I just got my server up and running after having some problems upgrading the kernel. Now my iptables scripts will not route traffic. I have been using this script for a while without any problems before. What would cause the problem? The script is: #!/bin/bash # ----------------------------------------------------------------------| # This is it...MonMotha's Firewall 2.3.7! | # All your hackers are belong to Linux/Netfilter! | # ----------------------------------------------------------------------| # 2.3 RELEASE NOTES: This is the 2.2 series with some extra stuff, | # including MAC address matching, stateful matching, port forwarding, | # per-proto accept behavior, and some other stuff that I might think | # about adding later. | # ----------------------------------------------------------------------| # COMMENTS from MonMotha: | # | # I will be entering "feature freeze" when 2.3.8 goes final. Please | # make sure to have any patches or feature requests in by then. | # I expect 2.3.7 to be closing in on deserving the "stable" marking. | # --MonMotha | # | # Please note the change of my e-mail address. The new address is: | # obi-wan@starwarsfan.com. The old address (bvmopen@usa.net) will be | # discontinued as of July 31, 2001. | # --MonMotha | # | # When e-mailing to report a bug, please check first that it has not | # already been fixed in the next prerelease (which can be found at the | # distribution site). | # --MonMotha | # | # Before e-mailing me, please check the distribution site (which can be | # found at http://freshmeat.net/projects/mothafirewall as it changes | # sometimes) for a new version. | # --MonMotha | # | # Please...PLEASE give me feedback on your experiences with this script | # I would really like to know what everyone wants, what works, and | # about the inevitable bugs present in anything. | # | # Direct all feedback to: obi-wan@starwarsfan.com | # --MonMotha | # | # When e-mailing with problems, please include firewall script version, | # iptables version, kernel version, and GNU BASH version. If you think | # your problem might be related to kernel configuration, please attach | # the .config file for your kernel. | # --MonMotha | # | # ----------------------------------------------------------------------| # SYSTEM REQUIREMENTS: You must have either compiled the appropriate | # iptables support into your 2.4 kernel or have loaded all the | # applicable modules BEFORE you run this script. This script will not | # load modules for you. | # | # You will need (at least) the following kernel options to use | # this firewall: CONFIG_NETFILTER, CONFIG_IP_NF_IPTABLES, | # CONFIG_IP_NF_FILTER, CONFIG_IP_NF_MATCH_STATE and | # CONFIG_IP_NF_TARGET_REJECT. | # To use the masquerading you will also need (at least): | # CONFIG_IP_NF_CONNTRACK, CONFIG_IP_NF_NAT, CONFIG_IP_NF_NAT_NEEDED | # and CONFIG_IP_NF_TARGET_MASQUERADE. | # Additional options may be needed to use other features. | # | # You need iptables. Get it at "http://netfilter.filewatcher.org". | # Some of the features will need patches only present in the CVS | # | # This script was written (and partially tested) with iptables CVS | # and kernel 2.4.x (non testing) in mind. | # | # Also, this is a BASH shell script...any 2.x version of GNU BASH | # should work. | # ----------------------------------------------------------------------| # | # ALL USERS, READ THE FOLLOWING: | # | # This is distributed under the modified BSD liscense: | # | # Redistribution and use in source and binary forms, with or without | # modification, are permitted provided that the following conditions | # are met: | # | # 1.Redistributions of source code must retain the above copyright | # notice, this list of conditions and the following disclaimer. | # 2.Redistributions in binary form must reproduce the above | # copyright notice, this list of conditions and the following | # disclaimer in the documentation and/or other materials provided | # with the distribution. | # 3.The name of the author may not be used to endorse or promote | # products derived from this software without specific prior | # written permission. | # | # THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR | # IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED | # WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | # ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY | # DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE | # GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS | # INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER | # IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR | # OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN | # IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE | # | # While this may be used freely for commercial use, I do REQUEST that | # any commercial users please tell me via e-mail at | # obi-wan@starwarsfan.com that they are using it, why they chose it, | # how well it works, etc. | # | # ----------------------------------------------------------------------| # IMPORTANT: This firewall is currently in beta! It may be too | # restrictive or insecure. | # ----------------------------------------------------------------------| # CHANGELOG: (Since 2.3.0-pre1a only) | # version 2.3.7: No changes from pre8 | # version 2.3.7-pre8: Change email address on liscense | # Revert to pre6 behavior of dropping ICMP | # echo-request (take global DROP= policy) | # Allow everything from interface lo | # Correct pre7 changelog | # Special rules for DHCP servers | # version 2.3.7-pre7: Fix version number in changelog entry below | # Fix 127.0.0.1 INPUT handling. | # Only enable IP forwarding if it's needed | # (INTERNAL_LAN defined) | # Tweak flood parameters | # Hostwise allows now override explicit, | # denies but not blackholes | # ICMP echo-request will no longer take the | # specified drop policy when it doesn't | # comply with limits, straight DROP will | # be used instead | # Fix REJECT handling in TREJECT and LTREJECT | # Add transparent proxy support (Joshua Link) | # version 2.3.7-pre6: Fix status reporting on SSR SysCtl loop | # Fix the SSR SysCtl loop | # Remove stateful match from forward chain | # version 2.3.7-pre5: Make the default policy actually be DROP | # instead of just saying it is | # Add stateful matching to forward chain to | # prevent people from routing onto your | # internal network (please tell me if | # breaks anything). Thanks to Martin | # Mosny for noticing this | # Block Source Routed Packets to help with | # the above problem | # Add option for TCP SynCookies on or off | # Fix BLACKHOLE directive (was being applied | # to INPUT/OUTPUT after the jump for | # INETIN/INETOUT so didn't apply for | # the internet). Thanks to Gerry Doris | # for noticing this | # Add DHCP client to default UDP port allows | # Note email address change | # Changed emphesis in comments | # Forwarding of port ranges (Vinny and Eddie) | # version 2.3.7-pre4: Line 414, missing subnet match caused all | # packets from anywhere to be allowed. | # Fixed. | # version 2.3.7-pre3: Fix missing fi (fatal syntax error) | # Fix logging in TCPACCEPT chain | # version 2.3.7-pre2: Add route verification (thanks to Jeremy | # Frank) | # Add blackhole directive | # Updated configuration sanity checks | # Ripped out SSH Stuff as it isn't needed | # True default DROP on INPUT | # Don't run the INTERNAL_LAN loop if no nets | # Upped the default SYN limit as large | # numbers of small FTP transfers would | # overload it quickly | # Form cleanups | # version 2.3.7-pre1: Maybe the FTP will work now (fixes for the | # RELATED state) | # Now works with both LAN and DMZ iface null | # Moved static NAT to stable options | # Change parser to /bin/bash not /bin/sh | # version 2.3.6: Add TTL mangling | # Added some more EFNet servers to the list | # Fix in the DMZOUT chain | # Fix FTP stuff | # version 2.3.5: Fixes to make port forwarding work again | # version 2.3.4: USE_MASQ has been changed to MASQ_LAN in port fw | # Fix syntax error in TCP port forwards | # General cleanup | # Fixes in port forwarding | # It's LTREJECT, not TLREJECT | # More TOS mangling | # version 2.3.3: Fatal syntax error in IP forward detect fix | # Don't bail on no IP forward for no LAN | # version 2.3.3-pre1: Reject with tcp-reset for TCP option | # Removed the huge list of censorship | # Moved the port forwards to stable options | # Moved the TOS mangling to stable options | # Check before enabling IP Forwarding and | # IP SynCookies | # Don't run censorship loop if no rules | # Request low latency TOS on UDP packets for | # games on ports 4000-7000 (Diablo II) | # Fix bad syntax in the port forwarding loops | # Reversed DMZIN and DMZOUT fixed | # Various syntax fixes | # Stateful inspection on forward chain | # Other stateful matching changes | # version 2.3.2: Fixed bad syntax in DMZ_IFACE loop | # version 2.3.2-pre2: Put a real liscense on it (BSD liscense) | # Changed format of ALLOW_HOSTWISE and | # DENY_HOSTWISE to be less confusing | # (the ":" was changed to ">") | # Added LOG_FLOOD option to tweak log limit | # Added SYN_FLOOD option to tweak SYN limit | # Added PING_FLOOD option to tweak PING limit | # version 2.3.2-pre1: Stateful matching on active FTP and SSH | # rules (thanks to Len Padilla) | # Fixed a minor bug in chain creation order | # (thanks to Peter Lindman) | # TOS Optimizations (thanks to vesa alatalo) | # Begin DMZ Support | # Proofread comments and correct | # Use BASH builtins instead of sed | # (thanks to Craig Ludington) | # Fixed "USE_SNAT" bug in port forwarding | # (has been changed to "SNAT_LAN") | # (thanks to Frédéric Marchand) | # Tuned down default TCP allows (remove POP3) | # version 2.3.1: Option for 1:1 or subnet:1 static NAT | # Internet censorship options | # version 2.3.1-pre2: Added option to deny specific ports from | # specific hosts | # Added limiting to logging chains to prevent | # log DoSing | # Spiffed up comments | # Changed the "AUTH_ALLOW" and "DNS" options | # to be more generic and flexible | # version 2.3.1-pre1: Updated comments for new kernel version | # Removed double drop setting | # Updated for iptables-1.2 | # Began a kernel option list | # version 2.3.0: No changes from pre1g | # version 2.3.0-pre1g: Tuned down default TCP allows | # Restructure to SSH loop | # Status Reporting Fixes (newlines, etc.) | # Fix log prefix length on accept loops | # version 2.3.0-pre1f: Moved the ICMP echo-request limit to where | # it should have been | # Allows the rest of the ICMP like it should | # Remove the interface matching from ICMP | # echo-request (not needed) | # version 2.3.0-pre1e: Fixed an issue in the invalid matching | # version 2.3.0-pre1d: Spiffed up comments | # Port Forwarding | # Moved the deny setting to normal options | # version 2.3.0-pre1c: Minor fixes that don't (currently) affect | # functionality | # version 2.3.0-pre1b: Security fix documented in 2.1.13 | # Slight logic change in TCP_ALLOW loop | # Don't print allow messages if nothign is | # allowed by that loop | # Changed IPTables download URL | # version 2.3.0-pre1a: Initial branch from 2.1.12 | # Add stuff in release notes except port fw | # ----------------------------------------------------------------------| # USE AT YOUR OWN RISK! THIS IS WITHOUT ANY WARRANTY STATED OR IMPLIED! | # That means NONE whatsoever, not even fitfullness for purpose! | # ----------------------------------------------------------------------| # You NEED to set this! | # Configuration follows: | # | # Notes about configuration: | # Some things take more than one option; separate with spaces. | # You probably don't want all the ports I have under here open, portscan| # yourself to find what you want open. | # If you want to used host-based identd allowing, do NOT put 113 in | # TCP_ALLOW and DO set ALLOW_TCP_HOSTWISE (using 113 as the port).| # Of course, you can also put 113 in TCP_ALLOW to allow anyone. | # The same applies to DNS zone transfers (only use port 53 and UDP). | # MAC_MASQ is ONLY used for the purposes of masquerading and it will | # override the MASQ_LAN setting for masquerading. However, you | # must still define MASQ_LAN properly. | # INTERNAL_LAN must always be properly defined. | # You can use hostnames anywhere, but you'll need to have access to the | # DNS server when the script runs and you might not get the expected | # results since the DNS lookup is only done once. | # You can mix and match hosts with public IPs and masqueraded hosts in | # INTERNAL_LAN as long as you define the ones to use NAT later. | # DMZ support can currently be considered (at best) PREALPHA. | # It should work without a LAN, leave INTERNAL_LAN blank. | # Main configuration, modify to suit your setup. Help can be found at: # http://www.mplug.org/phpwiki/index.php?MonMothaReferenceGuide IPTABLES="/sbin/iptables" # set to your iptables location, must be set TCP_ALLOW="22" # TCP ports to allow UDP_ALLOW="68 6112 6119 4000" # UDP ports to allow INET_IFACE="eth1" # the interface your internet's on (one only), must be set LAN_IFACE="eth0" # the interface your LAN's on (one only) INTERNAL_LAN="192.168.0.0/24 192.168.1.0/24" # The internal LAN (including DMZs but not censored hosts) MASQ_LAN="192.168.0.0/24 192.168.1.0/24" # the internal network(s) to be masqueraded (this is overridden by MAC_MASQ) SNAT_LAN="" # Internal networks/hosts to use static NAT (format is :) (this is overridden by MAC_SNAT) DROP="TREJECT" # What to do with packets we don't want: DROP, REJECT, TREJECT (Reject with tcp-reset for TCP), LDROP (log and drop), LREJECT (log and reject), LTREJECT (log and reject with tcp-reset) DENY_ALL="" # Internet hosts to explicitly deny from accessing your system at all DENY_HOSTWISE_TCP="" # Specific hosts to deny access to specific TCP ports; format is "IP>PORT" DENY_HOSTWISE_UDP="" # Specific hosts to deny access to specific UDP ports; format is "IP>PORT" BLACKHOLE="" # People you don't want to have anything to do with (equivlent of my old TK_DROP). This is a bidirectional drop. BLACKHOLE_DROP="DROP" # What to do for the blackholes (same options as DROP directive above) ALLOW_HOSTWISE_TCP="" # Specific hosts allowed access to specific TCP ports; format is "IP>PORT" ALLOW_HOSTWISE_UDP="" # Specific hosts allowed access to specific UDP ports; format is "IP>PORT" TCP_FW="" # TCP port forwards, form is "SPORT:DPORT>IP" UDP_FW="" # UDP port forwards, form is "SPORT:DPORT>IP" MANGLE_TOS_OPTIMIZE="TRUE" # TOS "optimizations" on or off (TRUE/FALSE toggle) ENABLE="Y" # Set to 'Y' when it's configured; this is for your own saftey # Flood Params. You will still recieve the packets and the bandwidth will be used, but this will cause floods to be ignored (useful against SYNFLOODS especially) LOG_FLOOD="2/s" # Limit on logging (for LTREJECT, LREJECT and LDROP, the packet will always take the policy regardless of logging) SYN_FLOOD="20/s" # GLOBAL limit on SYN packets (servers will probably need even higher sustained rates as this isn't on a per IP basis) PING_FLOOD="1/s" # GLOBAL limit on ICMP echo-requests to reply to # Outbound filters (they work, but are of limited functionality), probably better to use a proxy here ALLOW_OUT_TCP="" # Internal hosts allowed to be forwarded out on TCP (internet censorship!) (do not put this/these host/s in INTERNAL_LAN, but do define their method of access [snat, masq] if not a public ip) # Below here is experimental (please report your successes/failures) MAC_MASQ="" # MAC addresses permitted to use masquerading, leave blank to not use MAC_SNAT="" # MAC addresses permitted to use static NAT, leave blank to not use (format is :) TTL_SAFE="" # How many hops packets need to make once they get on your LAN (null disables the mangling) (requires patch from patch-o-matic) USE_SYNCOOKIES="TRUE" # TCP SynCookies on or off (TRUE/FALSE toggle) PROXY="" # Redirect for Squid or other transparent proxy. Syntax to specify the proxy is "host:port". DHCP_SERVER="FALSE" # Set to true if you run a DHCP server. DHCP clients do not need this. This allows broadcasts to the server from potential clients on the LAN to succeede. MUST DEFINE LAN_IFACE IF YOU USE THIS! # Only touch these if you're daring (PREALPHA stuff, as in basically non-functional) DMZ_IFACE="" # Interface your DMZ is on (leave blank if you don't have one) MUST DEFINE LAN_IFACE IF YOU USE THIS! # ----------------------------------------------------------------------| # These control basic script behavior, there should be no need to | # any of these settings for normal use. | # ----------------------------------------------------------------------| FILTER_CHAINS="INETIN INETOUT DMZIN DMZOUT TCPACCEPT UDPACCEPT LDROP LREJECT TREJECT LTREJECT" LOOP_IFACE="lo" # ----------------------------------------------------------------------| # You shouldn't need to modify anything below here | # Main Script Starts | # ----------------------------------------------------------------------| # Let's load it! echo "Loading iptables firewall:" # Configuration Sanity Checks echo -n "Checking configuration..." if ! [ -x $IPTABLES ] ; then echo echo "ERROR IN CONFIGURATION: IPTABLES doesn't exist or isn't executable!" exit 1 fi if [ "$DMZ_IFACE" = "$LAN_IFACE" ] && [ "$LAN_IFACE" != "" ]; then echo echo "ERROR IN CONFIGURATION: DMZ_IFACE and LAN_IFACE can't be the same!" exit 1 fi if [ "$DROP" = "" ] ; then echo echo "There needs to be a DROP policy (try TREJECT)!" exit 1 fi if [ "$DROP" = "ACCEPT" ] ; then echo echo "The DROP policy is set to ACCEPT; there is no point in loading the firewall as there wouldn't be one." exit 2 fi if [ "$BLACKHOLE" != "" ] && [ "$BLACKHOLE_DROP" = "" ] ; then echo echo "You can't use blackholes and not have a policy for them!" exit 1 fi if ! [ "$ENABLE" = "Y" ] ; then echo echo "You need to edit your configuration and set ENABLE to Y!" exit 99 fi echo "passed" # =============================================== # -------Set some Kernel stuff via SysCTL-------- # =============================================== # Turn on IP forwarding if [ "$INTERNAL_LAN" != "" ] ; then echo -n "Checking IP Forwarding..." if [ -e /proc/sys/net/ipv4/ip_forward ] ; then echo 1 > /proc/sys/net/ipv4/ip_forward echo "enabled." else echo "support not found! This will cause problems if you need to do any routing." fi fi # Enable TCP Syncookies if [ "$USE_SYNCOOKIES" = "TRUE" ] ; then echo -n "Checking IP SynCookies..." if [ -e /proc/sys/net/ipv4/tcp_syncookies ] ; then echo 1 > /proc/sys/net/ipv4/tcp_syncookies echo "enabled." else echo "support not found, but that's OK." fi fi # Enable Route Verification to prevent martians and other such crud that # seems to be commonplace on the internet today echo -n "Checking Route Verification..." if [ "$INET_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$INET_IFACE/rp_filter ] ; then echo 1 > /proc/sys/net/ipv4/conf/$INET_IFACE/rp_filter echo -n "activated:${INET_IFACE} " else echo "not found:${INET_IFACE} " fi fi if [ "$LAN_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$LAN_IFACE/rp_filter ] ; then echo 1 > /proc/sys/net/ipv4/conf/$LAN_IFACE/rp_filter echo -n "activated:${LAN_IFACE} " else echo "not found:${LAN_IFACE} " fi fi if [ "$DMZ_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$DMZ_IFACE/rp_filter ] ; then echo 1 > /proc/sys/net/ipv4/conf/$DMZ_IFACE/rp_filter echo -n "activated:${DMZ_IFACE} " else echo "not found:${DMZ_IFACE} " fi fi echo # Tell the Kernel to Ignore Source Routed Packets echo -n "Refusing SSR Packets via SysCtl..." if [ "$INET_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$INET_IFACE/accept_source_route ] ; then echo "0" > /proc/sys/net/ipv4/conf/$INET_IFACE/accept_source_route echo -n "activated:${INET_IFACE} " else echo "not found:${INET_IFACE} " fi fi if [ "$LAN_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$LAN_IFACE/accept_source_route ] ; then echo "0" > /proc/sys/net/ipv4/conf/$LAN_IFACE/accept_source_route echo -n "activated:${LAN_IFACE} " else echo "not found:${LAN_IFACE} " fi fi if [ "$DMZ_IFACE" != "" ] ; then if [ -e /proc/sys/net/ipv4/conf/$DMZ_IFACE/accept_source_route ] ; then echo "0" > /proc/sys/net/ipv4/conf/$DMZ_IFACE/accept_source_route echo -n "activated:${DMZ_IFACE} " else echo "not found:${DMZ_IFACE} " fi fi echo # =============================================== # --------Actual NetFilter Stuff Follows--------- # =============================================== # Flush everything # If you need compatability, you can comment some or all of these out, # but remember, if you re-run it, it'll just add the new rules in, it # won't remove the old ones for you then, this is how it removes them. echo -n "Flush: " ${IPTABLES} -t filter -F INPUT echo -n "INPUT " ${IPTABLES} -t filter -F OUTPUT echo -n "OUTPUT1 " ${IPTABLES} -t filter -F FORWARD echo -n "FORWARD " ${IPTABLES} -t nat -F PREROUTING echo -n "PREROUTING1 " ${IPTABLES} -t nat -F OUTPUT echo -n "OUTPUT2 " ${IPTABLES} -t nat -F POSTROUTING echo -n "POSTROUTING " ${IPTABLES} -t mangle -F PREROUTING echo -n "PREROUTING2 " ${IPTABLES} -t mangle -F OUTPUT echo -n "OUTPUT3" echo # Create new chains # Output to /dev/null in case they don't exist from a previous invocation echo -n "Creating chains: " for chain in ${FILTER_CHAINS} ; do ${IPTABLES} -t filter -F ${chain} > /dev/null 2>&1 ${IPTABLES} -t filter -X ${chain} > /dev/null 2>&1 ${IPTABLES} -t filter -N ${chain} echo -n "${chain} " done echo # Default Policies # INPUT policy is drop as of 2.3.7-pre5 # Policy can't be reject becuase of kernel limitations echo -n "Default Policies: " ${IPTABLES} -t filter -P INPUT DROP echo -n "INPUT:DROP " ${IPTABLES} -t filter -P OUTPUT ACCEPT echo -n "OUTPUT:ACCEPT " ${IPTABLES} -t filter -P FORWARD DROP echo -n "FORWARD:DROP " echo # Local traffic to internet or crossing subnets # This should cover what we need if we don't use masquerading # Unfortunately, MAC address matching isn't bidirectional (for # obvious reasons), so IP based matching is done here echo -n "Local Traffic Rules: " if [ "$INTERNAL_LAN" != "" ] ; then for subnet in ${INTERNAL_LAN} ; do ${IPTABLES} -t filter -A FORWARD -s ${subnet} -j ACCEPT ${IPTABLES} -t filter -A FORWARD -d ${subnet} -j ACCEPT ${IPTABLES} -t filter -A INPUT -s ${subnet} -j ACCEPT echo -n "${subnet}:ACCEPT " done echo fi # 127.0.0.0/8 used to need an entry in INTERNAL_LAN, but routing of that isn't needed # so an allow is placed on INPUT so that the computer can talk to itself :) ${IPTABLES} -t filter -A INPUT -i ${LOOP_IFACE} -j ACCEPT echo -n "loopback:ACCEPT " # DHCP server magic # Allow broadcasts from LAN to UDP port 67 (DHCP server) if [ "$DHCP_SERVER" = "TRUE" ] ; then ${IPTABLES} -t filter -A INPUT -i ${LAN_IFACE} -p udp --dport 67 -j ACCEPT fi echo #newline from local traffic rules if [ "$PROXY" != "" ] ; then echo -n "Setting up Transparent Proxy to ${PROXY}: " for subnet in ${INTERNAL_LAN} ; do echo "$PROXY" | { IFS=':' read host port if [ "$host" = "localhost" ] || [ "$host" = "127.0.0.1" ] ; then ${IPTABLES} -t nat -A PREROUTING -s ${subnet} -p tcp --dport 80 -j REDIRECT --to-port ${port} echo -n "${subnet}:PROXY " else ${IPTABLES} -t nat -A PREROUTING -s ${subnet} -p tcp --dport 80 -j DNAT --to ${host}:${port} echo -n "${subnet}:PROXY " fi } done echo fi if [ "$ALLOW_OUT_TCP" != "" ] ; then echo -n "Internet censorship TCP allows: " for rule in ${ALLOW_OUT_TCP} ; do echo "$rule" | { IFS=':' read intip destip ${IPTABLES} -t filter -A FORWARD -s ${intip} -d ${destip} -o ${INET_IFACE} -j ACCEPT ${IPTABLES} -t filter -A FORWARD -d ${intip} -s ${destip} -i ${INET_IFACE} -j ACCEPT echo -n "${intip}:${destip} " } done echo fi # Set up basic NAT if the user wants it if [ "$MASQ_LAN" != "" ] ; then echo -n "Setting up masquerading: " if [ "$MAC_MASQ" = "" ] ; then for subnet in ${MASQ_LAN} ; do ${IPTABLES} -t nat -A POSTROUTING -s ${subnet} -o ${INET_IFACE} -j MASQUERADE echo -n "${subnet}:MASQUERADE " done else for address in ${MAC_MASQ} ; do ${IPTABLES} -t nat -A POSTROUTING -m mac --mac-source ${address} -o ${INET_IFACE} -j MASQUERADE echo -n "${address}:MASQUERADE " done fi echo fi if [ "$SNAT_LAN" != "" ] ; then #Static NAT used echo -n "Setting up static NAT: " if [ "$MAC_SNAT" = "" ] ; then for rule in ${SNAT_LAN} ; do echo "$rule" | { IFS=':' read host destip ${IPTABLES} -t nat -A POSTROUTING -s ${host} -o ${INET_IFACE} -j SNAT --to-source ${destip} echo -n "${subnet}:SNAT " } done else for rule in ${MAC_SNAT} ; do echo "$rule" | { IFS=':' read address destip ${IPTABLES} -t nat -A POSTROUTING -m mac --mac-source ${address} -o ${INET_IFACE} -j SNAT --to-source ${destip} echo -n "${address}:SNAT " } done fi echo fi #TCP Port-Forwards if [ "$TCP_FW" != "" ] ; then echo -n "TCP Port Forwards: " for rule in ${TCP_FW} ; do echo "$rule" | { IFS=':>' read srcport destport host echo "$srcport" | { IFS='-' read fsp lsp if [ "$lsp" != "" ] ; then echo "$destport" | { IFS='-' read fdp ldp ${IPTABLES} -t nat -A PREROUTING -p tcp -i ${INET_IFACE} --dport ${fsp}:${lsp} -j DNAT --to-destination ${host}:${destport} ${IPTABLES} -t filter -A FORWARD -p tcp -d ${host} --dport ${fdp}:${ldp} -j ACCEPT } else ${IPTABLES} -t nat -A PREROUTING -p tcp -i ${INET_IFACE} --dport ${srcport} -j DNAT --to-destination ${host}:${destport} ${IPTABLES} -t filter -A FORWARD -p tcp -d ${host} --dport ${destport} -j ACCEPT fi echo -n "${rule} " } } done echo fi #UDP Port Forwards if [ "$UDP_FW" != "" ] ; then echo -n "UDP Port Forwards: " for rule in ${UDP_FW} ; do echo "$rule" | { IFS=':>' read srcport destport host echo "$srcport" | { IFS='-' read fsp lsp if [ "$lsp" != "" ] ; then echo "$destport" | { IFS='-' read fdp ldp ${IPTABLES} -t nat -A PREROUTING -p udp -i ${INET_IFACE} --dport ${fsp}:${lsp} -j DNAT --to-destination ${host}:${destport} ${IPTABLES} -t filter -A FORWARD -p udp -d ${host} --dport ${fdp}:${ldp} -j ACCEPT } else ${IPTABLES} -t nat -A PREROUTING -p udp -i ${INET_IFACE} --dport ${srcport} -j DNAT --to-destination ${host}:${destport} ${IPTABLES} -t filter -A FORWARD -p udp -d ${host} --dport ${destport} -j ACCEPT fi echo -n "${rule} " } } done echo fi # =============================================== # -------Chain setup before jumping to them------ # =============================================== # Set up INET chains echo -n "Setting up INET chains: " ${IPTABLES} -t filter -A INPUT -i ${INET_IFACE} -j INETIN echo -n "INETIN " ${IPTABLES} -t filter -A OUTPUT -o ${INET_IFACE} -j INETOUT echo -n "INETOUT " echo # For now we'll subject the DMZ to the same rules as the internet when going onto the trusted LAN # And we'll let it go anywhere on the internet if [ "$DMZ_IFACE" != "" ] ; then echo -n "Setting up DMZ Chains: " ${IPTABLES} -A OUTPUT -o ${DMZ_IFACE} -j DMZOUT echo -n "DMZOUT " ${IPTABLES} -A INPUT -i ${DMZ_IFACE} -j DMZIN echo -n "DMZIN " echo echo -n "DMZ for LAN Forwarding to INETIN..." ${IPTABLES} -A DMZOUT -o ${LAN_IFACE} -j INETIN echo "done" echo -n "DMZ for Internet Forwarding to INETOUT..." ${IPTABLES} -A DMZOUT -o ${INET_IFACE} -j INETOUT echo -n "done" fi #These logging chains are valid to specify in DROP= above #Set up LDROP echo -n "Setting up drop chains chains: " ${IPTABLES} -t filter -A LDROP -p tcp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "TCP Dropped " ${IPTABLES} -t filter -A LDROP -p udp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "UDP Dropped " ${IPTABLES} -t filter -A LDROP -p icmp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "ICMP Dropped " ${IPTABLES} -t filter -A LDROP -f -m limit --limit ${LOG_FLOOD} -j LOG --log-level warning --log-prefix "FRAGMENT Dropped " ${IPTABLES} -t filter -A LDROP -j DROP echo -n "LDROP " #And LREJECT too ${IPTABLES} -t filter -A LREJECT -p tcp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "TCP Rejected " ${IPTABLES} -t filter -A LREJECT -p udp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "UDP Rejected " ${IPTABLES} -t filter -A LREJECT -p icmp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "ICMP Dropped " ${IPTABLES} -t filter -A LREJECT -f -m limit --limit ${LOG_FLOOD} -j LOG --log-level warning --log-prefix "FRAGMENT Rejected " ${IPTABLES} -t filter -A LREJECT -j REJECT echo -n "LREJECT " #Don't forget TREJECT ${IPTABLES} -t filter -A TREJECT -p tcp -j REJECT --reject-with tcp-reset ${IPTABLES} -t filter -A TREJECT -p udp -j REJECT --reject-with icmp-port-unreachable ${IPTABLES} -t filter -A TREJECT -p icmp -j DROP ${IPTABLES} -t filter -A TREJECT -j REJECT echo -n "TREJECT " #And LTREJECT ${IPTABLES} -t filter -A LTREJECT -p tcp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "TCP Rejected " ${IPTABLES} -t filter -A LTREJECT -p udp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "UDP Rejected " ${IPTABLES} -t filter -A LTREJECT -p icmp -m limit --limit ${LOG_FLOOD} -j LOG --log-level info --log-prefix "ICMP Dropped " ${IPTABLES} -t filter -A LTREJECT -f -m limit --limit ${LOG_FLOOD} -j LOG --log-level warning --log-prefix "FRAGMENT Rejected " ${IPTABLES} -t filter -A LTREJECT -p tcp -j REJECT --reject-with tcp-reset ${IPTABLES} -t filter -A LTREJECT -p udp -j REJECT --reject-with icmp-port-unreachable ${IPTABLES} -t filter -A LTREJECT -p icmp -j DROP ${IPTABLES} -t filter -A LTREJECT -j REJECT echo -n "LTREJECT " #newline echo # Set up the per-proto ACCEPT chains echo -n "Setting up per-proto ACCEPT: " # TCPACCEPT # SYN Flood "Protection" ${IPTABLES} -t filter -A TCPACCEPT -p tcp --syn -m limit --limit ${SYN_FLOOD} -j ACCEPT ${IPTABLES} -t filter -A TCPACCEPT -p tcp --syn -m limit --limit ${LOG_FLOOD} -j LOG --log-prefix "Possible SynFlood " ${IPTABLES} -t filter -A TCPACCEPT -p tcp --syn -j ${DROP} ${IPTABLES} -t filter -A TCPACCEPT -p tcp ! --syn -j ACCEPT # Log anything that hasn't matched yet and ${DROP} it since it isn't TCP and shouldn't be here ${IPTABLES} -t filter -A TCPACCEPT -m limit --limit ${LOG_FLOOD} -j LOG --log-prefix "Mismatch in TCPACCEPT " ${IPTABLES} -t filter -A TCPACCEPT -j ${DROP} echo -n "TCPACCEPT " #UDPACCEPT ${IPTABLES} -t filter -A UDPACCEPT -p udp -j ACCEPT # Log anything not UDP and ${DROP} it since it's not supposed to be here ${IPTABLES} -t filter -A UDPACCEPT -m limit --limit ${LOG_FLOOD} -j LOG --log-prefix "Mismatch on UDPACCEPT " ${IPTABLES} -t filter -A UDPACCEPT -j ${DROP} echo -n "UDPACCEPT " #Done echo # =============================================== # --------------Apply configuration-------------- # =============================================== #Blackholes will not be overridden by hostwise allows if [ "$BLACKHOLE" != "" ] ; then echo -n "Blackholes: " for host in ${BLACKHOLE} ; do ${IPTABLES} -t filter -A INETIN -s ${host} -j ${BLACKHOLE_DROP} ${IPTABLES} -t filter -A INETOUT -d ${host} -j ${BLACKHOLE_DROP} echo -n "${host} " done echo fi #Hostwise allows (to bypass explicit denies) if [ "$ALLOW_HOSTWISE_TCP" != "" ] ; then echo -n "Hostwise TCP Allows: " for rule in ${ALLOW_HOSTWISE_TCP} ; do echo "$rule" | { IFS='>' read host port ${IPTABLES} -t filter -A INETIN -p tcp -s ${host} --dport ${port} -j ACCEPT echo -n "${rule} " } done echo fi if [ "$ALLOW_HOSTWISE_UDP" != "" ] ; then echo -n "Hostwise UDP Allows: " for rule in ${ALLOW_HOSTWISE_UDP} ; do echo "$rule" | { IFS='>' read host port ${IPTABLES} -t filter -A INETIN -p udp -s ${host} --dport ${port} -j ACCEPT echo -n "${rule} " } done echo fi # ================================================= # ----------------Explicit Denies------------------ # ================================================= if [ "$DENY_ALL" != "" ] ; then echo -n "Denying hosts: " for host in ${DENY_ALL} ; do ${IPTABLES} -t filter -A INETIN -s ${host} -j ${DROP} echo -n "${host}:${DROP}" done echo fi if [ "$DENY_HOSTWISE_TCP" != "" ] ; then echo -n "Hostwise TCP Denies: " for rule in ${DENY_HOSTWISE_TCP} ; do echo "$rule" | { IFS='>' read host port ${IPTABLES} -t filter -A INETIN -p tcp -s ${host} --dport ${port} -j ${DROP} echo -n "${rule} " } done echo fi if [ "$DENY_HOSTWISE_UDP" != "" ] ; then echo -n "Hostwise UDP Denies: " for rule in ${DENY_HOSTWISE_UDP} ; do echo "$rule" | { IFS='>' read host port ${IPTABLES} -t filter -A INETIN -p udp -s ${host} --dport ${port} -j ${DROP} echo -n "${rule} " } done echo fi #Invalid packets are always annoying echo -n "${DROP}ing invalid packets..." ${IPTABLES} -t filter -A INETIN -m state --state INVALID -j ${DROP} echo "done" # ================================================================ # ------------Allow stuff we have chosen to allow in-------------- # ================================================================ # Flood "security" # You'll still respond to these if they comply with the limits (set in config) # There is a more elegant way to set this using sysctl, however this has the # advantage that the kernel ICMP stack never has to process it, lessening # the chance of a very serious flood overloading your kernel. # This is just a packet limit, you still get the packets on the interface and # still may experience lag if the flood is heavy enough echo -n "Flood limiting: " # Ping Floods (ICMP echo-request) ${IPTABLES} -t filter -A INETIN -p icmp --icmp-type echo-request -m limit --limit ${PING_FLOOD} -j ACCEPT ${IPTABLES} -t filter -A INETIN -p icmp --icmp-type echo-request -j ${DROP} echo -n "ICMP-PING " echo # Probably shouldn't let em all back in, someone give me a good list of the # WORTHWHILE ICMP messages. echo -n "Allowing the rest of the ICMP messages in..." ${IPTABLES} -t filter -A INETIN -p icmp --icmp-type ! echo-request -j ACCEPT echo "done" if [ "$TCP_ALLOW" != "" ] ; then echo -n "TCP Input Allow: " for port in ${TCP_ALLOW} ; do ${IPTABLES} -t filter -A INETIN -p tcp --dport ${port} -j TCPACCEPT echo -n "${port} " done echo fi if [ "$UDP_ALLOW" != "" ] ; then echo -n "UDP Input Allow: " for port in ${UDP_ALLOW} ; do ${IPTABLES} -t filter -A INETIN -p udp --dport ${port} -j UDPACCEPT echo -n "${port} " done echo fi echo -n "Allowing established outbound connections back in..." ${IPTABLES} -t filter -A INETIN -m state --state ESTABLISHED -j ACCEPT echo "done" # RELATED on high ports only for security echo -n "Allowing related inbound connections..." ${IPTABLES} -t filter -A INETIN -p tcp --dport 1024:65535 -m state --state RELATED -j TCPACCEPT ${IPTABLES} -t filter -A INETIN -p udp --dport 1024:65535 -m state --state RELATED -j UDPACCEPT echo "done" # ================================================= # ----------------Packet Mangling------------------ # ================================================= # TTL mangling # This is probably just for the paranoid, but hey, isn't that what # all security guys are? :) if [ "$TTL_SAFE" != "" ] ; then ${IPTABLES} -t mangle -A PREROUTING -i ${INET_IFACE} -j TTL --ttl-set ${TTL_SAFE} fi # Type of Service mangle optimizations (the ACTIVE FTP one will only work for uploads) if [ "$MANGLE_TOS_OPTIMIZE" = "TRUE" ] ; then echo -n "Optimizing traffic: " ${IPTABLES} -t mangle -A OUTPUT -p tcp --dport 23 -j TOS --set-tos Minimize-Delay echo -n "telnet " ${IPTABLES} -t mangle -A OUTPUT -p tcp --dport 22 -j TOS --set-tos Minimize-Delay echo -n "ssh " ${IPTABLES} -t mangle -A OUTPUT -p tcp --dport 20 -j TOS --set-tos Minimize-Cost echo -n "ftp-data " ${IPTABLES} -t mangle -A OUTPUT -p tcp --dport 21 -j TOS --set-tos Minimize-Delay echo -n "ftp-control " ${IPTABLES} -t mangle -A OUTPUT -p udp --dport 4000:7000 -j TOS --set-tos Minimize-Delay echo -n "diablo2 " echo fi #What to do on those INET chains when we hit the end echo -n "Setting up INET policies: " #Drop if we cant find a valid inbound rule. ${IPTABLES} -t filter -A INETIN -j ${DROP} echo -n "INETIN:${DROP} " #We can send what we want to the internet ${IPTABLES} -t filter -A INETOUT -j ACCEPT echo -n "INETOUT:ACCEPT " echo #All done! echo "Done loading the firewall!" Anyone have an idea? David