Copyright (C) 2005-2007 Pablo Neira Ayuso 0.Introduction ============== The conntrack-tools package contains two programs: - conntrack: the command line interface to interact with the connection tracking system. - conntrackd: the connection tracking userspace daemon that can be used to deploy highly available GNU/Linux firewalls and collect statistics of the firewall use. 1. Requirements =============== You have to install the following software in order to get the conntrack-tools working, make sure that you have installed them correctly before going forward: o linux kernel version >= 2.6.18 (http://www.kernel.org) with support for: - connection tracking system (quite obvious ;) - nfnetlink - ctnetlink (ip_conntrack_netlink) - connection tracking event notification API o libnfnetlink: the netfilter netlink library available at: o libnetfilter_conntrack: the netfilter conntrack library available at: 2.Basic Installation ==================== To compile and install conntrack-tools just follow the classical steps: $ ./configure $ make # make install Up to this point, the command line interface `conntrack' is ready for use. However, the userspace daemon so-called `conntrackd' requires some magic speells to get it working. 3.Setting up conntrackd ======================= conntrackd currently have two working modes: statistics and synchronization modes, both details here below. 3.1. Synchronization Mode ========================= Conntrackd can replicate the status of the connections that are currently being processed by your stateful firewall based on Linux. This section describes how to setup the daemon in synchronization mode: o Keepalived version 1.x (http://www.keepalived.org) check if your distribution comes with a recent version 3.1.2. Configuration 1) Setting up keepalived There is an example file available inside the conntrackd tarball: For node 1: conntrackd-x.x.x/examples/sync/node1/keepalived.conf For node 2: conntrackd-x.x.x/examples/sync/node2/keepalived.conf These files can be used to set up a simple VRRP cluster composed of two machines that hold the virtual IPs 192.168.0.100 on eth0 and 192.168.1.100 on eth1. If you are not familiar with keepalived, please read the official docs available at http://www.keepalived.org Please, make sure that keepalived is correctly working before passing to step 2) 2) Setting up conntrackd To setup 'conntrackd' in synchronization mode, you have to put the configuration file in the directory /etc/conntrackd. On node 1: # cp examples/sync/_type_/node1/conntrackd.conf /etc/conntrackd.conf On node 2: # cp examples/sync/_type_/node1/conntrackd.conf /etc/conntrackd.conf Where _type_ is the synchronization type selected, currently there are two: the persistent mode and the NACK mode. The persistent mode consumes more resources than the NACK mode, however the NACK mode is still experimental Do not forget to edit the files in order to adapt them to the setting that you are deploying. Note: If you don't want to put the config file under /etc/conntrackd, just tell conntrackd where to find it passing the option -C 3) Running conntrackd Conntrackd can run in console mode, in that case just type 'conntrackd', otherwise, if you want to run it in daemon mode the type 'conntrackd -d'. 4) Checking that conntrackd is working fine Conntrackd comes with several facilities to check its status: - Dump the cache of connections that are currently being processed by this node (aka. internal cache): # conntrackd -i - Dump the cache of connections that has been transfered from others active nodes in the network (aka. external cache) # conntrackd -e - Dump statistics collected by the replication daemon: # conntrackd -s 5) Setting up interaction with keepalived If keepalived detects the failure of the active node, then it designates a candidate node that will replace the failing active. On such event, the external cache, eg. the cache that contains the connections processed by other nodes, must be commited. To commit the external cache, just type: # conntrackd -c See that keepalived provides a shell script interface to interact with other programs, so we can automate the process of commiting the external cache by introducing the following line in the keepalived file: notify_master /etc/conntrackd/script_master.sh The script 'script_master.sh' just the following: #!/bin/sh /usr/sbin/conntrackd -c Therefore, on failure event, the candidate node takes over the virtual IPs and the connections that the failing active was processing. Observe that this file differs for the NACK mode. 6) Disable TCP window tracking Until the appropiate patches don't go into kernel mainline, you will have to disable TCP window tracking, consider this as a temporary solution: # echo 1 > /proc/sys/net/ipv4/netfilter/ip_conntrack_tcp_be_liberal 3.2. Statistics mode ==================== Conntrackd can also run as statistics daemon, if you are not interested in this mode, just skip it. It is not required in order to get the synchronization mode working. This section details how to setup the daemon in statistics mode: 3.2.1. Requirements No extra requirements to set up the statistics mode apart from those detailed in section 1. 3.2.2. Configuration Setting up conntrackd in statistics mode is rather easy. Just copy the configuration file # cp examples/stats/conntrackd.conf /etc/conntrackd.conf 3.2.3. Running conntrackd in statistics mode To run conntrackd in statistics mode: # conntrackd -S Alternatively, you can run conntrackd in daemon mode: # conntrackd -S -d In order to dump the statistics, just type: # conntrackd -s