vzbr0
DEV=veth102.0
while sleep 1; do
/sbin/ifconfig $DEV 0 >/dev/null 2>&1
if [ $? -eq 0 ]; then
/usr/sbin/brctl addif $BRIDGE $DEV
break
fi
done
} &
Below are exerpts from the configuration files as they pertain to the OpenVZ environment. Again, bold font indicates change from the prior configuration.
/etc/shorewall/zones
:
#ZONE TYPE OPTIONS IN OUT # OPTIONS OPTIONS fw firewall net ipv4 #Internet loc ipv4 #Local wired Zone dmz ipv4 #DMZ ...
/etc/shorewall/params
:
NET_IF=eth3
INT_IF=eth1
VPS_IF=vzbr0
...
/etc/shorewall/interfaces
:
#ZONE INTERFACE BROADCAST OPTIONS net $NET_IF detect dhcp,blacklist,tcpflags,optional,routefilter=0,nosmurfs,logmartions=0 loc $INT_IF detect dhcp,logmartians=1,routefilter=1,nets=(172.20.1.0/24),tcpflags dmz $VPS_IF detect logmartians=0,routefilter=0,nets=(206.124.146.177,206.124.146.178),routeback ...
/etc/shorewall/proxyarp:
#ADDRESS INTERFACE EXTERNAL HAVEROUTE PERSISTENT
206.124.146.177 DMZ_IF eth2 no yes
206.124.146.178 DMZ_IF eth2 no yes
This is a multi-ISP configuration so entries are required in
/etc/shorewall/route_rules
:
#SOURCE DEST PROVIDER PRIORITY - 172.20.0.0/24 main 1000 - 206.124.146.177 main 1001 - 206.124.146.178 main 1001
I have set up Shorewall on VE 101 (206.124.146.178) just to have an environment to test with. It is a quite vanilla one-interface configuration.
/etc/shorewall/zones:
#ZONE TYPE OPTIONS IN OUT # OPTIONS OPTIONS fw firewall net ipv4
/etc/shorewall/interfaces:
#ZONE INTERFACE BROADCAST OPTIONS
net eth0 detect dhcp,tcpflags,logmartians,nosmurfs