Shorewall 3.x Reference Tom Eastep 2005-09-30 2001-2005 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. This documentation is intended primarily for reference. Step-by-step instructions for configuring Shorewall in common setups may be found in the QuickStart Guides. This article applies to Shorewall 3.0 and later. If you are running a version of Shorewall earlier than Shorewall 3.0.0 then please see the documentation for that release.
Components Shorewall consists of the following components: params a parameter file installed in /etc/shorewall that can be used to establish the values of shell variables for use in other files. shorewall.conf a parameter file installed in /etc/shorewall that is used to set several firewall parameters. zones a parameter file installed in /etc/shorewall that defines a network partitioning into zones policy a parameter file installed in /etc/shorewall that establishes overall firewall policy. rules a parameter file installed in /etc/shorewall and used to express firewall rules that are exceptions to the high-level policies established in /etc/shorewall/policy. blacklist a parameter file installed in /etc/shorewall and used to list blacklisted IP/subnet/MAC addresses. ecn a parameter file installed in /etc/shorewall and used to selectively disable Explicit Congestion Notification (ECN - RFC 3168). functions a set of shell functions used by both the firewall and shorewall shell programs. Installed in /usr/share/shorewall. modules a parameter file installed in /etc/shorewall and that specifies kernel modules and their parameters. Shorewall will automatically load the modules specified in this file. tos a parameter file installed in /etc/shorewall that is used to specify how the Type of Service (TOS) field in packets is to be set. init.sh and init.debian.sh a shell script installed in /etc/init.d to automatically start Shorewall during boot. The particular script installed depends on which distribution you are running. interfaces a parameter file installed in /etc/shorewall and used to describe the interfaces on the firewall system. hosts a parameter file installed in /etc/shorewall and used to describe individual hosts or subnetworks in zones. maclist a parameter file installed in /etc/shorewall and used to verify the MAC address (and possibly also the IP address(es)) of devices. masq This file also describes IP masquerading under Shorewall and is installed in /etc/shorewall. firewall a shell program that reads the configuration files in /etc/shorewall and configures your firewall. This file is installed in /usr/share/shorewall. nat a parameter file in /etc/shorewall used to define one-to-one NAT. proxyarp a parameter file in /etc/shorewall used to define Proxy Arp. rfc1918 a parameter file in /usr/share/shorewall used to define the treatment of packets under the norfc1918 interface option. routestopped a parameter file in /etc/shorewall used to define those hosts that can access the firewall when Shorewall is stopped. tcrules a parameter file in /etc/shorewall used to define rules for classifying packets for Traffic Shaping/Control. tcdevices a parameter file in /etc/shorewall used to define the bandwidth for interfaces on which you want traffic shaping to be enabled. tcclasses a parameter file in /etc/shorewall used to define classes for traffic shaping. tcstart a set of shell functions used by Shorewall to setup traffic shaping.This file is installed in /usr/share/shorewall. tunnels a parameter file in /etc/shorewall used to define IPSec tunnels. shorewall a shell program (requiring a Bourne shell or derivative) used to control and monitor the firewall. This should be placed in /sbin or in /usr/sbin (the install.sh script and the rpm install this file in /sbin). accounting a parameter file in /etc/shorewall used to define traffic accounting rules. version a file created in /usr/share/shorewall that describes the version of Shorewall installed on your system. actions and action.template files in /etc/shorewall and /usr/share/shorewall respectively that allow you to define your own actions for rules in /etc/shorewall/rules. actions.std and action.* files in /usr/share/shorewall that define the actions included as a standard part of Shorewall. providers file in /etc/shorewall that is used to define multiple Internet Service Providers and load-balancing. routes file in /etc/shorewall that is used to interface to the experimental ROUTE target from Netfilter patch-o-matic-ng.
/etc/shorewall/params You may use the file /etc/shorewall/params file to set shell variables that you can then use in some of the other configuration files. It is suggested that variable names begin with an upper case letter to distinguish them from variables used internally within the Shorewall programs shell variables NET_IF=eth0 NET_BCAST=130.252.100.255 NET_OPTIONS=blacklist,norfc1918 /etc/shorewall/interfaces record net $NET_IF $NET_BCAST $NET_OPTIONS The result will be the same as if the record had been written net eth0 130.252.100.255 blacklist,norfc1918 Variables may be used anywhere in the other configuration files.
/etc/shorewall/zones This file is used to define the network zones. There is one entry in /etc/shorewall/zones for each zone; Columns in an entry are: ZONE short name for the zone. The name should be 5 characters or less in length and consist of lower-case letters or numbers. Short names must begin with a letter and the name assigned to the firewall is reserved for use by Shorewall itself. Note that the output produced by iptables is much easier to read if you select short names that are three characters or less in length. The name all may not be used as a zone name nor may the zone name assigned to the firewall itself via the FW variable in . TYPE ipsec - All traffic to/from this zone is encrypted. ipv4 - By default, traffic to/from some of the hosts in this zone is not encrypted. Any encrypted hosts are designated using the ipsec option in /etc/shorewall/hosts. firewall - Designates the firewall itself. You must have exactly one 'firewall' zone. No options are permitted with a 'firewall' zone. OPTIONS, IN OPTIONS, OUT OPTIONS Optional parameters that identify the security policy and security associations used in communication with hosts in this zone. A comma-separated list of the following: proto[!]=ah|esp|ipcomp mode[!]=transport|tunnel reqid[!]=<number> — A number assigned to a security policy using the unique:<number> as the SPD level. See setkey(8). tunnel-src[!]=<address>[/<mask>] — Tunnel Source; may only be included with mode=tunnel. Since tunnel source and destination are dependent on the direction of the traffic, this option and the following one should only be included in the IN OPTIONS and OUT OPTIONS columns. tunnel-dst[!]=<address>[/<mask>] — Tunnel Destination; may only be included with mode=tunnel. mss=<number> — Sets the MSS field in TCP syn packets forwarded to/from this zone. May be used to compensate for the lack of IPSEC pseudo-devices with their own MTU in the 2.6 Kernel IPSEC implementation. If specified in the IN OPTIONS, TCP SYN packets from the zone will have MSS altered; if specified in the OUT OPTIONS, TCP SYN packets to the zone will have MSS altered. spi[!]=<number> — The security parameter index of the Security Association. Since a different SA is used for incoming and outgoing traffic, this option should only be listed in the IN OPTIONS and OUT OPTIONS columns. strict — Must be specified when SPD rules are used (e.g., esp encapsulated with ah). next — Separates rules when strict is used. The order of entries in the /etc/shorewall/zones file is significant in some cases.
/etc/shorewall/interfaces This file is used to tell the firewall which of your firewall's network interfaces are connected to which zone. There will be one entry in /etc/shorewall/interfaces for each of your interfaces. Columns in an entry are: ZONE A zone defined in the file or -. If you specify -, you must use the file to define the zones accessed via this interface. INTERFACE the name of the interface (examples: eth0, ppp0, ipsec+). Each interface can be listed on only one record in this file. You do not need to include the loopback interface (lo) in this file. BROADCAST the broadcast address(es) for the sub-network(s) attached to the interface. This should be left empty for P-T-P interfaces (ppp*, ippp*); if you need to specify options for such an interface, enter - in this column. If you supply the special value detect in this column, the firewall will automatically determine the broadcast address. In order to use detect: the interface must be up before you start your firewall the interface must only be attached to a single sub-network (i.e., there must have a single broadcast address). OPTIONS a comma-separated list of options. Possible options include: arp_filter This option causes /proc/sys/net/ipv4/conf/<interface>/arp_filter to be set with the result that this interface will only answer ARP who-has requests from hosts that are routed out of that interface. Setting this option facilitates testing of your firewall where multiple firewall interfaces are connected to the same HUB/Switch (all interface connected to the single HUB/Switch should have this option specified). Note that using such a configuration in a production environment is strongly recommended against. arp_ignore[=<number>] respond to arp requests based on the value of <number>. 1 - reply only if the target IP address is local address configured on the incoming interface 2 - reply only if the target IP address is local address configured on the incoming interface and both with the sender's IP address are part from same subnet on this interface 3 - do not reply for local addresses configured with scope host, only resolutions for global and link addresses are replied 4-7 - reserved 8 - do not reply for all local addresses If no <number> is given then the value 1 is assumed DO NOT SPECIFY arp_ignore FOR ANY INTERFACE INVOLVED IN PROXY ARP. newnotsyn This option overrides NEWNOTSYN=No for packets arriving on this interface. In other words, packets coming in on this interface are processed as if NEWNOTSYN=Yes had been specified in . routeback This option causes Shorewall to set up handling for routing packets that arrive on this interface back out the same interface. If this option is specified, the ZONE column may not contain -. tcpflags This option causes Shorewall to make sanity checks on the header flags in TCP packets arriving on this interface. Checks include Null flags, SYN+FIN, SYN+RST and FIN+URG+PSH; these flag combinations are typically used for silent port scans. Packets failing these checks are logged according to the TCP_FLAGS_LOG_LEVEL option in and are disposed of according to the TCP_FLAGS_DISPOSITION option. blacklist This option causes incoming packets on this interface to be checked against the blacklist. dhcp The interface is assigned an IP address via DHCP or is used by a DHCP server running on the firewall. The firewall will be configured to allow DHCP traffic to and from the interface even when the firewall is stopped. You may also wish to use this option if you have a static IP but you are on a LAN segment that has a lot of Laptops that use DHCP and you select the norfc1918 option (see below). norfc1918 Packets arriving on this interface and that have a source or destination address that is reserved in RFC 1918 will be dropped after being optionally logged. Beware that as IPv4 addresses become in increasingly short supply, ISPs are beginning to use RFC 1918 addresses within their own infrastructure. Also, many cable and DSL modems have an RFC 1918 address that can be used through a web browser for management and monitoring functions. If you want to specify norfc1918 on your external interface but need to allow access to certain addresses from the above list, see FAQ 14. routefilter Invoke the Kernel's route filtering (anti-spoofing) facility on this interface. The kernel will reject any packets incoming on this interface that have a source address that would be routed outbound through another interface on the firewall. If you specify this option for an interface then the interface must be up prior to starting the firewall. proxyarp This option causes Shorewall to set /proc/sys/net/ipv4/conf/<interface>/proxy_arp and is used when implementing Proxy ARP Sub-netting as described at http://www.tldp.org/HOWTO/mini/Proxy-ARP-Subnet/. Do not set this option if you are implementing Proxy ARP through entries in . maclist If this option is specified, all connection requests from this interface are subject to MAC Verification. May only be specified for ethernet interfaces. detectnets If this option is specified, the zone named in the ZONE column will contain only the hosts routed through the interface named in the INTERFACE column. Do not set this option on your external (Internet) interface! The interface must be in the UP state when Shorewall is [re]started. nosmurfs If this option is specified, incoming connection requests will be checked to ensure that they do not have a broadcast or multicast address as their source. Any such packets will be dropped after being optionally logged according to the setting of SMURF_LOG_LEVEL in /etc/shorewall/shorewall.conf. logmartians If this option is specified, the kernel's martian logging facility will be enabled on this interface (/proc/sys/net/ipv4/conf/<interface>/log_martians will be set to 1). See also the LOG_MARTIANS option in shorewall.conf. sourceroute If this option is not specified for an interface, then source-routed packets will not be accepted from that interface (sets /proc/sys/net/ipv4/conf/<interface> to 1).Only set this option if you know what are you doing. upnp Incoming requests from this interface may be remapped via UPNP (upnpd). My recommendations concerning options: External Interface -- tcpflags,blacklist,norfc1918,routefilter,nosmurfs,logmartians Wireless Interface -- maclist,routefilter,tcpflags,detectnets,nosmurfs Use dhcp and proxyarp when needed. You have a conventional firewall setup in which eth0 connects to a Cable or DSL modem and eth1 connects to your local network and eth0 gets its IP address via DHCP. You want to check all packets entering from the internet against the <link linkend="Blacklist">black list</link>. Your /etc/shorewall/interfaces file would be as follows: #ZONE INTERFACE BROADCAST OPTIONS net eth0 detect dhcp,norfc1918,blacklist You have a standalone dialup GNU/Linux System. Your /etc/shorewall/interfaces file would be: #ZONE INTERFACE BROADCAST OPTIONS net ppp0 You have local interface eth1 with two IP addresses - 192.168.1.1/24 and 192.168.12.1/24 #ZONE INTERFACE BROADCAST OPTIONS loc eth1 192.168.1.255,192.168.12.255
/etc/shorewall/hosts Configuration For most applications, specifying zones entirely in terms of network interfaces is sufficient. There may be times though where you need to define a zone to be a more general collection of hosts. This is the purpose of the /etc/shorewall/hosts file. The only time that you need entries in /etc/shorewall/hosts is where you have more than one zone connecting through a single interface. IF YOU DON'T HAVE THIS SITUATION THEN DON'T TOUCH THIS FILE!! Columns in this file are: ZONE A zone defined in the file. HOST(S) The name of an interface defined in the /etc/shorewall/interfaces file followed by a colon (":") and a comma-separated list whose elements are either: The IP address of a host A subnetwork in the form <subnet-address>/<mask width> A physical port name; only allowed when the interface names a bridge created by the brctl addbr command. This port must not be defined in /etc/shorewall/interfaces and may optionally followed by a colon (":") and a host or network IP. See the bridging documentation for details. OPTIONS A comma-separated list of option routeback This option causes Shorewall to set up handling for routing packets sent by this host group back back to the same group. maclist If specified, connection requests from the hosts specified in this entry are subject to MAC Verification. This option is only valid for ethernet interfaces. tcpflags This option causes Shorewall to make sanity checks on the header flags in TCP packets arriving from these hosts. Checks include Null flags, SYN+FIN, SYN+RST and FIN+URG+PSH; these flag combinations are typically used for silent port scans. Packets failing these checks are logged according to the TCP_FLAGS_LOG_LEVEL option in and are disposed of according to the TCP_FLAGS_DISPOSITION option. blacklist (only makes sense for bridge ports) This option causes incoming packets on this port to be checked against the blacklist. norfc1918 -- only makes sense for bridge ports) Packets arriving on this port and that have a source address that is reserved in RFC 1918 will be dropped after being optionally logged as specified in the section of RFC1918_LOG_LEVEL in shorewall.conf. nosmurfs (only makes sense for bridge ports) If this option is specified, incoming connection requests will be checked to ensure that they do not have a broadcast or multicast address as their source. Any such packets will be dropped after being optionally logged according to the setting of SMURF_LOG_LEVEL in /etc/shorewall/shorewall.conf. ipsec The hosts are accessed via a kernel 2.6 ipsec SA. Note that if the zone named in the ZONE column is specified as an IPSEC zone in the /etc/shorewall/zones file then you do NOT need to specify 'ipsec' here. If you don't define any hosts for a zone, the hosts in the zone default to i0:0.0.0.0/0 , i1:0.0.0.0/0, ... where i0, i1, ... are the interfaces to the zone. You probably DON'T want to specify any hosts for your internet zone since the hosts that you specify will be the only ones that you will be able to access without adding additional rules. Your local interface is eth1 and you have two groups of local hosts that you want to make into separate zones: 192.168.1.0/25 192.168.1.128/25 Your /etc/shorewall/interfaces file might look like: #ZONE INTERFACE BROADCAST OPTIONS net eth0 detect dhcp,norfc1918 - eth1 192.168.1.127,192.168.1.255 The - in the ZONE column for eth1 tells Shorewall that eth1 interfaces to multiple zones. #ZONE HOST(S) OPTIONS loc1 eth1:192.168.1.0/25 loc2 eth1:192.168.1.128/25 You have local interface eth1 with two IP addresses - 192.168.1.1/24 and 192.168.12.1/24 Your /etc/shorewall/interfaces file might look like: #ZONE INTERFACE BROADCAST OPTIONS net eth0 detect dhcp,norfc1918 - eth1 192.168.1.255,192.168.12.255 Your /etc/shorewall/hosts file might look like: #ZONE HOST(S) OPTIONS loc eth1:192.168.1.0/24,192.168.12.0/24
Nested and Overlapping Zones The /etc/shorewall/interfaces and /etc/shorewall/hosts file allow you to define nested or overlapping zones. Such overlapping/nested zones are allowed and Shorewall processes zones in the order that they appear in the /etc/shorewall/zones file. So if you have nested zones, you want the sub-zone to appear before the super-zone and in the case of overlapping zones, the rules that will apply to hosts that belong to both zones is determined by which zone appears first in /etc/shorewall/zones. Hosts that belong to more than one zone may be managed by the rules of all of those zones. This is done through use of the special CONTINUE policy described below.
/etc/shorewall/policy Configuration This file is used to describe the firewall policy regarding establishment of connections. Connection establishment is described in terms of clients who initiate connections and servers who receive those connection requests. Policies defined in /etc/shorewall/policy describe which zones are allowed to establish connections with other zones. Policies established in /etc/shorewall/policy can be viewed as default policies. If no rule in /etc/shorewall/rules applies to a particular connection request then the policy from /etc/shorewall/policy is applied. Five policies are defined: ACCEPT The connection is allowed. DROP The connection request is ignored. REJECT The connection request is rejected with an RST (TCP) or an ICMP destination-unreachable packet being returned to the client. QUEUE Send the connection request to a user-space process via the iptables QUEUE target (useful when you are using Snort-inline). CONTINUE The connection is neither ACCEPTed, DROPped nor REJECTed. CONTINUE may be used when one or both of the zones named in the entry are sub-zones of or intersect with another zone. For more information, see below. NONE Shorewall should not set up any infrastructure for handling traffic from the SOURCE zone to the DEST zone. When this policy is specified, the LOG LEVEL and BURST:LIMIT columns must be left blank. For each policy specified in /etc/shorewall/policy, you can indicate that you want a message sent to your system log each time that the policy is applied. Entries in /etc/shorewall/policy have four columns as follows: SOURCE The name of a client zone (a zone defined in the file , the name of the firewall zone or all). DEST The name of a destination zone (a zone defined in the file , the name of the firewall zone or all). Shorewall automatically allows all traffic from the firewall to itself so the name of the firewall zone cannot appear in both the SOURCE and DEST columns. POLICY The default policy for connection requests from the SOURCE zone to the DESTINATION zone. LOG LEVEL Optional. If left empty, no log message is generated when the policy is applied. Otherwise, this column should contain an integer or name indicating a syslog level. LIMIT:BURST - optional If left empty, TCP connection requests from the SOURCE zone to the DEST zone will not be rate-limited. Otherwise, this column specifies the maximum rate at which TCP connection requests will be accepted followed by a colon (:) followed by the maximum burst size that will be tolerated. Example: 10/sec:40 specifies that the maximum rate of TCP connection requests allowed will be 10 per second and a burst of 40 connections will be tolerated. Connection requests in excess of these limits will be dropped. See the rules file documentation for an explanation of how rate limiting works. In the SOURCE and DEST columns, you can enter all to indicate all zones. The default /etc/shorewall/policy file is as follows. #SOURCE DEST POLICY LOG LEVEL LIMIT:BURST loc net ACCEPT net all DROP info all all REJECT info This table may be interpreted as follows: All connection requests from the local network to hosts on the internet are accepted. All connection requests originating from the internet are ignored and logged at level KERNEL.INFO. All other connection requests are rejected and logged. The firewall script processes the /etc/shorewall/policy file from top to bottom and uses the first applicable policy that it finds. For example, in the following policy file, the policy for (loc, loc) connections would be ACCEPT as specified in the first entry even though the third entry in the file specifies REJECT. #SOURCE DEST POLICY LOG LEVEL LIMIT:BURST loc all ACCEPT net all DROP info loc loc REJECT info
Intra-Zone Traffic Shorewall allows a zone to be associated with more than one interface or with multiple networks that interface through a single interface. Shorewall will ACCEPT all traffic from a zone to itself provided that there is no explicit policy governing traffic from that zone to itself (an explicit policy does not specify all in either the SOURCE or DEST column). The idea is this: A zone should be homogeneous with respect to security requirements. Traffic within a zone should not require rules or policies. Shorewall will not restrict traffic within a zone. Any time that you have multiple interfaces associated with a single zone, you should ask yourself if you really want traffic routed between those interfaces. Cases where you might not want that behavior are: Multiple net interfaces to different ISPs. You don't want to route traffic from one ISP to the other through your firewall. Multiple VPN clients. You don't necessarily want them to all be able to communicate between themselves using your gateway/router. You can control the traffic from the firewall to itself. As with any zone, fw->fw traffic is enabled by default. It is not necessary to define the loopback interface (lo) in /etc/shorewall/interfaces in order to define fw->fw rules or a fw->fw policy.
The CONTINUE policy Where zones are nested or overlapping, the CONTINUE policy allows hosts that are within multiple zones to be managed under the rules of all of these zones. Let's look at an example: /etc/shorewall/zones: #ZONE TYPE OPTION $FW firewall sam ipv4 net ipv4 loc ipv4 /etc/shorewall/interfaces: #ZONE INTERFACE BROADCAST OPTIONS - eth0 detect dhcp,norfc1918 loc eth1 detect /etc/shorewall/hosts: #ZONE HOST(S) OPTIONS net eth0:0.0.0.0/0 sam eth0:206.191.149.197 Sam's home system is a member of both the sam zone and the net zone and as described above , that means that sam must be listed before net in /etc/shorewall/zones. /etc/shorewall/policy: #SOURCE DEST POLICY LOG LEVEL loc net ACCEPT sam all CONTINUE net all DROP info all all REJECT info The second entry above says that when Sam is the client, connection requests should first be process under rules where the source zone is sam and if there is no match then the connection request should be treated under rules where the source zone is net. It is important that this policy be listed BEFORE the next policy (net to all). Partial /etc/shorewall/rules: #ACTION SOURCE DEST PROTO DEST PORT(S) ... DNAT sam loc:192.168.1.3 tcp ssh DNAT net loc:192.168.1.5 tcp www ... Given these two rules, Sam can connect to the firewall's internet interface with ssh and the connection request will be forwarded to 192.168.1.3. Like all hosts in the net zone, Sam can connect to the firewall's internet interface on TCP port 80 and the connection request will be forwarded to 192.168.1.5. The order of the rules is not significant. Sometimes it is necessary to suppress port forwarding for a sub-zone. For example, suppose that all hosts can SSH to the firewall and be forwarded to 192.168.1.5 EXCEPT Sam. When Sam connects to the firewall's external IP, he should be connected to the firewall itself. Because of the way that Netfilter is constructed, this requires two rules as follows: #ACTION SOURCE DEST PROTO DEST PORT(S) ... DNAT sam $FW tcp ssh DNAT net loc:192.168.1.3 tcp ssh ... The first rule allows Sam SSH access to the firewall. The second rule says that any clients from the net zone with the exception of those in the sam zone should have their connection port forwarded to 192.168.1.3. If you need to exclude more than one zone in this way, you can list the zones separated by commas (e.g., net!sam,joe,fred). This technique also may be used when the ACTION is REDIRECT.
/etc/shorewall/rules The /etc/shorewall/rules file defines exceptions to the policies established in the /etc/shorewall/policy file. There is one entry in /etc/shorewall/rules for each of these rules. Entries in this file only govern the establishment of new connections — packets that are part of an existing connection or that establish a connection that is related to an existing connection are automatically accepted. Rules for each pair of zones (source zone, destination zone) are evaluated in the order that they appear in the file — the first match determines the disposition of the connection request with a couple of caveats: LOG rules cause the connection request to be logged then processing continues with the next rule in the file. QUEUE rules cause the connection request to be passed to user-space -- the user-space application can later insert them back into the stream for further processing by following rules. CONTINUE rules may cause the connection request to be reprocessed using a different (source zone, destination zone) pair. The /etc/shorewall/rules file may now be into sections. Each section is introduced by a line that begins with the keyword SECTION which is followed by the section name. Sections are as listed below and must appear in the order shown. ESTABLISHED Rules in this section apply to packets in the ESTABLISHED state. RELATED Rules in this section apply to packets in the RELATED state. NEW Rules in this section apply to packets in the NEW and INVALID states. Rules in the ESTABLISHED and RELATED sections are limited to the following ACTIONs:
ACCEPT, DROP, REJECT, QUEUE, LOG and User-defined actions.
Macros may be used in these sections provided that they expand to only these ACTIONs. At the end of the ESTABLISHED and RELATED sections, there is an implicit ACCEPT rule. RESTRICTION: If you specify FASTACCEPT=Yes in /etc/shorewall.shorewall.conf then the ESTABLISHED and RELATED sections must be empty. Unless you understand Netfilter well enough to be comfortable with the difference between ESTABLISHED, RELATED, INVALID and NEW connection tracking states, you should omit the ESTABLISHED and RELATED sections and place all of your rules in the NEW section. Entries in the file have the following columns: ACTION ACCEPT, DROP, REJECT, CONTINUE These have the same meaning here as in the policy file above. ACCEPT+ Works like ACCEPT but also exempts the connection from matching DNAT and REDIRECT rules later in the file. NONAT Exempts matching connections from DNAT and REDIRECT rules later in the file. DNAT Causes the connection request to be forwarded to the system specified in the DEST column (port forwarding). DNAT stands for Destination Network Address Translation DNAT- The above ACTION (DNAT) generates two iptables rules: a header-rewriting rule in the Netfilter nat table an ACCEPT rule in the Netfilter filter table. DNAT- works like DNAT but only generates the header-rewriting rule. SAME SAME is useful when more than one server IP address (an address range, for example) is given in the DEST column below. SAME works similar to DNAT with the exception that when multiple connections from an internet host match a SAME rule then all of the connections will be sent to the same internal server. Unlike when using DNAT rules, SAME rules may not alter the destination port number used for the connection. SAME- SAME generates two iptables rules: a header-rewriting rule in the Netfilter nat table an ACCEPT rule in the Netfilter filter table. SAME- works like SAME but only generates the header-rewriting rule. REDIRECT Causes the connection request to be redirected to a port on the local (firewall) system. REDIRECT- The above ACTION (REDIRECT) generates two iptables rules: a header-rewriting rule in the Netfilter nat table an ACCEPT rule in the Netfilter filter table. REDIRECT- works like REDIRECT but only generates the header-rewriting rule. LOG Log the packet -- requires a syslog level (see below). QUEUE Forward the packet to a user-space application. This facility is provided to allow interfacing to ftwall for Kazaa filtering. When the protocol specified in the PROTO column is TCP (tcp, TCP or 6), Shorewall will only pass connection requests (SYN packets) to user space. This is for compatibility with ftwall. <defined action> An action defined in the /etc/shorewall/actions or /usr/share/shorewall/actions.std files. <macro> The name of a macro defined using a file with name macro.<name>. Macro files are usually placed in /etc/shorewall but may reside in any directory on the CONFIG_PATH. The ACTION may optionally be followed by : and a syslog level (example: REJECT:info or ACCEPT:debug). This causes the packet to be logged at the specified level prior to being processed according to the specified ACTION. Note: if the ACTION is LOG then you MUST specify a syslog level. The log level may be optionally followed by a log tag. A log tag is a string of alphanumeric characters and is specified by following the log level with ":" and the log tag. Example:ACCEPT:info:ftp net dmz tcp 21 The log tag is appended to the log prefix generated by the LOGPREFIX variable in /etc/shorewall/conf. If "ACCEPT:info" generates the log prefix "Shorewall:net2dmz:ACCEPT:" then "ACCEPT:info:ftp" will generate "Shorewall:net2dmz:ACCEPT:ftp " (note the trailing blank). The maximum length of a log prefix supported by iptables is 29 characters; if a larger prefix is generated, Shorewall will issue a warning message and will truncate the prefix to 29 characters. Specifying a log level for a <defined action> will log all invocations of the action. For example: AllowFTP:info net dmz will log all net->dmz traffic that has not been handled by earlier rules. That's probably not what you want. If you want to log the FTP connections that are actually accepted, you need to log within the action itself. One way to do that would be to copy /usr/share/shorewall/action.AllowFTP to /etc/shorewall and modify the copy as follows: #TARGET SOURCE DEST PROTO DEST SOURCE RATE USER/ # PORT PORT(S) LIMIT GROUP ACCEPT:info - - tcp 21 #LAST LINE -- ADD YOUR ENTRIES BEFORE THIS ONE -- DO NOT REMOVE The use of DNAT or REDIRECT requires that you have NAT enabled in your kernel configuration. SOURCE Describes the source hosts to which the rule applies.. The contents of this field must begin with the name of a zone defined in /etc/shorewall/zones, $FW, all or "none". If the ACTION is DNAT or REDIRECT, sub-zones may be excluded from the rule by following the initial zone name with ! and a comma-separated list of those sub-zones to be excluded. There is an example above. If the source is "none" then the rule is ignored. This is most commonly used with Shell Variables where a shell variable is set to "none" if a rule is to be omitted. If the source is not all then the source may be further restricted by adding a colon (:) followed by a comma-separated list of qualifiers. Qualifiers are may include: interface name refers to any connection requests arriving on the specified interface (example loc:eth4). The interface name may optionally be followed by a colon (:) and an IP address or subnet (examples: loc:eth4:192.168.4.22, net:eth0:192.0.2.0/24). IP address refers to a connection request from the host with the specified address (example net:155.186.235.151). If the ACTION is DNAT, this must not be a DNS name. MAC Address in Shorewall format. subnet refers to a connection request from any host in the specified subnet (example net:155.186.235.0/24). IP address ranges of the form <first address>-<last address> may be specified. This requires that your kernel and iptables have iprange match support. DEST Describes the destination host(s) to which the rule applies. May take most of the forms described above for SOURCE plus the following two additional forms: An IP address followed by a colon and the port number that the server is listening on (service names from /etc/services are not allowed - example loc:192.168.1.3:80). A single port number (again, service names are not allowed) -- this form is only allowed if the ACTION is REDIRECT and refers to a server running on the firewall itself and listening on the specified port. Restrictions: MAC addresses may not be specified. In DNAT rules, only IP addresses may be given -- DNS names are not permitted. You may not specify both an IP address and an interface name in the DEST column. Like in the SOURCE column, a range of IP addresses may be specified in the DEST column as <first address>-<last address>. When the ACTION is DNAT or DNAT-, connections will be assigned to the addresses in the range in a round-robin fashion (load-balancing). PROTO Protocol. Must be a protocol name from /etc/protocols, a number, or all. Specifies the protocol of the connection request. DEST PORT(S) Port or port range (<low port>:<high port>) being connected to. May only be specified if the protocol is tcp, udp or icmp. For icmp, this column's contents are interpreted as an icmp type. If you don't want to specify DEST PORT(S) but need to include information in one of the columns to the right, enter - in this column. You may give a list of ports and/or port ranges separated by commas. Port numbers may be either integers or service names from /etc/services. SOURCE PORTS(S) May be used to restrict the rule to a particular client port or port range (a port range is specified as <low port number>:<high port number>). If you don't want to restrict client ports but want to specify something in the next column, enter - in this column. If you wish to specify a list of port number or ranges, separate the list elements with commas (with no embedded white space). Port numbers may be either integers or service names from /etc/services. ORIGINAL DEST This column may only be non-empty if the ACTION is DNAT or REDIRECT. If DNAT or REDIRECT is the ACTION and the ORIGINAL DEST column is left empty, any connection request arriving at the firewall from the SOURCE that matches the rule will be forwarded or redirected. This works fine for connection requests arriving from the internet where the firewall has only a single external IP address. When the firewall has multiple external IP addresses or when the SOURCE is other than the internet, there will usually be a desire for the rule to only apply to those connection requests directed to particular IP addresses (see Example 2 below for another usage). Those IP addresses are specified in the ORIGINAL DEST column as a comma-separated list. If this list begins with ! then the rule will only apply if the original destination address matches none of the addresses listed. RATE LIMIT You may rate-limit ACCEPT, DNAT[-], REDIRECT[-] or LOG rules with an entry in this column. Entries have the form <rate>/<interval>[:<burst>] where <rate> is the number of connections per <interval> (sec or min) and <burst> is the largest burst permitted. If no burst value is given, a value of 5 is assumed. There may be no whitespace embedded in the specification. Let's take ACCEPT<2/sec:4> net dmz tcp 80 The first time this rule is reached, the packet will be accepted; in fact, since the burst is 4, the first four packets will be accepted. After this, it will be 500ms (1 second divided by the rate of 2) before a packet will be accepted from this rule, regardless of how many packets reach it. Also, every 500ms which passes without matching a packet, one of the bursts will be regained; if no packets hit the rule for 2 second, the burst will be fully recharged; back where we started. When rate limiting is specified on a rule with all in the SOURCE or DEST fields below, the limit will apply to each pair of zones individually rather than as a single limit for all pairs of zones covered by the rule. If you want to specify any following columns but no rate limit, place - in this column. USER/GROUP Output rules from the firewall itself may be restricted to a particular set of users and/or user groups. See the User Set Documentation for details. You wish to forward all ssh connection requests from the internet to local system 192.168.1.3. You wish to limit the number of connections to 4/minute with a burst of 8: #ACTION SOURCE DEST PROTO DEST PORT(S) DNAT<4/min:8> net loc:192.168.1.3 tcp ssh You want to redirect all local www connection requests EXCEPT those to your own http server (206.124.146.177) to a Squid transparent proxy running on the firewall and listening on port 3128. Squid will of course require access to remote web servers. This example shows yet another use for the ORIGINAL DEST column; here, connection requests that were NOT (notice the <quote>!</quote>) originally destined to 206.124.146.177 are redirected to local port 3128. #ACTION SOURCE DEST PROTO DEST PORT(S) SOURCE ORIGINAL # PORT(S) DEST REDIRECT loc 3128 tcp www - !206.124.146.177 ACCEPT $FW net tcp www You want to run a web server at 155.186.235.222 in your DMZ and have it accessible remotely and locally. the DMZ is managed by Proxy ARP or by classical sub-netting. #ACTION SOURCE DEST PROTO DEST PORT(S) ACCEPT net dmz:155.186.235.222 tcp www ACCEPT loc dmz:155.186.235.222 tcp www You want to run wu-ftpd on 192.168.2.2 in your masqueraded DMZ. Your internet interface address is 155.186.235.151 and you want the FTP server to be accessible from the internet in addition to the local 192.168.1.0/24 and dmz 192.168.2.0/24 subnetworks. since the server is in the 192.168.2.0/24 subnetwork, we can assume that access to the server from that subnet will not involve the firewall (but see FAQ 2) unless you have more than one external IP address, you can leave the ORIGINAL DEST column blank in the first rule. You cannot leave it blank in the second rule though because then all ftp connections originating in the local subnet 192.168.1.0/24 would be sent to 192.168.2.2 regardless of the site that the user was trying to connect to. That is clearly not what you want. #ACTION SOURCE DEST PROTO DEST PORT(S) SOURCE ORIGINAL # PORT(S) DEST DNAT net dmz:192.168.2.2 tcp ftp DNAT loc:192.168.1.0/24 dmz:192.168.2.2 tcp ftp - 155.186.235.151 If you are running wu-ftpd, you should restrict the range of passive in your /etc/ftpaccess file. I only need a few simultaneous FTP sessions so I use port range 65500-65535. In /etc/ftpaccess, this entry is appropriate: passive ports 0.0.0.0/0 65500 65534 If you are running pure-ftpd, you would include -p 65500:65534 on the pure-ftpd runline. The important point here is to ensure that the port range used for FTP passive connections is unique and will not overlap with any usage on the firewall system. You wish to allow unlimited DMZ access to the host with MAC address 02:00:08:E3:FA:55. #ACTION SOURCE DEST PROTO DEST PORT(S) ACCEPT loc:~02-00-08-E3-FA-55 dmz all You wish to allow access to the SMTP server in your DMZ from all zones. #ACTION SOURCE DEST PROTO DEST PORT(S) ACCEPT all dmz tcp 25 When all is used as a source or destination, intra-zone traffic is not affected. In this example, if there were two DMZ interfaces then the above rule would NOT enable SMTP traffic between hosts on these interfaces. Your firewall's external interface has several IP addresses but you only want to accept SSH connections on address 206.124.146.176. #ACTION SOURCE DEST PROTO DEST PORT(S) ACCEPT net fw:206.124.146.176 tcp 22 (For advanced users). From the internet, you with to forward tcp port 25 directed to 192.0.2.178 and 192.0.2.179 to host 192.0.2.177 in your DMZ. You also want to allow access from the internet directly to tcp port 25 on 192.0.2.177. #ACTION SOURCE DEST PROTO DEST PORT(S) SOURCE ORIGINAL # PORT(S) DEST DNAT- net dmz:192.0.2.177 tcp 25 - 192.0.2.178 DNAT- net dmz:192.0.2.177 tcp 25 - 192.0.2.179 ACCEPT net dmz:192.0.2.177 tcp 25 Using DNAT- rather than DNAT avoids two extra copies of the third rule from being generated. You have 9 http servers behind a Shorewall firewall and you want connection requests to be distributed among your servers. The servers are 192.168.1.101-192.168.1.109. #ACTION SOURCE DEST PROTO DEST PORT(S) DNAT net loc:192.168.1.101-192.168.1.109 tcp 80 You want to redirect all local www connection requests EXCEPT those from 192.168.1.4 and 192.168.1.199 to a Squid transparent proxy running on the firewall and listening on port 3128. #ACTION SOURCE DEST PROTO DEST PORT(S) SOURCE ORIGINAL # PORT(S) DEST NONAT loc:192.168.1.4,192.168.1.199 \ net tcp www REDIRECT loc 3128 tcp www - ACCEPT $FW net tcp www The reason that NONAT is used in the above example rather than ACCEPT+ is that the example is assuming the usual ACCEPT loc->net policy. Since traffic from the local zone to the internet zone is accepted anyway, adding an additional ACCEPT rule is unnecessary and all that is required is to avoid the REDIRECT rule for HTTP connection requests from the two listed IP addresses. Look here for information on other services.
/etc/shorewall/masq The /etc/shorewall/masq file is used to define classical IP Masquerading and Source Network Address Translation (SNAT). There is one entry in the file for each subnet that you want to masquerade. In order to make use of this feature, you must have NAT enabled. Columns are: INTERFACE The interface that will masquerade the subnet; this is normally your internet interface. This interface name can be optionally qualified by adding : and a subnet or host IP. When this qualification is added, only packets addressed to that host or subnet will be masqueraded. The interface name can be qualified with ":" followed by a comma separated list of hosts and/or subnets. If this list begins with ! (e.g., eth0:!192.0.2.8/29,192.0.2.32/29) then only packets addressed to destinations not listed will be masqueraded; otherwise (e.g., eth0:192.0.2.8/29,192.0.2.32/29), traffic will be masqueraded if it does match one of the listed addresses. If you have set ADD_SNAT_ALIASES=Yes in , you can cause Shorewall to create an alias label of the form interfacename:digit (e.g., eth0:0) by placing that label in this column. See example 5 below. Alias labels created in this way allow the alias to be visible to the ipconfig utility. THAT IS THE ONLY THING THAT THIS LABEL IS GOOD FOR AND IT MAY NOT APPEAR ANYWHERE ELSE IN YOUR SHOREWALL CONFIGURATION. Normally MASQUERADE/SNAT rules are evaluated after one-to-one NAT rules defined in the /etc/shorewall/nat file. If you preceed the interface name with a plus sign ("+") then the rule will be evaluated before one-to-one NAT. Examples: +eth0 +eth1:192.0.2.32/27 The effect of ADD_SNAT_ALIASES=Yes can be negated for an entry by following the interface name by ":" but no digit. Examples: eth0: eth1::192.0.2.32/27 +eth3 SUBNET The subnet that you want to have masqueraded through the INTERFACE. This may be expressed as a single IP address, a subnet or an interface name. In the latter instance, the interface must be configured and started before Shorewall is started as Shorewall will determine the subnet based on information obtained from the ip utility. The subnet may be optionally followed by ! and a comma-separated list of addresses and/or subnets that are to be excluded from masquerading. ADDRESS The source address to be used for outgoing packets. This column is optional and if left blank, the current primary IP address of the interface in the first column is used. If you have a static IP on that interface, listing it here makes processing of output packets a little less expensive for the firewall. If you specify an address in this column, it must be an IP address configured on the INTERFACE or you must have ADD_SNAT_ALIASES enabled in . You may include a range of IP addresses in this column to indicate that Netfilter should use the addresses in the range in round-robin fashion. Beginning with Shorewall version 1.4.7, you may include a list of ranges and/or addresses in this column; again, Netfilter will use all listed ranges/addresses in rounded-robin fashion. You may also specify the source port range to be used (the PROTO column must specify tcp or udp) by following the address or address range if any with ":" and the port range (in the format <low port>-<high port>). Examples: #INTERFACE SUBNET ADDRESS PROTO eth0 10.0.0.0/8 192.0.2.44:7000-8000 udp #INTERFACE SUBNET ADDRESS PROTO eth0 192.168.1.0/24 :4000-5000 tcp Some internet application that establish multiple connections from a client assume that when SNAT is being used that all connections between the client and a particular client and a remote server will appear to the server to come from the same external IP address. You can ensure that this is the case by preceding the ADDRESS range by "SAME:". Example: #INTERFACE SUBNET ADDRESS eth0 10.0.0.0/8 SAME:192.0.2.44-192.168.2.50 If you want all connections from an internal system to use the same external IP address regardless of the remote server that they are connecting to then precede the ADDRESS range by "SAME:nodst:". Example: #INTERFACE SUBNET ADDRESS eth0 10.0.0.0/8 SAME:nodst:192.0.2.44-192.168.2.50 PROTO If specified, must be a protocol number of a protocol name from /etc/protocols. Restricts the SNAT or Masquerade to that protocol. PORT(S) If the PROTO column specifies TCP (6) or UDP (17) then this column may be used to restrict to SNAT or Masquerade to traffic with a certain destination port or a set of destination ports. The column may contain: A port number or a port name from /etc/services. A comma-separated list of port numbers and/or port names. Your kernel must have Multiport match support. You can tell if your kernel has this support by issuing a shorewall check command and looking at the output under Shorewall has detected the following iptables/netfilter capabilities:. A range of port numbers of the form <low port>:<high port> IPSEC If you specify a value other than "-" in this column, you must be running kernel 2.6 and your kernel and iptables must include policy match support. The value in this column is a comma-separated list of options from the following. Only packets that will be encrypted via an SA that matches these options will have their source address changed. Yes or yes ― Match any SA. Normally used as the only option. reqid=<number> where <number> is specified using setkey(8) using the 'unique:<number>' option for the SPD level. spi=<number> where <number> is the SPI of the SA. proto=ah|esp|ipcomp mode=transport|tunnel tunnel-src=<address>[/<mask>] (only available with mode=tunnel) tunnel-dst=<address>[/<mask>] (only available with mode=tunnel) strict — Means that packets must match all rules. next — Separates rules; can only be used with strict. You have eth0 connected to a cable modem and eth1 connected to your local subnetwork 192.168.9.0/24. Your /etc/shorewall/masq file would look like: #INTERFACE SUBNET ADDRESS eth0 192.168.9.0/24 You have a number of IPSEC tunnels through ipsec0 and you want to masquerade traffic from your 192.168.9.0/24 subnet to the remote subnet 10.1.0.0/16 only. #INTERFACE SUBNET ADDRESS ipsec0:10.1.0.0/16 192.168.9.0/24 You have a DSL line connected on eth0 and a local network (192.168.10.0/24) connected to eth1. You want all local->net connections to use source address 206.124.146.176. #INTERFACE SUBNET ADDRESS eth0 192.168.10.0/24 206.124.146.176 Same as example 3 except that you wish to exclude 192.168.10.44 and 192.168.10.45 from the SNAT rule. #INTERFACE SUBNET ADDRESS eth0 192.168.10.0/24!192.168.10.44,192.168.10.45 206.124.146.176 You have a second IP address (206.124.146.177) assigned to you and wish to use it for SNAT of the subnet 192.168.12.0/24. You want to give that address the name eth0:0. You must have ADD_SNAT_ALIASES=Yes in <xref linkend="Conf" />. #INTERFACE SUBNET ADDRESS eth0:0 192.168.12.0/24 206.124.146.177 You want to use both 206.124.146.177 and 206.124.146.179 for SNAT of the subnet 192.168.12.0/24. Each address will be used on alternate outbound connections. #INTERFACE SUBNET ADDRESS eth0 192.168.12.0/24 206.124.146.177,206.124.146.179 You want all outgoing SMTP traffic entering the firewall on eth1 to be sent from eth0 with source IP address 206.124.146.177. You want all other outgoing traffic from eth1 to be sent from eth0 with source IP address 206.124.146.176. #INTERFACE SUBNET ADDRESS PROTO PORT(S) eth0 eth1 206.124.146.177 tcp 25 eth0 eth1 206.124.146.176 Note that the order of the entries in the above example is important.
/etc/shorewall/proxyarp If you want to use proxy ARP on an entire sub-network, I suggest that you look at the Proxy ARP Subnet Mini HOWTO. If you decide to use the technique described in that HOWTO, you can set the proxy_arp flag for an interface (/proc/sys/net/ipv4/conf/<interface>/proxy_arp) by including the proxyarp option in the interface's record in . When using Proxy ARP sub-netting, you do NOT include any entries in /etc/shorewall/proxyarp. The /etc/shorewall/proxyarp file is used to define Proxy ARP. The file is typically used for enabling Proxy ARP on a small set of systems since you need one entry in this file for each system using proxy ARP. Columns are: ADDRESS address of the system. INTERFACE the interface that connects to the system. If the interface is obvious from the subnetting, you may enter - in this column. EXTERNAL the external interface that you want to honor ARP requests for the ADDRESS specified in the first column. HAVEROUTE If you already have a route through INTERFACE to ADDRESS, this column should contain Yes or yes. If you want Shorewall to add the route, the column should contain No or no. PERSISTENT If you specify "No" or "no" in the HAVEROUTE column, Shorewall will automatically add a route to the host in the ADDRESS column through the interface in the INTERFACE column. If you enter No or no in the PERSISTENT column or if you leave the column empty, that route will be deleted if you issue a shorewall stop or shorewall clear command. If you place Yes or yes in the PERSISTENT column, then those commands will not cause the route to be deleted. After you have made a change to the /etc/shorewall/proxyarp file, you may need to flush the ARP cache of all routers on the LAN segment connected to the interface specified in the EXTERNAL column of the change/added entry(s). If you are having problems communicating between an individual host (A) on that segment and a system whose entry has changed, you may need to flush the ARP cache on host A as well. ISPs typically have ARP configured with long TTL (hours!) so if your ISPs router has a stale cache entry (as seen using tcpdump -nei <external interface> host <IP addr>), it may take a long while to time out. I personally have had to contact my ISP and ask them to delete a stale entry in order to restore a system to working order after changing my proxy ARP settings. You have public IP addresses 155.182.235.0/28. You configure your firewall as follows: eth0 - 155.186.235.1 (internet connection) eth1 - 192.168.9.0/24 (masqueraded local systems) eth2 - 192.168.10.1 (interface to your DMZ) In your DMZ, you want to install a Web/FTP server with public address 155.186.235.4. On the Web server, you subnet just like the firewall's eth0 and you configure 155.186.235.1 as the default gateway. In your /etc/shorewall/proxyarp file, you will have: #ADDRESS INTERFACE EXTERNAL HAVEROUTE 155.186.235.4 eth2 eth0 NO You may want to configure the servers in your DMZ with a subnet that is smaller than the subnet of your internet interface. See the Proxy ARP Subnet Mini HOWTO for details. In this case you will want to place Yes in the HAVEROUTE column.
/etc/shorewall/nat The /etc/shorewall/nat file is used to define one-to-one NAT. There is one entry in the file for each one-to-one NAT relationship that you wish to define. In order to make use of this feature, you must have NAT enabled. If all you want to do is forward ports to servers behind your firewall, you do NOT want to use one-to-one NAT. Port forwarding can be accomplished with simple entries in the rules file. Also, in most cases Proxy ARP provides a superior solution to one-to-one NAT because the internal systems are accessed using the same IP address internally and externally. Columns in an entry are: EXTERNAL External IP address This should NOT be the primary IP address of the interface named in the next column. INTERFACE Interface that you want the EXTERNAL IP address to appear on. If you have set ADD_IP_ALIASES=Yes in , you can specify an alias label of the form interfacename:digit (e.g., eth0:0) and Shorewall will create the alias with that label. Alias labels created in this way allow the alias to be visible to the ipconfig utility. THAT IS THE ONLY THING THAT THIS LABEL IS GOOD FOR AND IT MAY NOT APPEAR ANYWHERE ELSE IN YOUR SHOREWALL CONFIGURATION. The effect of ADD_IP_ALIASES=Yes can be negated for an entry by following the interface name by ":" but no digit. Example: eth0: INTERNAL Internal IP address. ALL INTERFACES If Yes or yes, NAT will be effective from all hosts. If No or no (or if left empty) then NAT will be effective only through the interface named in the INTERFACE column. LOCAL If Yes or yes, NAT will be effective from the firewall system. Note that with Shorewall 2.0.1 and earlier versions, this column was ignored if the ALL INTERFACES column did not contain "Yes" or "yes". This column's contents are independent of the value in ALL INTERFACES. Look here for additional information and an example.
/etc/shorewall/tunnels The /etc/shorewall/tunnels file allows you to define IPSec, GRE, IPIP, OpenVPN, PPTP, 6to4 and other tunnels with end-points on your firewall. For an overview of Shorewall's VPN support, try this article. Instructions for setting up IPSEC tunnels may be found here (if you are using kernel 2.6 with native IPSEC support, look here), instructions for IPIP and GRE tunnels are here, instructions for OpenVPN tunnels are here, instructions for PPTP tunnels are here, instructions for 6to4 tunnels are here, and instructions for integrating Shorewall with other types of tunnels are here.
/etc/shorewall/shorewall.conf This file is used to set the following firewall parameters: FASTACCEPT Normally, Shorewall accepting ESTABLISHED/RELATED packets until these packets reach the chain in which the original connection was accepted. So for packets going from the 'loc' zone to the 'net' zone, ESTABLISHED/RELATED packets are ACCEPTED in the 'loc2net' chain. If you set FASTACCEPT=Yes, then ESTABLISHED/RELEATED packets are accepted early in the INPUT, FORWARD and OUTPUT chains. If you set FASTACCEPT=Yes then you may not include rules in the ESTABLISHED or RELATED sections of /etc/shorewall/rules. STARTUP_ENABLED When set to Yes or yes, Shorewall may be started. Used as guard against Shorewall being accidentally started before it has been configured. MACLIST_TTL The performance of configurations with a large numbers of entries in /etc/shorewall/maclist can be improved by setting the MACLIST_TTL variable in /etc/shorewall/shorewall.conf. If your iptables and kernel support the "Recent Match" (see the output of "shorewall check" near the top), you can cache the results of a 'maclist' file lookup and thus reduce the overhead associated with MAC Verification. When a new connection arrives from a 'maclist' interface, the packet passes through then list of entries for that interface in /etc/shorewall/maclist. If there is a match then the source IP address is added to the 'Recent' set for that interface. Subsequent connection attempts from that IP address occurring within $MACLIST_TTL seconds will be accepted without having to scan all of the entries. After $MACLIST_TTL from the first accepted connection request from an IP address, the next connection request from that IP address will be checked against the entire list. If MACLIST_TTL is not specified or is specified as empty (e.g, MACLIST_TTL="" or is specified as zero then 'maclist' lookups will not be cached). RFC1918_STRICT Traditionally, the RETURN target in the 'rfc1918' file has caused norfc1918 processing to cease for a packet if the packet's source IP address matches the rule. Thus, if you have this entry in /etc/shorewall/rfc1918: #SUBNETS TARGET 192.168.1.0/24 RETURN then traffic from 192.168.1.4 to 10.0.3.9 will be accepted even though you also have: #SUBNETS TARGET 10.0.0.0/8 logdrop Setting RFC1918_STRICT=Yes in shorewall.conf will cause such traffic to be logged and dropped since while the packet's source matches the RETURN rule, the packet's destination matches the 'logdrop' rule. If not specified or specified as empty (e.g., RFC1918_STRICT="") then RFC1918_STRICT=No is assumed. RFC1918_STRICT=Yes requires that your kernel and iptables support 'Connection Tracking' match. LOGALLNEW When set to a log level, this option causes Shorewall to generate a logging rule as the first rule in each builtin chain. The table name is used as the chain name in the log prefix. The chain name is used as the target in the log pref Example: Using the default LOGFORMAT, the log prefix for logging from the nat table's PREROUTING chain is: Shorewall:nat:PREROUTING There is no rate limiting on these logging rules so use LOGALLNEW at your own risk; it may cause high CPU and disk utilization and you may not be able to control your firewall after you enable this option. DO NOT USE THIS OPTION IF THE RESULTING LOG MESSAGES WILL BE SENT TO ANOTHER SYSTEM. DYNAMIC_ZONES When set to Yes or yes, enables dynamic zones. CONFIG_PATH Specifies where configuration files other than shorewall.conf may be found. CONFIG_PATH is specifies as a list of directory names separated by colons (":"). When looking for a configuration file other than shorewall.conf: If the command is "try" or if "-c <configuration directory>" was specified in the command then the directory given in the command is searched first. Next, each directory in the CONFIG_PATH setting is searched in sequence. If CONFIG_PATH is not given or if it is set to the empty value then the contents of /usr/share/shorewall/configpath are used. As released from shorewall.net, that file sets the CONFIG_PATH to /etc/shorewall:/usr/share/shorewall but your particular distribution may set it differently. Note that the setting in /usr/share/shorewall/configpath is always used to locate shorewall.conf. PKTTYPE Normally Shorewall attempts to use the iptables packet type match extension to determine broadcast and multicast packets. This can cause a message to appear during shorewall start (modprobe: cant locate module ipt_pkttype). Some users have found problems with the packet match extension with the result that their firewall log is flooded with messages relating to broadcast packets. If you are experiencing either of these problems, setting PKTTYPE=No will prevent Shorewall from trying to use the packet type match extension and to use IP address matching to determine which packets are broadcasts or multicasts. RESTOREFILE The simple name of a file in /var/lib/shorewall to be used as the default restore script in the shorewall save, shorewall restore, shorewall forget and shorewall -f start commands. See the Saved Configuration documentation for details. BRIDGING When set to Yes or yes, enables Shorewall Bridging support. SMURF_LOG_LEVEL Specifies the logging level for smurf packets (see the nosmurfs option in /etc/shorewall/interfaces). If set to the empty value ( SMURF_LOG_LEVEL="" ) then smurfs are not logged. MODULE_SUFFIX The value of this variable determines the possible file extensions of kernel modules. The default value is "o gz ko and o.gz". See for more details. ADMINISABSENTMINDED The value of this variable affects Shorewall's stopped state. When ADMINISABSENTMINDES=No, only traffic to/from those addresses listed in /etc/shorewall/routestopped is accepted when Shorewall is stopped. When ADMINISABSENTMINDED=Yes, in addition to traffic to/from addresses in /etc/shorewall/routestopped, connections that were active when Shorewall stopped continue to work and all new connections from the firewall system itself are allowed. If this variable is not set or is given the empty value then ADMINISABSENTMINDED=No is assumed. SHOREWALL_SHELL This parameter is used to specify the shell program to be used to interpret the firewall script (/usr/share/shorewall/firewall). If not specified or specified as a null value, /bin/sh is assumed. IPTABLES This parameter names the iptables executable to be used by Shorewall. If not specified or if specified as a null value, then the iptables executable located using the PATH option is used. LOGFORMAT The value of this variable generate the --log-prefix setting for Shorewall logging rules. It contains a printf formatting template which accepts three arguments (the chain name, logging rule number (optional) and the disposition). To use LOGFORMAT with fireparse, set it as: LOGFORMAT="fp=%s:%d a=%s " If the LOGFORMAT value contains the substring %d then the logging rule number is calculated and formatted in that position; if that substring is not included then the rule number is not included. If not supplied or supplied as empty (LOGFORMAT="") then Shorewall:%s:%s: is assumed. /sbin/shorewall uses the leading part of the LOGFORMAT string (up to but not including the first %) to find log messages in the show log, status and hits commands. This part should not be omitted (the LOGFORMAT should not begin with %) and the leading part should be sufficiently unique for /sbin/shorewall to identify Shorewall messages. CLEAR_TC If this option is set to No then Shorewall won't clear the current traffic control rules during [re]start. This setting is intended for use by people that prefer to configure traffic shaping when the network interfaces come up rather than when the firewall is started. If that is what you want to do, set TC_ENABLED=Yes and CLEAR_TC=No and do not supply an /etc/shorewall/tcstart file. That way, your traffic shaping rules can still use the fwmark classifier based on packet marking defined in /etc/shorewall/tcrules. If not specified, CLEAR_TC=Yes is assumed. MARK_IN_FORWARD_CHAIN If your kernel has a FORWARD chain in the mangle table, you may set MARK_IN_FORWARD_CHAIN=Yes to cause the marking specified in the tcrules file to occur in that chain rather than in the PREROUTING chain. This permits you to mark inbound traffic based on its destination address when SNAT or Masquerading are in use. To determine if your kernel has a FORWARD chain in the mangle table, use the /sbin/shorewall show mangle command; if a FORWARD chain is displayed then your kernel will support this option. If this option is not specified or if it is given the empty value (e.g., MARK_IN_FORWARD_CHAIN="") then MARK_IN_FORWARD_CHAIN=No is assumed. RFC1918_LOG_LEVEL This parameter determines the level at which packets logged under the norfc1918 mechanism are logged. The value must be a valid syslog level and if no level is given, then info is assumed. TCP_FLAGS_DISPOSITION Determines the disposition of TCP packets that fail the checks enabled by the tcpflags interface option and must have a value of ACCEPT (accept the packet), REJECT (send an RST response) or DROP (ignore the packet). If not set or if set to the empty value (e.g., TCP_FLAGS_DISPOSITION="") then TCP_FLAGS_DISPOSITION=DROP is assumed. TCP_FLAGS_LOG_LEVEL Determines the syslog level for logging packets that fail the checks enabled by the tcpflags interface option. The value must be a valid syslogd log level. If you don't want to log these packets, set to the empty value (e.g., TCP_FLAGS_LOG_LEVEL=""). MACLIST_DISPOSITION Determines the disposition of connections requests that fail MAC Verification and must have the value ACCEPT (accept the connection request anyway), REJECT (reject the connection request) or DROP (ignore the connection request). If not set or if set to the empty value (e.g., MACLIST_DISPOSITION="") then MACLIST_DISPOSITION=REJECT is assumed. MACLIST_LOG_LEVEL Determines the syslog level for logging connection requests that fail MAC Verification. The value must be a valid syslogd log level. If you don't want to log these connection requests, set to the empty value (e.g., MACLIST_LOG_LEVEL=""). NEWNOTSYN When set to Yes or yes, Shorewall will filter TCP packets that are not part of an established connection and that are not SYN packets (SYN flag on - ACK flag off). If set to No, Shorewall will silently drop such packets. If not set or set to the empty value (e.g., NEWNOTSYN=), NEWNOTSYN=No is assumed. If you have a HA setup with failover to another firewall, you should have NEWNOTSYN=Yes on both firewalls. You should also select NEWNOTSYN=Yes if you have asymmetric routing. I find that NEWNOTSYN=No tends to result in lots of "stuck" connections because any network timeout during TCP session tear down results in retries being dropped (Netfilter has removed the connection from the conntrack table but the end-points haven't completed shutting down the connection). I therefore have chosen NEWNOTSYN=Yes as the default value and I advise caution in using NEWNOTSYN=Yes. If you are looking for a way to defeat "stealth TCP scans" then I recommend the tcpflags interface option in /etc/shorewall/interfaces rather than NEWNOTSYN=No. LOGNEWNOTSYN Shorewall can be configured to drop non-SYN TCP packets that are not part of an existing connection (see NEWNOTSYN above). If you would like to log these packets, set LOGNEWNOTSYN to the syslog level at which you want the packets logged. Example: LOGNEWNOTSYN=ULOG| Packets logged under this option are usually the result of a "stuck" connection rather than as the result of an attempt to breach your firewall. LOG_MARTIANS If set to Yes or yes, sets /proc/sys/net/ipv4/conf/all/log_martians and /proc/sys/net/ipv4/conf/default/log_martians to 1. Default is which sets both of the above to zero. If you do not enable martian logging for all interfaces, you may still enable it for individual interfaces using the logmartians interface option in /etc/shorewall/interfaces. DETECT_DNAT_ADDRS If set to Yes or yes, Shorewall will detect the first IP address of the interface to the source zone and will include this address in DNAT rules as the original destination IP address. If set to No or no, Shorewall will not detect this address and any destination IP address will match the DNAT rule. If not specified or empty, DETECT_DNAT_ADDRS=Yes is assumed. NAT_BEFORE_RULES If set to No or no, port forwarding rules can override the contents of the file. If set to Yes or yes, port forwarding rules cannot override one-to-one NAT. If not set or set to an empty value, Yes is assumed. SUBSYSLOCK This parameter should be set to the name of a file that the firewall should create if it starts successfully and remove when it stops. Creating and removing this file allows Shorewall to work with your distribution's initscripts. For RedHat, this should be set to /var/lock/subsys/shorewall. For Debian, the value is /var/state/shorewall and in LEAF it is /var/run/shorwall. Example: SUBSYSLOCK=/var/lock/subsys/shorewall. MODULESDIR This parameter specifies the directory where your kernel netfilter modules may be found. If you leave the variable empty, Shorewall will supply the value "/lib/modules/`uname -r`/kernel/net/ipv4/netfilter. LOGRATE and LOGBURST These parameters set the match rate and initial burst size for logged packets. Please see the iptables man page for a description of the behavior of these parameters (the iptables option --limit is set by LOGRATE and --limit-burst is set by LOGBURST). If both parameters are set empty, no rate-limiting will occur. LOGRATE=10/minute LOGBURST=5 For each logging rule, the first time the rule is reached, the packet will be logged; in fact, since the burst is 5, the first five packets will be logged. After this, it will be 6 seconds (1 minute divided by the rate of 10) before a message will be logged from the rule, regardless of how many packets reach it. Also, every 6 seconds which passes without matching a packet, one of the bursts will be regained; if no packets hit the rule for 30 seconds, the burst will be fully recharged; back where we started. LOGFILE This parameter tells the /sbin/shorewall program where to look for Shorewall messages when processing the show log, monitor, status and hits commands. If not assigned or if assigned an empty value, /var/log/messages is assumed. IP_FORWARDING This parameter determines whether Shorewall enables or disables IPV4 Packet Forwarding (/proc/sys/net/ipv4/ip_forward). Possible values are: On or on packet forwarding will be enabled. Off or off packet forwarding will be disabled. Keep or keep Shorewall will neither enable nor disable packet forwarding. If this variable is not set or is given an empty value (IP_FORWARD="") then IP_FORWARD=On is assumed. ADD_IP_ALIASES This parameter determines whether Shorewall automatically adds the external address(es) in . If the variable is set to Yes or yes then Shorewall automatically adds these aliases. If it is set to No or no, you must add these aliases yourself using your distribution's network configuration tools. If this variable is not set or is given an empty value (ADD_IP_ALIASES="") then ADD_IP_ALIASES=Yes is assumed. Addresses added by ADD_IP_ALIASES=Yes are deleted and re-added during shorewall restart. As a consequence, connections using those addresses may be severed. ADD_SNAT_ALIASES This parameter determines whether Shorewall automatically adds the SNAT ADDRESS in . If the variable is set to Yes or yes then Shorewall automatically adds these addresses. If it is set to No or no, you must add these addresses yourself using your distribution's network configuration tools. If this variable is not set or is given an empty value (ADD_SNAT_ALIASES="") then ADD_SNAT_ALIASES=No is assumed. Addresses added by ADD_SNAT_ALIASES=Yes are deleted and re-added during shorewall restart. As a consequence, connections using those addresses may be severed. RETAIN_ALIASES During "shorewall start", IP addresses to be added as a consequence of ADD_IP_ALIASES=Yes and ADD_SNAT_ALIASES=Yes are quietly deleted when /etc/shorewall/nat and /etc/shorewall/masq are processed then are re-added later. This is done to help ensure that the addresses can be added with the specified labels but can have the undesirable side effect of causing routes to be quietly deleted. When RETAIN_ALIASES is set to Yes, existing addresses will not be deleted. Regardless of the setting of RETAIN_ALIASES, addresses added during "shorewall start" are still deleted at a subsequent "shorewall stop" or "shorewall restart". LOGUNCLEAN This parameter determines the logging level of mangled/invalid packets controlled by the dropunclean" and "logunclean interface options. If LOGUNCLEAN is empty (LOGUNCLEAN=) then packets selected by dropclean are dropped silently (logunclean packets are logged under the info log level). Otherwise, these packets are logged at the specified level (Example: LOGUNCLEAN=debug). BLACKLIST_DISPOSITION This parameter determines the disposition of packets from blacklisted hosts. It may have the value DROP if the packets are to be dropped or REJECT if the packets are to be replied with an ICMP port unreachable reply or a TCP RST (tcp only). If you do not assign a value or if you assign an empty value then DROP is assumed. BLACKLIST_LOGLEVEL This parameter determines if packets from blacklisted hosts are logged and it determines the syslog level that they are to be logged at. Its value is a syslog level (Example: BLACKLIST_LOGLEVEL=debug). If you do not assign a value or if you assign an empty value then packets from blacklisted hosts are not logged. DELAYBLACKLISTLOAD Users with a large static black list (/etc/shorewall/blacklist) may want to set the DELAYBLACKLISTLOAD option to Yes. When DELAYBLACKLISTLOAD=Yes, Shorewall will enable new connections before loading the blacklist rules. While this may allow connections from blacklisted hosts to slip by during construction of the blacklist, it can substantially reduce the time that all new connections are disabled during shorewall [re]start. CLAMPMSS[=<value>] This parameter enables the TCP Clamp MSS to PMTU feature of Netfilter and is usually required when your internet connection is through PPPoE or PPTP. If set to Yes or yes, the feature is enabled. If left blank or set to No or no, the feature is not enabled. This option requires CONFIG_IP_NF_TARGET_TCPMSS in your kernel. You may also set CLAMPMSS to a numeric value (e.g., CLAMPMSS=1400). This will set the MSS field in TCP SYN packets going through the firewall to the value that you specify. ROUTE_FILTER If this parameter is given the value Yes or yes then route filtering (anti-spoofing) is enabled on all network interfaces which are brought up while Shorewall is in the started state. The default value is no.
/etc/shorewall/modules Configuration The file /etc/shorewall/modules contains commands for loading the kernel modules required by Shorewall-defined firewall rules. Shorewall will source this file during start/restart provided that it exists and that the directory specified by the MODULESDIR parameter exists (see above). The file that is released with Shorewall calls the Shorewall function loadmodule for the set of modules that I load. The loadmodule function is called as follows: loadmodule <modulename> [ <module parameters> ] where <modulename> is the name of the modules without the trailing .o (example ip_conntrack). <module parameters> Optional parameters to the insmod utility. The function determines if the module named by <modulename> is already loaded and if not then the function determines if the .o file corresponding to the module exists in the <moduledirectory>; if so, then the following command is executed: insmod <moduledirectory>/<modulename>.o <module parameters> If the file doesn't exist, the function determines of the .o.gz file corresponding to the module exists in the moduledirectory. If it does, the function assumes that the running configuration supports compressed modules and execute the following command: insmod <moduledirectory>/<modulename>.o.gz <module parameters> The value of the MODULE_SUFFIX option in determines which files the loadmodule function looks for if the named module doesn't exist. For each file <extension> listed in MODULE_SUFFIX (default "o gz ko o.gz"), the function will append a period (".") and the extension and if the resulting file exists then the following command will be executed: insmod moduledirectory/<modulename>.<extension> <module parameters>
/etc/shorewall/tos Configuration The /etc/shorewall/tos file allows you to set the Type of Service field in packet headers based on packet source, packet destination, protocol, source port and destination port. In order for this file to be processed by Shorewall, you must have mangle support enabled. Entries in the file have the following columns: SOURCE The source zone. May be qualified by following the zone name with a colon (:) and either an IP address, an IP subnet, a MAC address in Shorewall Format or the name of an interface. This column may also contain the name of the firewall zone to indicate packets originating on the firewall itself or all to indicate any source. DEST The destination zone. May be qualified by following the zone name with a colon (:) and either an IP address or an IP subnet. Because packets are marked prior to routing, you may not specify the name of an interface. This column may also contain all to indicate any destination. PROTOCOL The name of a protocol in /etc/protocols or the protocol's number. SOURCE PORT(S) The source port or a port range. For all ports, place a hyphen (-) in this column. DEST PORT(S) The destination port or a port range. To indicate all ports, place a hyphen (-) in this column. TOS The type of service. Must be one of the following: Minimize-Delay (16) Maximize-Throughput (8) Maximize-Reliability (4) Minimize-Cost (2) Normal-Service (0) Here's a sample /etc/shorewall/tos file: #SOURCE DEST PROTOCOL SOURCE PORTS(S) DEST PORTS(S) TOS all all tcp - ssh 16 all all tcp ssh - 16 all all tcp - ftp 16 all all tcp ftp - 16 all all tcp - ftp-data 8 all all tcp ftp-data - 8 Users have reported that odd routing problems result from adding the ESP and AH protocols to the /etc/shorewall/tos file.
/etc/shorewall/blacklist Each line in /etc/shorewall/blacklist contains an IP address, a MAC address in Shorewall Format or subnet address. 130.252.100.69 206.124.146.0/24 Packets from hosts listed in the blacklist file will be disposed of according to the value assigned to the BLACKLIST_DISPOSITION and BLACKLIST_LOGLEVEL variables in /etc/shorewall/shorewall.conf. Only packets arriving on interfaces that have the blacklist option in /etc/shorewall/interfaces are checked against the blacklist. The black list is designed to prevent listed hosts/subnets from accessing services on your network. The blacklist file has three columns: ADDRESS/SUBNET As described above. PROTOCOL Optional. If specified, only packets specifying this protocol will be blocked. PORTS Optional; may only be given if PROTOCOL is tcp, udp or icmp. Expressed as a comma-separated list of port numbers or service names (from /etc/services). If present, only packets destined for the specified protocol and one of the listed ports are blocked. When the PROTOCOL is icmp, the PORTS column contains a comma-separated list of ICMP type numbers or names (see iptables -h icmp). Shorewall also has a dynamic blacklist capability. The Shorewall blacklist file is NOT designed to police your users' web browsing -- to do that, I suggest that you install and configure Squid with SquidGuard.
/usr/share/shorewall/rfc1918 This file lists the subnets affected by the norfc1918 interface option. Columns in the file are: SUBNET The subnet using VLSM notation (e.g., 192.168.0.0/16). TARGET What to do with packets to/from the SUBNET: RETURN Process the packet normally thru the rules and policies. See also RFC1918_STRICT above. DROP Silently drop the packet. logdrop Log then drop the packet -- see the RFC1918_LOG_LEVEL parameter above. If you want to modify this file, DO NOT MODIFY /usr/share/shorewall/rfc1918. Rather copy that file to /etc/shorewall/rfc1918 and modify the copy.
/etc/shorewall/netmap Network mapping is defined using the /etc/shorewall/netmap file. Columns in this file are: TYPE Must be DNAT or SNAT. If DNAT, traffic entering INTERFACE and addressed to NET1 has it's destination address rewritten to the corresponding address in NET2. If SNAT, traffic leaving INTERFACE with a source address in NET1 has it's source address rewritten to the corresponding address in NET2. NET1 Must be expressed in CIDR format (e.g., 192.168.1.0/24). INTERFACE A firewall interface. This interface must have been defined in /etc/shorewall/interfaces. NET2 A second network expressed in CIDR format. For more information, see the Network Mapping documentation.
/etc/shorewall/routestopped This file defines the hosts that are accessible from the firewall when the firewall is stopped. Entries in this file are also active while Shorewall is being stopped or [re]started. Columns in the file are: INTERFACE The firewall interface through which the host(s) communicate with the firewall. HOST(S) - (Optional) A comma-separated list of IP/Subnet addresses. If not supplied or supplied as - then 0.0.0.0/0 is assumed. When your firewall is stopped, you want firewall accessibility from local hosts 192.168.1.0/24 and from your DMZ. Your DMZ interfaces through eth1 and your local hosts through eth2. #INTERFACE HOST(S) eth2 192.168.1.0/24 eth1 - Shorewall allows connections defined by the contents of /etc/shorewall/routestopped during the potentially lengthy processing of the commands shorewall start, shorewall restart, shorewall try and shorewall stop.
/etc/shorewall/maclist This file is described in the MAC Validation Documentation.
/etc/shorewall/ecn This file is described in the ECN Control Documentation.
/etc/shorewall/accounting This file is described in the Traffic Accounting Documentation.
Revision History 1.25 2005-08-28 TE Changes for 3.0. 1.24 2005-03-15 TE Update for use of /etc/shorewall/routestopped during [re]start. 1.23 2005-03-10 TE Changes for Shorewall 2.2.2. 1.20 2004-10-25 TE Changes for Shorewall 2.2 Beta 1. 1.19 2004-09-12 TE Changes for Shorewall 2.1. 1.18 2004-08-22 TE Add /etc/shorewall/ipsec documentation. 1.17 2004-04-05 TE Update for Shorewall 2.0.2 1.16 2004-03-17 TE Clarified LOGBURST and LOGLIMIT. 1.15 2004-02-16 TE Move the rfc1918 file to /usr/share/shorewall. 1.14 2004-02-13 TE Add a note about the order of rules. 1.13 2004-02-03 TE Update for Shorewall 2.0. 1.12 2004-01-21 TE Add masquerade destination list. 1.12 2004-01-18 TE Correct typo. 1.11 2004-01-05 TE Standards Compliance 1.10 2004-01-05 TE Improved formatting of DNAT- and REDIRECT- for clarity 1.9 2003-12-25 MN Initial Docbook Conversion Complete