NAME LaBrea::Tarpit - Utilities and web displays for Tom Liston's LaBrea scanner/worm disruptor See: http://www.hackbusters.net SYNOPSIS use LaBrea::Tarpit qw( [exportable functions] ); or require LaBrea::Tarpit; daemon(%hash or \%hash); $bandwidth = bandwidth(\%tarpit); $midnight = midnight($epoch_time,$tz); $timezone = timezone($now); $sec = $tz2_sec($tz); $time_string = their_date($gmtime,$tz); $rv = restore_tarpit(\%tarpit,path2cache_file); $rv = log2_mem(\%tarpit,log_line,is_daemon,port_intvls,DShield); $rv = process_log(\%tarpit,path2log_file,is_daemon,port_intvls); $rv = cull_threads(\%tarpit,timeout,scanners,port_intvls,DShield); $rv = write_cache_file(\%tarpit,path2cache_file,umask,flag); prep_report(\%tarpit,\%hash); $rv = find_old_threads(\%tarpit,\%report,$age); INSTALL * Package Untar the package perl Makefile.PL make make test make install To use examples/daemon.pl, configure the array at the beginning of the script and set the locations for the cache files. ...typically /var/tmp/labrea.cache and /var/tmp/DShield.cache * Report/examples/html_report.plx * Report/examples/paged_report.plx html_report and paged_report will run as a cgi scripts by simply renaming them xxx_report.cgi. It is highly recommend that you enable the file caching to minimize load on your system. Read the comments in the file itself for configuration. The defaults should work fine, but you must create the temporary directory used for file caching AND it must be writable by the web server. html_report and paged_report are configured to provide other_site reporting. You must set up the cron job maintain the site_stats file for reporting. See webscan.pl below: * Get/examples/web_scan.pl Run web_scan.pl from a cron job hourly or daily to update the statistics from all know sites running LaBrea::Tarpit. A report can then be generated showing the activity worldwide. # MIN HOUR DAY MONTH DAYOFWEEK COMMAND 30 * * * * ./web_scan.pl ./other_sites.txt ./tmp/site_stats Also see: LaBrea::Tarpit::Report::other_sites * examples/tell_me.pl AGE Run tell_me.pl from a cron job daily to send yourself an email detailing teergrubed hosts that have been held longer than AGE days. You might actually want to tell the bad guys that they have a rogue machine. # MIN HOUR DAY MONTH DAYOFWEEK COMMAND 30 * * * * ./tell_me.pl 60 # default * DShield/examples/mail_dshield.pm Configure with your DShield UserID, email address, mail agent and the location of the daemon - DShield cache file, then run periodically from cron to send reports to DShield. DESCRIPTION - LaBrea::Tarpit A comprehensive Hack Attack reporting module when used in conjunction with Tom Liston's LaBrea scanner/worm disruptor. When configured with reporting and stat collection it provides a detailed HTML page containing: * Bandwidth consumed by attack/disruption daemon * Summary of previous 5 days of attack/disruption * All IP addresses currently attacking * IP address, port attacked/held, attack start time * As above, but history of terminated attacks * By day detail graphs on port attack intensity * Active summary of known LaBrea::Tarpit sites For more information on LaBrea see: http://www.hackbusters.net/ or contact the author of LaBrea, Tom Liston the tliston@hackbusters.net entry elsewhere in this document. The parsed output of either syslog data or STDOUT from LaBrea using -o or -O options is readily turned into text reports or an html output page. Basically there are two methods of operation. You can use the daemon mode to create an almost realtime cache that may be parsed using the report routines, or you can use the update and report routines to parse the syslog files on an as needed basis. If you plan to create web page reports, the daemon model will use less system resources in the long run and avoids running syslog with the high volume output of LaBrea. Improvements VERSION 1.00 As of version 1.00, daemon.pl uses network sockets to provide data for the report modules. This means that the daemon can run on a remote machine and the report scripts and web server can be somewhere else. For those of you upgrading from older versions, you MUST upgrade all of your report scripts as well. Older versions use a pipe or FIFO and this is no longer supported as there were problems maintaining separate sessions. * recurse_hash2txt(\$txt_buffer,\%hash,$keys_so_far,flag) Appends to txt_buffer. Generates a text tree of a hash. %hash{lvl1}->{lvl2}->{lvl3} = 5; this real hash flag = 0, ksf = '' with this input lvl1:lvl2:lvl3:5 produces this text flag = 1, ksf = ptr this input ptr->{lvl1}->{lvl2}->{lvl3} = 5; this txt * ($LBfh,$version,$kid) = lbd_open($LaBrea,$DEBUG); Core daemon start routine. Not exported, but can be replaced externally with: *LaBrea::Tarpit::lbd_open = sub { stuff }; Returns the pid of the underlying process (if any) and the version number of that process. It also sets the command line shown by 'ps' like this: $0 = 'stuff'; input: path to daemon, STDERR switch returns: LaBrea file handle, version, pid of kid * lbd_close($LBfh,$kid); Core daemon close routine, not exported but can be replaced externally with: *LaBrea::Tarpit::lbd_close = sub { sutff }; Close the daemon and kill off $kid with sig 15 input: filehandle, pid of kid returns: nothing * daemon(&hash | \%hash) input parameters: from hash or pointer to hash { 'LaBrea' => '/usr/local/bin/LaBrea -z -v -p 1000 -h -i eth0 -b -O 2>&1', # 'd_port' => '8686', # default local comm port 'd_host' => 'localhost', # defaults to ALL interfaces # NOT recommended 'allowed' => 'localhost,remote.com', # default is ALL # recommend only 'localhost' 'pid' => '/path/to/pid/file_name', 'cache' => '/path/to/cache/file', 'DShield' => '/path/to/DShield/out_file', # 'kids' => default 5 # kids to deliver net msgs # why would you need more?? # 'umask' => default 033, # cache_file umask # 'cull' => default 600, # seconds to keep old threads 'scanners' => 100, # keep this many dead threads # 'port_timer' => default 86400, # seconds per collection period 'port_intvls' => 30, # keep #nintvls of port stats # 0 or missing disables # this can take lots of memory # optional exclusion information (required if files exist) 'config' => '/etc/LaBreaConfig', # or # 'config' => 'LaBrea.cfg', # windoze (untested) # or # 'config' => ['/etc/LaBreaExclude','/etc/LaBreaHardExclude'], }; The daemon can be run on a remote host with restricted client access and the data retrieved by another host that has web server capabilities scanners is enabled by setting to a positive number. Since all IP's that are seen but not captured can potentially be saved, this list could grow very large. You can limit the amount of memory used by setting the number of items that can be saved. There is no default, a value <= 0 turns of this feature. Scanners are saved on a fifo basis, when full, the oldest will be deleted first. Signals: HUP cull then write new cache file TERM cull, write cache, exit Killing the daemon with SIG_KILL (-9) will NOT write a new cache file and will leave LaBrea running. YOU SHOULD NOT DO THIS daemon operation: The daemon parses the output of LaBrea in real time and collects the information in its memory cache, periodically pruning away threads that are no longer active to minimize the memory footprint. Upon receiving a HUP, it immediately prunes memory of old threads and writes its cache to file. data retrieval Usage: connect to TCP port 8686 send "standard" (endline) or send "active" (endline) or send "short" (endline) or send "config" (endline) to receive the complete memory cache described above or only active threads or a truncated version suitable for making a short report. config sends the daemon configuration file information to the client. * $bandwidth = bandwidth(\%tarpit); Returns bandwidth reported by LaBrea or zero if the -b option is not used or bw is unknown. * $time = midnight($epoch_time,$tz); Returns epoch time at 00:00:00 of current day from any epoch time submitted. Time zone is calculated (inefficently) each time if omitted. * $seconds = $tz2_sec($tz); Convert time zone into seconds input: timezone i.e. -0800 returns: seconds i.e. -28800 * $time_string = their_date($gmtime,$tz); Returns date string like perl's localtime(time) for the specified time zone * $timezone = timezone($now); Returns the local timezone as a text string i.e. -0800 uses current time if $now is omitted, this is the normal method of usage. * $rv = restore_tarpit(\%tarpit,path2cache_file); Restore the memory cache from the file cache. returns true if successful false if cache_file won't open File Cache is of the form: _VERSION:Package::Name version daemon | static _CACHEDUMP:date of last cache dump # for each src host at:ip_addr:sport:dest at:ip_addr:sport:dport at:ip_addr:sport:ct at:ip_addr:sport:last at:ip_addr:sport:persist at:ip_addr:sport:DShield:dest:dp:time bw:number # for each scanning (gone) host dt:ip_addr:sport:dport dt:ip_addr:sport:last dt:ip_addr:persist ph:time:dport:count now:1234567890 tz:-0800 * array2_tarpit(\%tarpit,\@array); Restore the memory cache from an array of lines as described for restore_tarpit. The lines must already be stripped of return characters Always returns true; * $rv = log2_mem(\%tarpit,log_line,is_daemon,port_intvls,DShield); Update memory cache from log output line. Set is_daemon if the log output is from daemon STDOUT In order of minimum CPU overhead: Daemon mode or logs created from STDOUT require the least cpu overhead. LaBrea -O is more efficient than Labrea -o. Logs from STDOUT are more efficient than syslogs. Standalone syslogs are more efficient than mixed. All log lines used are of the form: epoch time (seconds) or date text followed by [...LaBrea:] # syslog only one of these info text bw: bandwidth (bytes/sec) or info text: src_ip src_port txt dest_ip dest_port Or more succinctly: time text: bandwidth or time text: src_ip src_port txt dest_ip dest_port Returns: true / false on success / fail * $rv = process_log(\%tarpit,path2log_file,is_daemon,port_intvls); Update the memory cache from a file with lines of the form described for log2_mem Set is_daemon if the output log was created from STDOUT of LaBrea or if you can guarantee that there is nothing in the file except LaBrea lines. Returns true on success, false if file fails to open * $rv=cull_threads(\%tarpit,timeout,scanners,port_intvls,DShield); Cull aged threads from memory cache. Default time is 600 seconds (10 min). On startup, no culls are done for the cull time interval to retain old capture time information for any lingering black hats that may fall back into the tarpit. See daemon description for scanners, port_intvls cull_threads updates the time zone of the tarpit cache appends DShield info to file specified in DShield if present returns: true if threads removed, else false * $rv = write_cache_file(\%tarpit,path2cache_file,umask,flag); Write memory cache to file. returns cache text on success, false if file fails to open. umask defaults to 033 if not supplied File Cache is of the form: _VERSION:Package::Name version daemon | static _CACHEDUMP:date of last cache dump see description above in restore_tarpit flag = true, append 'daemon' to version flag = false, append 'static' to version * prep_report(\%tarpit,\%hash); Prepare arrays of report values from the tarpit memory cache. Only the values requested will be filled. %hash values: times in seconds since epoch { # teergrubed hosts 'tg_srcIP' => \@tgsip, # B 'tg_sPORT' => \@tgsp, # B 'tg_dstIP' => \@tgdip, 'tg_dPORT' => \@tgdp, 'tg_captr' => \@tgcap, # capture epoch time 'tg_last' => \@tglst, # last contact 'tg_prst' => \@tgpst, # type / persistent [true|false] # # threads per teergrubed host 'th_srcIP' => \@thsip, # B 'th_numTH' => \@thnum, # number threads this IP # # capture statistics # all fields B 'cs_days' => number of days to show, 'cs_date' => \@csdate, # epoch midnight of capt date 'cs_ctd' => \@csctd, # captured this date # # phantom IP's used (from our IP block) 'ph_dstIP' => \@phdip, # B 'ph_prst' => \@phpst, # type / persistent [true|false] # # scanning hosts lost 'sc_srcIP' => \@scsip, # B 'sc_dPORT' => \@scdp, # attacked port 'sc_prst' => \@scpst, # type / persistent [true|false] 'sc_last' => \@sclst, # last contact # # port statistics # all fields B 'port_intvls' => number of periods to show, 'ports' => \@ports, # scanned port list 'portstats' => \@portstats, # where @portstats = @stats_port1, @stats_port2, etc... # always returned $hash{tz} = timezone, always filled if not present $hash{now} = epoch time of last load from cache $hash{bw} = bandwidth always filled $hash{total_IPs} = total teergrubed hosts $hash{threads} = total # of threads # conditionally returned $hash{LaBrea} = version if known $hash{pt} = port activity collection interval $hash{tg_capt} = active hard captured (need tg_prst) $hash{phantoms} = total phantoms $hash{ph_capt} = phantoms that were hard captures $hash{sc_total} = total dropped scans $hash{sc_capt} = dropped hard capture (need sc_prst) } NOTE: prep_report will fill any subset of the specified or all if they are all specified * $rv = find_old_threads(\%tarpit,\%report,$age); Report only aged threads input: \%tarpit, \%report, age_in_days returns: false = fail, or nothing to report true = number of items and fills \%report %report = ( [text string] [time since epoch] ip.addr:sp -> dp => time captured, .... DEPENDENCIES Net::Whois::IP version 0.35 Net::Netmask version 1.8 or higher LaBrea version 2.4b3 or higher See the INSTALL document for complete information EXPORT None by default. EXPORT_OK recurse_hash2txt daemon bandwidth midnight timezone tz2_sec their_date restore_tarpit log2_mem process_log cull_threads write_cache_file prep_report find_old_threads array2_tarpit COPYRIGHT Copyright 2002, 2003, 2004, Michael Robinton & BizSystems This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. AUTHOR Michael Robinton, michael@bizsystems.com SEE ALSO perl(1), LaBrea::Codes(3), LaBrea::Tarpit::Get(3), LaBrea::Tarpit::Report(3), LaBrea::Tarpit::Util(3), LaBrea::Tarpit::DShield(3)