############################################################################### # Copyright 2006-2012, Way to the Web Limited # URL: http://www.configserver.com # Email: sales@waytotheweb.com ############################################################################### ConfigServer Security & Firewall ################################ This suite of scripts provides: 1. A straight-forward SPI iptables firewall script 2. A daemon process that checks for Login Authentication 3. A Control Panel configuration interface 4. ... and much more! The reason we have developed this suite is that we have found over the years of providing server management services that many of the tools available for the task are either over-complex, not very friendly, or simply aren't as effective as they could or should be. This document contains: 1. Introduction 2. csf Principles 3. lfd Principles 4. csf Command Line Options 5. lfd Command Line Options 6. Login Tracking 7. Script Email Alerts 8. Process Tracking 9. Directory Watching 10. Advanced Allow/Deny Filters 11. Multiple Ethernet Devices 12. Installation on a Generic Linux Server 13. A note about FTP Connection Issues 14. Messenger Service 15. Block Reporting 16. Port Flood Protection 17. External Pre- and Post- Scripts 18. lfd Clustering 19. Watching IP addresses 20. Port Knocking 21. Connection Limit Protection 22. Port/IP address Redirection 23. Integrated User Interface Feature 1. Introduction ############### ConfigServer Firewall (csf) =========================== We have developed an SPI iptables firewall that is straight-forward, easy and flexible to configure and secure with extra checks to ensure smooth operation. csf can be used on any (supported - see the website) generic Linux OS. The csf installation includes preconfigured configurations and control panel UI's for cPanel, DirectAdmin and Webmin Login Failure Daemon (lfd) ========================== To complement the ConfigServer Firewall, we have developed a daemon process that runs all the time and periodically (every X seconds) scans the latest log file entries for login attempts against your server that continually fail within a short period of time. Such attempts are often called "Brute-force attacks" and the daemon process responds very quickly to such patterns and blocks offending IP's quickly. Other similar products run every x minutes via cron and as such often miss break-in attempts until after they've finished, our daemon eliminates such long waits and makes it much more effective at performing its task. There are an array of extensive checks that lfd can perform to help alert the server administrator of changes to the server, potential problems and possible compromises. On cPanel servers, lfd is integrated into the WHM > Service Manager, which will restart lfd if it fails for any reason. Control Panel Interface ======================= To help with the ease and flexibility of the suite we have developed a front-end to both csf and lfd for cPanel, DirectAdmin and Webmin. From there you can modify the configuration files and stop, start and restart the applications and check their status. This makes configuring and managing the firewall very simple indeed. There is also an abbreviated UI for mobile phone access to Quick Allow, Quick Deny and Remove Deny. Direct URLs: cPanel: https://1.2.3.4:2087/cgi/addon_csf.cgi?mobi=1 DA: https://1.2.3.4:2222/CMD_PLUGINS_ADMIN/csf/index.html?mobi=1 Webmin: https://1.2.3.4:10000/csf/?mobi=1 There is, of course, a comprehensive Command Line Interface (CLI) for csf. 2. csf Principles ################# The idea with csf, as with most iptables firewall configurations, is to block everything and then allow through only those connections that you want. This is done in iptables by DROPPING all connections in and out of the server on all protocols. Then allow traffic in and out from existing connections. Then open ports up in and outgoing for both TCP and UDP individually. This way we can control exactly what traffic is allowed in and out of the server and helps protect the server from malicious attack. In particular it prevents unauthorised access to network daemons that we want to restrict access by IP address, and also should a service suffer a compromise, it can help prevent access to compromise networks daemons, a typical example being a hackers sshd daemon running on a random open port. Perhaps the greatest of reasons is to help mitigate the effects of suffering a root compromise where often they only way to take advantage of such a failure is to open a daemon for the hacker to access the server on. While this won't prevent root compromises, it can help slow them down enough for you to notice and react. Another way that a port filtering firewall can help is when a user level compromise occurs and a hacker installs DOS tools to effect other servers. A firewall configured to block outgoing connections except on specific ports can help prevent DOS attacks from working and make it immediately apparent to you from the system logs. csf has been designed to keep this configuration simple, but still flexible enough to give you options to suit your server environment. Often firewall scripts can become cumbersome of complex making it impossible to identify where problems lie and to easily fix them. To take advantage of kernel logging of iptables dropped connections you should ensure that kernel logging daemon (klogd) is enabled. Typically, VPS servers have this disabled and you should check /etc/init.d/syslog and make sure that any klogd lines are not commented out. If you change the file, remember to restart syslog. 3. lfd Principles ################# One of the best ways to protect the server from inbound attack against network daemons is to monitor their authentication logs. Invalid login attempts which happen in a short space of time from the same source can often mean someone is attempting to brute-force their way into the server, usually by guessing usernames and passwords and therefore generating authentication and login failures. lfd can monitor the most commonly abused protocols, SSHD, POP3, IMAP, FTP and HTTP password protection. Unlike other applications, lfd is a daemon process that monitors logs continuously and so can react within seconds of detecting such attempts. It also monitors across protocols, so if attempts are made on different protocols in a short space of time, all those attempts will be counted against the threshold. Once the number of failed login attempts is reached, lfd immediately forks a sub-process and uses csf to block the offending IP address from both in and outgoing connections. Stopping the attack in its tracks in a quick and timely manner. Other applications that use cron job timings to run usually completely miss brute force attacks as they run usually every 5 minutes or by which time the attack could be over, or simply biding its time. In the meantime lfd will have block the offenders IP address. By running the block and alert email actions in a sub-process, the main daemon can continue monitoring the logs without delay. If you want to know when lfd blocks an IP address you can enable the email alert (which is on by default) and you should watch the log file in /var/log/lfd.log. If you use logcheck, you can add it to your log monitoring by editing logcheck.sh and adding the line: $LOGTAIL /var/log/lfd.log >> $TMPDIR/check.$$ Add it in amongst the other logs that you have selected. 4. csf Command Line Options ########################### Before configuring and starting csf for the first time, it is a good idea to run the script /etc/csf/csftest.pl using: perl /etc/csf/csftest.pl This script will test whether the required iptables modules are functioning on the server. Don't worry if it cannot run all the features, so long as the script doesn't report any FATAL errors. You can view the csf command line options by using: # csf -h Usage: /usr/sbin/csf [option] [value] Option Meaning -h, --help Show this message -l, --status List/Show iptables configuration -l6, --status6 List/Show ip6tables configuration -s, --start Start firewall rules -f, --stop Flush/Stop firewall rules (Note: lfd may restart csf) -r, --restart Restart firewall rules -q, --startq Quick restart (csf restarted by lfd) -sf, --startf Force CLI restart regardless of LF_QUICKSTART setting -a, --add ip Allow an IP and add to /etc/csf.allow -ar, --addrm ip Remove an IP from /etc/csf.allow and delete rule -d, --deny ip Deny an IP and add to /etc/csf.deny -dr, --denyrm ip Unblock an IP and remove from /etc/csf.deny -df, --denyf Remove and unblock all entries in /etc/csf.deny -g, --grep ip Search the iptables rules for an IP match (incl. CIDR) -t, --temp Displays the current list of temp IP entries and their TTL -tr, --temprm ip Remove an IPs from the temp IP ban and allow list -td, --tempdeny ip ttl [-p port] [-d direction] Add an IP to the temp IP ban list. ttl is how long to blocks for (default:seconds, can use one suffix of h/m/d). Optional port. Optional direction of block can be one of: in, out or inout (default:in) -ta, --tempallow ip ttl [-p port] [-d direction] Add an IP to the temp IP allow list (default:inout) -tf, --tempf Flush all IPs from the temp IP entries -cp, --cping PING all members in an lfd Cluster -cd, --cdeny ip Deny an IP in a Cluster and add to /etc/csf.deny -ca, --callow ip Allow an IP in a Cluster and add to /etc/csf.allow -cr, --crm ip Unblock an IP in a Cluster and remove from /etc/csf.deny -cc, --cconfig [name] [value] Change configuration option [name] to [value] in a Cluster -cf, --cfile [file] Send [file] in a Cluster to /etc/csf/ -crs, --crestart Cluster restart csf and lfd -m, --mail [addr] Display Server Check in HTML or email to [addr] if present -lr, --logrun Initiate Log Scanner report via lfd -c, --check Check for updates to csf but do not upgrade -u, --update Check for updates to csf and upgrade if available -uf Force an update of csf -x, --disable Disable csf and lfd -e, --enable Enable csf and lfd if previously disabled -v, --version Show csf version These options allow you to easily and quickly control and view csf. All the configuration files for csf are in /etc/csf and include: csf.conf - the main configuration file, it has helpful comments explaining what each option does csf.allow - a list of IP's and CIDR addresses that should always be allowed through the firewall csf.deny - a list of IP's and CIDR addresses that should never be allowed through the firewall csf.ignore - a list of IP's and CIDR addresses that lfd should ignore and not not block if detected csf.*ignore - various ignore files that list files, users, IP's that lfd should ignore. See each file for their specific purpose and tax If you modify any of the files listed above, you will need to restart csf to have them take effect. If you use the command line options to add or deny IP addresses, then csf automatically does this for you. Both csf.allow and csf.deny can have comments after the IP address listed. The comments must be on the same line as the IP address otherwise the IP rotation of csf.deny will remove them. If editing the csf.allow or csf.deny files directly, either from shell or the WHM UI, you should put a # between the IP address and the comment like this: 11.22.33.44 # Added because I don't like them You can also include comments when using the csf -a or csf -d commands, but in those cases you must not use a # like this: csf -d 11.22.33.44 Added because I don't like them If you use the shell commands then each comment line will be timestamped. You will also find that if lfd blocks an IP address it will add a descriptive comment plus timestamp. If you don't want csf to rotate a particular IP in csf.deny if the line limit is reach you can do so by adding "do not delete" within the comment field, e.g.: 11.22.33.44 # Added because I don't like them. do not delete You can also use an Include statement in either csf.allow or csf.deny to include other files that conform to the above. You must specify the full path to the included file, e.g. in /etc/csf/csf.allow: Include /etc/csf/csf.alsoallow Note: None of the csf commands for adding or removing IP addresses from csf.allow or csf.deny work on included files, they are treated as read-only. 5. lfd Command Line Options ########################### lfd doesn't have any command line options of its own but is controlled through the init script /etc/init.d/lfd which stops and starts the daemon. It is configured using the /etc/csf/csf.conf file. The best way to see what lfd is up to is to take a look in /var/log/lfd.log where its activities are logged. The various email alert templates follow, care should be taken if you modify that file to maintain the correct format: /etc/csf/accounttracking.txt - for account tracking alert emails /etc/csf/alert.txt - for port blocking emails /etc/csf/connectiontracking.txt - for connection tracking emails /etc/csf/cpanelalert.txt - for WHM/cPanel account access emails /etc/csf/exploitalert.txt - for system exploit alert emails /etc/csf/filealert.txt - for suspicious file alert emails /etc/csf/integrityalert.txt - for system integrity alert emails /etc/csf/loadalert.txt - for high load average alert emails /etc/csf/logalert.txt - for log scanner report emails /etc/csf/logfloodalert.txt - for log file flooding alert emails /etc/csf/netblock.txt - for netblock alert emails /etc/csf/permblock.txt - for temporary to permanent block alert emails /etc/csf/portknocking.txt - for Port Knocking alert emails /etc/csf/portscan.txt - for port scan tracking alert emails /etc/csf/processtracking.txt - for process tracking alert emails /etc/csf/queuealert.txt - for email queue alert emails /etc/csf/relayalert.txt - for email relay alert emails /etc/csf/resalert.txt - for process resource alert emails /etc/csf/scriptalert.txt - for script alert emails /etc/csf/sshalert.txt - for SSH login emails /etc/csf/sualert.txt - for SU alert emails /etc/csf/tracking.txt - for POP3/IMAP blocking emails /etc/csf/uialert.txt - for UI alert emails /etc/csf/usertracking.txt - for user process tracking alert emails /etc/csf/watchalert.txt - for watched file and directory change alert emails 6. Login Tracking ################# Login tracking is an extension of lfd, it keeps track of POP3 and IMAP logins and limits them to X connections per hour per account per IP address. It uses iptables to block offenders to the appropriate protocol port only and flushes them every hour and starts counting logins afresh. All of these blocks are temporary and can be cleared manually by restarting csf. There are two settings, one of POP3 and one for IMAP logins. It's generally not a good idea to track IMAP logins as many clients login each time to perform a protocol transaction (there's no need for them to repeatedly login, but you can't avoid bad client programming!). So, if you do have a need to have some limit to IMAP logins, it is probably best to set the login limit quite high. If you want to know when lfd temporarily blocks an IP address you can enable the email tracking alerts option (which is on by default) You can also add your own login failure tracking using regular expression matching. Please read /etc/csf/regex.custom.pm for more information Important Note: To enable successful SSHD login tracking you should ensure that UseDNS in /etc/ssh/sshd_config is disabled by using: UseDNS no and that sshd has then been restarted. 7. Script Email Alerts ###################### (cPanel installations of csf only) lfd can scan for emails being sent through exim from scripts on the server. To use this feature you must add an extended email logging line to WHM > Exim Configuration Editor > Switch to Advanced Mode > in the first textbox add the following line: log_selector = +arguments +subject +received_recipients If you already already use extended exim logging, then you need to either include +arguments or use +all This setting will then send an alert email if more than LF_SCRIPT_LIMIT lines appear with the same cwd= path in them within an hour. This can be useful in identifying spamming scripts on a server, especially PHP scripts running under the nobody account. The email that is sent includes the exim log lines and also attempts to find scripts that send email in the path that may be the culprit. This option uses the /etc/csf/scriptalert text file for alert emails. If you enable the option LF_SCRIPT_ALERT then lfd will disable the path using chattr +i and chmod 000 so that the user cannot re-enable it. The alert email also then includes the commands needed to re-enable the offending path. Any false-positives can be added to /etc/csf/csf.signore and lfd will then ignore those listed scripts. 8. Process Tracking ################### This option enables tracking of user and nobody processes and examines them for suspicious executables or open network ports. Its purpose is to identify potential exploit processes that are running on the server, even if they are obfuscated to appear as system services. If a suspicious process is found an alert email is sent with relevant information. It is then the responsibility of the recipient to investigate the process further as the script takes no further action. Processes (PIDs) are only reported once unless lfd is restarted. There is an ignore file /etc/csf/csf.pignore which can be used to whitelist either usernames or full paths to binaries. Care should be taken with ignoring users or files so that you don't force false-negatives. You must use the following format: exe:/full/path/to/file user:username cmd:command line The command line as reported in /proc has the trailing null character removed and all other occurrences replaced with a space. So, the line you specify in the file should have space separators for the command line arguments, not null characters. It is strongly recommended that you use command line ignores very carefully as any process can change what is reported to the OS. Don't list the paths to perl or php as this will prevent detection of suspicious web scripts. For more information on the difference between executable and command line, you should read and understand how the linux /proc pseudo-filesystem works: man proc man lsof It is beyond the scope of this application to explain how to investigate processes in the linux /proc architecture. The email alerts are sent using the processtracking.txt email template. It should be noted that this feature will not pickup a root compromise as root processes are ignored - you should use established IDS tools for such security considerations. *** NOTE *** You _will_ get false-positives with this particular feature. The reason for the feature is to bring to your attention processes that have either been running for a long time under a user account, or that have ports open outside of your server. You should satisfy yourself that they are indeed false- positives before either ignoring them or trapping them in the csf.pignore file. We've done our best to minimise false-positives, but there's a balance between being cautious and the sensitivity needed to pick up exploits. The script itself cannot distinguish between malicious intent and intended script function - that's your job as the server administrator ;-) The setting PT_SKIP_HTTP does reduce the number of false-positives by not checking scripts running directly or through CGI in Apache. However, disabling this setting will make a more thorough job of detecting active exploits of all varieties. Another alternative might be to disable PT_SKIP_HTTP and increase PT_LIMIT to avoid picking up web scripts, however this means that real exploits will run for longer before they're picked up. You can, of course, turn the feature off too - if you really want to. 9. Directory Watching ##################### Directory Watching enables lfd to check /tmp and /dev/shm and other pertinent directories for suspicious files, i.e. script exploits. If a suspicious file is found an email alert is sent using the template filealert.txt. NOTE: Only one alert per file is sent until lfd is restarted, so if you remove a suspicious file, remember to restart lfd To remove any suspicious files found during directory watching, enable corresponding setting the suspicious files will be appended to a tarball in /etc/csf/suspicious.tar and deleted from their original location. Symlinks are simply removed. If you want to extract the tarball to your current location, use: tar -xpf /etc/csf/suspicious.tar This will preserver the path and permissions of the original file. Any false-positives can be added to /etc/csf/csf.fignore and lfd will then ignore those listed files and directories. Within csf.fignore is a list of files that lfd directory watching will ignore. You must specify the full path to the file You can also use perl regular expression pattern matching, for example: /tmp/clamav.* /tmp/.*\.wrk Remember that you will need to escape special characters (precede them with a backslash) such as \. \? Pattern matching will only occur with strings containing an asterisk (*), otherwise full file path matching will be applied You can also add entries to ignore files owner by a particular user by preceding it with user:, for example: user:bob Note: files owned by root are ignored For information on perl regular expressions: http://www.perl.com/doc/manual/html/pod/perlre.html The second aspect of Directory Watching is enabled with LF_DIRWATCH_FILE. This option allows you to have lfd watch a particular file or directory for changes and should they change and email alert using watchalert.txt is sent. It uses a simple md5sum match from the output of "ls -laAR" on the entry and so will traverse directories if specified. 10. Advanced Allow/Deny Filters ############################### In /etc/csf.allow and /etc/csf.deny you can add more complex port and ip filters using the following format (you must specify a port AND an IP address): tcp/udp|in/out|s/d=port|s/d=ip|u=uid Broken down: tcp/udp : EITHER tcp OR udp OR icmp protocol in/out : EITHER incoming OR outgoing connections s/d=port : EITHER source OR destination port number (or ICMP type) (use a _ for a port range, e.g. 2000_3000) s/d=ip : EITHER source OR destination IP address u/g=UID : EITHER UID or GID of source packet, implies outgoing connections, s/d=IP value is ignored Note: ICMP filtering uses the "port" for s/d=port to set the ICMP type. Whether you use s or d is not relevant as either simply uses the iptables --icmp-type option. Use "iptables -p icmp -h" for a list of valid ICMP types. Only one type per filter is supported Examples: # TCP connections inbound to port 3306 from IP 11.22.33.44 tcp|in|d=3306|s=11.22.33.44 # TCP connections outbound to port 22 on IP 11.22.33.44 tcp|out|d=22|d=11.22.33.44 Note| If omitted, the default protocol is set to "tcp", the default connection direction is set to "in", so| # TCP connections inbound to port 22 from IP 44.33.22.11 d=22|s=44.33.22.11 # TCP connections outbound to port 80 from UID 99 tcp|out|d=80||u=99 # ICMP connections inbound for type ping from 44.33.22.11 icmp|in|d=ping|s=44.33.22.11 # TCP connections inbound to port 22 from Dynamic DNS address # www.configserver.com (for use in csf.dyndns only) tcp|in|d=22|s=www.configserver.com 11. Multiple Ethernet Devices ############################# If you have multiple ethernet NICs that you want to apply all rules to, then you can set ETH_DEVICE to the interface name immediately followed by a plus sign. For example, eth+ will apply all iptables rules to eth0, eth1, etc. That said, if you leave ETH_DEVICE blank all rules will be applied to all ethernet devices equally. 12. Installation on a Generic Linux Server ########################################## csf+lfd can be configured to run on a generic Linux server. There are some changes to the features available: 1. The default port range is for a typical non-cPanel web server and may need altering to suit the servers environment 2. The Process Tracking ignore file may need expanding in /etc/csf/csf.pignore to suit the server environment 3. A standard Webmin Module to configure csf is included - see the install.txt for more information The codebase is the same for a all installations, the csf.conf file simply has the cPanel specific options removed and the GENERIC option added 13. A note about FTP Connection Issues ###################################### It is important when using an SPI firewall to ensure FTP client applications are configured to use Passive (PASV) mode connections to the server. On servers running Monolithic kernels (e.g. VPS Virtuozzo/OpenVZ and custom built kernels) ip_conntrack and ip_conntrack_ftp iptables kernel modules may not be available or fully functional. If this happens, FTP passive mode (PASV) won't work. In such circumstances you will have to open a hole in your firewall and configure the FTP server to use that same hole. For example, with pure-ftpd you could add the port range 30000:35000 to TCP_IN and add the following line to /etc/pure-ftpd.conf and then restart pure-ftpd: PassivePortRange 30000 35000 For example, with proftpd you could add the port range 30000:35000 to TCP_IN and add the following line to /etc/proftpd.conf and then restart proftpd: PassivePorts 30000 35000 FTP over SSL/TLS will usually fail when using an SPI firewall. This is because of the way the FTP protocol established a connection between client and server. iptables fails to establish a related connection when using FTP over SSL because the FTP control connection is encrypted and so cannot track the relationship between the connection and the allocation of an ephemeral port. If you need to use FTP over SSL, you will have to open up a passive port block in both csf and your FTP server configuration (see above). Perversely, this makes your firewall less secure, while trying to make FTP connections more secure. 14. Messenger Service ##################### This feature allows the display of a message to a blocked connecting IP address to inform the user that they are blocked in the firewall. This can help when users get themselves blocked, e.g. due to multiple login failures. The service is provided by two daemons running on ports providing either an HTML or TEXT message. This services uses the iptables nat table and the associated PREROUTING chain. The ipt_REDIRECT module is used to redirect the incoming port to the relevant messenger service server port. Temporary and/or permanent (csf.deny) IP addresses can be serviced by this feature. It does NOT include redirection of any GLOBAL or BLOCK deny lists. It does require the IO::Socket::INET perl module. It does NOT work on servers that do not have the iptables module ipt_REDIRECT loaded. Typically, this will be with Monolithic kernels. VPS server admins should check with their VPS host provider that the iptables module is included. If you change any of the files in /etc/csf/messenger/ you must restart lfd as they are all cached in memory. HTML Messenger Server ===================== The HTML message that is displayed is provided by the file: /etc/csf/messenger/index.html The HTML server providing this page is very rudimentary but will accept the use of linked images that are stored in the /etc/csf/messenger/ directory. The images must be of either jpg, gif or png format. These images are loaded into memory so you should keep the number and size to a minimum. No other linked resource files are supported (e.g. .css, .js). As the HTML server requires interaction with the client, there is a timer on the connection to prevent port hogging. The server has a built-in function that will replace the text [IPADDRESS] in index.html with the IP address that is blocked by the firewall. This will help the blocked user know what their blocked IP address is. You can also use the text [HOSTAME] which will be replaced by the servers FQDN hostname. The HTML server does not support SSL connections, so redirecting port 443 will not work. The HTML server port should not be added to the TCP_IN list. There is a maximum of 15 port allowed in MESSENGER_HTML_IN. TEXT Messenger Server ===================== The TEXT message that is displayed is provided by the file: /etc/csf/messenger.text This file should only contain text. The TEXT server providing this file simply sends the contents to the connecting port and no protocol exchange takes place. this means that it may not be suitable for use with protocols such as POP3. The server has a built-in function that will replace the text [IPADDRESS] in index.text with the IP address that is blocked by the firewall. This will help the blocked user know what their blocked IP address is. You can also use the text [HOSTAME] which will be replaced by the servers FQDN hostname. The TEXT server does not support SSL connections, so redirecting port 995 will not work. The TEXT server port should not be added to the TCP_IN list. There is a maximum of 15 port allowed in MESSENGER_TEXT_IN. Messenger User ============== You should create a unique user that the messenger services will run under. This user should be disabled and have no shell access. For example, you can create such an account (in this example called "csf") from the root shell using: useradd csf -s /bin/false 15. Block Reporting ################### lfd can run an external script when it performs and IP address block following for example a login failure. This is done by setting the configuration variable BLOCK_REPORT to a script that must be executable. The following parameters are passed the the script as arguments: ARG 1 = IP Address # The IP address or CIDR being blocked ARG 2 = ports # Port, comma separated list or * for all ports ARG 3 = permanent # 0=temporary block, 1=permanent block ARG 4 = inout # Direction of block: in, out or inout ARG 5 = timeout # If a temporary block, TTL in seconds, otherwise 0 ARG 6 = message # Message containing reason for block ARG 7 = logs # The logs lines that triggered the block (will contain # line feeds between each log line) ARG 8 = trigger # The configuration settings triggered lfd launches the BLOCK_REPORT in a forked process which terminates after 10 seconds if not completed by then. It runs under the root account, so great care should be exercised with regard to security of the BLOCK_REPORT script. 16. Port Flood Protection ######################### This option configures iptables to offer protection from DOS attacks against specific ports. This option limits the number of connections per time interval that new connections can be made to specific ports. This feature does not work on servers that do not have the iptables module ipt_recent loaded. Typically, this will be with Monolithic kernels. VPS server admins should check with their VPS host provider that the iptables module is included. By default ipt_recent tracks only the last 100 IP addresses. The tracked IP addresses can be viewed in /proc/net/ipt_recent/* where the port number is the filename. Syntax for the PORTFLOOD setting: PORTFLOOD is a comma separated list of: port;protocol;hit count*;interval seconds So, a setting of PORTFLOOD = "22;tcp;5;300,80;tcp;20;5" means: 1. If more than 5 connections to tcp port 22 within 300 seconds, then block that IP address from port 22 for at least 300 seconds after the last packet is seen, i.e. there must be a "quiet" period of 300 seconds before the block is lifted 2. If more than 20 connections to tcp port 80 within 5 seconds, then block that IP address from port 80 for at least 5 seconds after the last packet is seen, i.e. there must be a "quiet" period of 5 seconds before the block is lifted More information about the ipt_recent module can be found in the iptables man page and at http://snowman.net/projects/ipt_recent/ Note: Blocked IP addresses do not appear in any of the iptables chains when using this module. You must manipulate the /proc/net/ipt_recent/* files as per the module documentation to view and remove IP addresses that are currently blocked if the blocks have not yet expired. Restarting csf resets the ipt_recent tables and removes all of its blocks. Note: There are some restrictions when using ipt_recent: 1. By default it only tracks 100 addresses per table (we try and increase this to 1000 via modprobe) 2. By default it only counts 20 packets per address remembered *This means that you need to keep the hit count to below 20. 17. External Pre- and Post- Scripts ################################### External commands (e.g. iptables rules not covered by csf) can be run before and/or after csf sets up the iptables chains and rules. 1. To run external commands before csf configures iptables create the file: /etc/csf/csfpre.sh Set that file as executable and add an appropriate shebang interpreter line and then whatever external commands you wish to execute. For example: #!/bin/sh /some/path/to/binary -a -b -c etc Then chmod +x /etc/csf/csfpre.sh 2. To run external commands after csf configures iptables create the file: /etc/csf/csfpost.sh Set that file as executable and add an appropriate shebang interpreter line and then whatever external commands you wish to execute. 18. lfd Clustering ################## This new set of options (CLUSTER*) in csf.conf allows the configuration of an lfd cluster environment where a group of servers can share blocks and, via the CLI, configuration option changes, allows and removes In the configuration there are two comma separated lists of IP addresses: CLUSTER_SENDTO = "" CLUSTER_RECVFROM = "" If you want all members of the lfd cluster to send block notifications to each other then both settings should be them same. You also need to enable CLUSTER_BLOCK (enabled by default) for lfd to automatically send blocks to all members in CLUSTER_SENDTO. However, you can also set up a cluster such that some members only provide notifications to others and do not accept blocks from others. For example, you may have a cluster of servers that includes one that hosts a support desk that you do not want to block clients from accessing. In such an example you might want to exclude the support desk server from the CLUSTER RECVFROM list, but include it in the CLUSTER_SENDTO list. The option CLUSTER_MASTER is the IP address of the master node in the cluster allowed to send CLUSTER_CONFIG changes to servers listed in the local CLUSTER_SENDTO list. Only cluster members that have CLUSTER_MASTER set to this IP address will accept CLUSTER_CONFIG changes. There is another option, CLUSTER_NAT that should be used if the IP address of the server does not appear in ifconfig, for example if it is a NAT configuration. If this is the case, add the IP address of the server that this configuration is on and used in CLUSTER_SENDTO/CLUSTER_RECVFROM to CLUSTER_NAT. CLUSTER_LOCALADDR can be set if you do not want to use the servers main IP, i.e. the first one listed via 0.0.0.0. The CLUSTER_PORT must be set to the same port on all servers. The port should NOT be opened in TCP_IN or TCP_OUT as csf will automatically add appropriate in and out bound rules to allow communication between cluster members. The CLUSTER_KEY is a secret key used to encrypt cluster communications using the Blowfish algorithm. It should be between 8 and 56 characters long, longer is better, and must be the same on all members of the cluster. This key must be kept secret! When blocks are sent around the cluster they will maintain their originals parameters, e.g. permanent/temporary, direction (in/out), ports, etc. All blocks are traded except for LT_POP3D and LT_IMAPD. The cluster uses 10 second timeouts in its communications, if the timeout is reached then that cluster members notification will be lost. lfd Cluster CLI and UI ====================== See csf --help for the list of new CLI commands. Additional options will automatically become available in the UI once CLUSTER_SENDTO has been configured. Only cluster members listed in CLUSTER_RECVFROM can send out requests to those members listed in CLUSTER_SENDTO. Only the server listed in CLUSTER_MASTER will be accepted as the source of CLUSTER_CONFIG configuration option requests, such as: --cconfig, --cfile, --crestart The CLI options --cfile and --cfiler allow you to synchronise csf configuration files throughout a cluster from the CLUSTER_MASTER server. There is currently only provision for permanent simple IP denies and allows from the CLI (i.e. not Allow/Deny Filters). The cluster PING sends a ping to each CLUSTER_SENDTO member which will report the request in their respective lfd.log files. This is intended as a test to confirm that cluster communications are functioning. The options to change the configuration option in csf.conf in cluster members should be used with caution to ensure that member specific options are not overwritten. The intention of the two options is that the --cconfig option be used if multiple changes are required and the final request is a --cconfigr to restart csf and lfd to effect the requested changes immediately. A Note on lfd Cluster Security ============================== The clustering option is undoubtedly powerful in allowing servers to pre-emptively block access attempts as one server is hit before the attack can spread to other members of the cluster. This communication, however, does introduce a security risk. Since communications are made over the network, they are open to interception. Also, there is nothing to stop any local user from accessing the network port and sending data to it, though it will be discarded unless properly encrypted[*]. There are security measures implemented to help mitigate attacks: 1. csf constructs iptables rules such that only cluster members can communicate over the cluster port with each other 2. The clustered servers will only accept data from connections from IPs listed in CLUSTER_RECVFROM or CLUSTER_MASTER 3. [*]All communications are encrypted using the Blowfish symmetric block cipher through a Pure Perl cpan module using the Cipher Block Chaining module and the configured CLUSTER_KEY 4. CLUSTER_CONFIG set to 0 prevents the processing of configuration option requests 5. Only CLUSTER_MASTER will be accepted as the source of CLUSTER_CONFIG configuration option requests Should the configured secret key (passphrase) be compromised or guessed or a flaw found in the encryption modules or their implementation in csf, a malicious connection could reconfigure the csf firewall and then leverage a local or remote root escalation. This should be considered if you decide to use this option. THERE ARE NO GUARANTEES OR WARRANTIES PROVIDED THAT THIS FACILITY IS SECURE AND ANY DAMAGE ARISING FROM THE EXPLOITATION OF THIS OPTION IS ENTIRELY AT YOUR OWN RISK. 19. Watching IP Addresses ######################### The CLI option csf --watch [ip] (csf -w [ip]) and configuration option WATCH_MODE logs TCP connection initiation (SYN) packets from a specified source as they traverse the iptables chains. This can be extremely useful in tracking where that IP address is being DROPed or ACCEPTed by iptables. WATCH_MODE should be used when watching IP addresses, although the csf -w [ip] option will still work without it but won't necessarily provide conclusive information on the final destination of the packet. WATCH_MODE is disabled by default and should be left as such unless actively watching an IP address as it will add an overhead to all accepted iptables traffic and increase overall iptables kernel logging through syslog. WATCH_MODE disables: DROP_NOLOG, PS_INTERVAL, DROP_ONLYRES WATCH_MODE enabled: DROP_LOGGING, DROP_IP_LOGGING, DROP_PF_LOGGING WATCH_MODE also logs iptables ACCEPT for watched IP addresses You should only watch a very small number of IP addresses at a time and for a very short period of time, otherwise the kernel log (usually /var/log/messages) will become flooded with entries. Also, any IP address rules added during the time of the watch will not necessarily be included in the logging rules for the watched IP addresses. IP address watches do not survive a csf (iptables) restart. You can use either an IP address or a CIDR address for csf -w [ip]. Recommended method to use this function: 1. Enable WATCH_MODE 2. Restart csf 3. Restart lfd 4. Use the following to watch an IP: csf -w 11.22.33.44 5. Watch the kernel iptables log for hits from the watched IP address Once you have finished watching an IP address you should: 1. Disable WATCH_MODE 2. Restart csf (which will also remove the watched ip rules) 3. Restart lfd The kernel iptables log lines for watching an IP (usually in /var/log/messages) contain the direction of the packet in the chain and the chain name, e.g. I:INPUT is Incoming to the chain INPUT, O:LOCALINPUT is Outgoing from chain LOCALINPUT. The following is a trimmed down example log watch of 192.168.254.4 connecting to port 22: Firewall: I:INPUT SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:LOCALINPUT SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:GDENYIN SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: O:GDENYIN SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:DSHIELD SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: O:DSHIELD SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:SPAMHAUS SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: O:SPAMHAUS SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: O:LOCALINPUT SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:INVALID SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: O:INVALID SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 Firewall: I:LOGACCEPT SRC=192.168.254.4 DST=192.168.254.71 PROTO=TCP DPT=22 20. Port Knocking ################# This option configures iptables to offer port knocking to open sensitive ports based on a sequence of knocked ports for the connecting IP address. For mor information on the idea of port knocking see: http://www.portknocking.org/ The feature requires that you list a random selection of unused ports (at least 3) with a timeout. The ports you choose must not be in use and not appear in TCP_IN (UDP_IN for udp packets). The port to be opened must also not appear in TCP_IN (UDP_IN for udp packets). This feature does not work on servers that do not have the iptables module ipt_recent loaded. Typically, this will be with Monolithic kernels. VPS server admins should check with their VPS host provider that the iptables module is included. By default ipt_recent tracks only the last 100 IP addresses. The tracked IP addresses can be viewed in /proc/net/ipt_recent/* Syntax for the PORTKNOCKING setting: PORTKNOCKING is a comma separated list of: openport;protocol;timeout;kport1;kport2;kport3[...;kportN] So, a setting of PORTKNOCKING = "22;TCP;20;100;200;300;400" means: Open Port 22 TCP for 20 seconds to the connecting IP address to new connections once ports 100, 200, 300 and 400 have been accessed (i.e. knocked with a SYN packet) each knock being less than 20 seconds apart. Access to port 22 remains active after 20 seconds until the connection is dropped, however new connections will not be allowed. More information about the ipt_recent module can be found in the iptables man page and at http://snowman.net/projects/ipt_recent/ Note: IP addresses do not appear in any of the iptables chains when using this module. You must view the /proc/net/ipt_recent/* files as per the module documentation to view IP addresses in the various stages of the knock. Restarting csf resets the ipt_recent tables and removes all of the knocks. 21. Connection Limit Protection ############################### This option configures iptables to offer protection from DOS attacks against specific ports. It can also be used as a way to simply limit resource usage by IP address to specific server services. This option limits the number of new concurrent connections per IP address that can be made to specific ports. This feature does not work on servers that do not have the iptables module xt_connlimit loaded. Typically, this will be with Monolithic kernels. VPS server admins should check with their VPS host provider that the iptables module is included. Also, although included in some older versions or RedHat/CentOS, it was only actually available from v5.3+ The protection can only be applied to the TCP protocol. Syntax for the CONNLIMIT setting: CONNLIMIT is a comma separated list of: port;limit So, a setting of CONNLIMIT = "22;5,80;20" means: 1. Only allow up to 5 concurrent new connections to port 22 per IP address 2. Only allow up to 20 concurrent new connections to port 80 per IP address Note: Existing connections are not included in the count, only new SYN packets, i.e. new connections Note: Run /etc/csf/csftest.pl to check whether this option will function on the server 22. Port/IP address Redirection ############################### This feature uses the file /etc/csf/csf.redirect which is a list of port and/or IP address assignments to direct traffic to alternative ports/IP addresses. Requirements: nat tables ipt_DNAT iptables module ipt_SNAT iptables module ipt_REDIRECT iptables module The following are the allowed redirection formats DNAT (redirect from one IP address to a different one): IPx|*|IPy|*|tcp/udp - To IPx redirects to IPy IPx|portA|IPy|portB|tcp/udp - To IPx to portA redirects to IPy portB DNAT examples: 192.168.254.62|*|10.0.0.1|*|tcp 192.168.254.62|666|10.0.0.1|25|tcp REDIRECT (redirect from port to a different one): IPx|portA|*|portB|tcp/udp - To IPx to portA redirects to portB *|portA|*|portB|tcp/udp - To portA redirects to portB REDIRECT examples: *|666|*|25|tcp 192.168.254.60|666|*|25|tcp 192.168.254.4|666|*|25|tcp Where a port is specified it cannot be a range, only a single port. All redirections to another IP address will always appear on the destination server with the source of this server, not the originating IP address. This feature is not intended to be used for routing, NAT, VPN, etc tasks Note: /proc/sys/net/ipv4/ip_forward must be set to 1 for DNAT connections to work. csf will set this where it can, but if the kernel value cannot be set then the DNAT redirection many not work. 23. Integrated User Interface Feature ##################################### Integrated User Interface. This feature provides a HTML UI to the features of csf and lfd, without requiring a control panel or web server. The UI runs as a sub process to the lfd daemon. As it runs under the root account and successful login provides root access to the server, great care should be taken when configuring and using this feature. There are additional restrictions to enhance secure access to the UI: 1. An SSL connection is required 2. Separate ban and allow files are provided to only allow access to listed IP addresses 3. Local IP addresses cannot connect to the UI (i.e. all IP addresses configured on the server NICs) 4. Unique sessions, session timeouts, session cookies and browser headers are used to identify and restrict active sessions Requirements: 1. openssl 2. Perl modules: Net::SSLeay, IO::Socket::SSL and dependent modules 4. SSL keys 5. Entries in /etc/csf/ui/ui.allow The SSL server uses the following files: SSL Key goes into /etc/csf/ui/server.key SSL Certificate goes into /etc/csf/ui/server.crt Preferably, real CA signed certificates should be used. You can use an existing domain and cert for accessing the UI by populating the two files mentioned. If the cert has a ca bundle, it should be appended to the server.crt file. lfd must be restarted after making any changes: http://httpd.apache.org/docs/2.2/ssl/ssl_faq.html#realcert Alternatively, you could generate your own self-signed certificate: http://httpd.apache.org/docs/2.2/ssl/ssl_faq.html#selfcert Any keys used must have their pass-phrase removed: http://httpd.apache.org/docs/2.2/ssl/ssl_faq.html#removepassphrase The login URL should use the domain you have listed in the self-signed cert: https://: For example: https://www.somedomain.com:6666 Your browser must accept session cookies to gain access. UI_ALLOW is enabled by default, so IP addresses (or CIDRs) allowed to use this UI must be listed in /etc/csf/ui/ui.allow before trying to connect to the UI. Only IP addresses can be listed/used in /etc/csf/ui/ui.ban - this file should only be used by the UI to prevent login. Use csf blocks to prevent access to the configured port and only use Advanced Allow/Deny Filters for access, i.e. do not list the port in TCP_IN. Logging for UI events are logged to the lfd /var/log/lfd.log file. Check this file if you are unable to access the UI. Required Perl Modules: For example, on Debian v6 the perl modules can be installed using: apt-get install libio-socket-ssl-perl libcrypt-ssleay-perl \ libnet-libidn-perl libio-socket-inet6-perl libsocket6-perl For example, on CentOS v6 the perl modules can be installed using: yum install perl-IO-Socket-SSL.noarch perl-Net-SSLeay perl-Net-LibIDN \ perl-IO-Socket-INET6 perl-Socket6