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<REQUIRED>
'tg_sPORT' => \@tgsp, # B<REQUIRED>
'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<REQUIRED>
'th_numTH' => \@thnum, # number threads this IP
#
# capture statistics # all fields B<REQUIRED>
'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<REQUIRED>
'ph_prst' => \@phpst, # type / persistent [true|false]
#
# scanning hosts lost
'sc_srcIP' => \@scsip, # B<REQUIRED>
'sc_dPORT' => \@scdp, # attacked port
'sc_prst' => \@scpst, # type / persistent [true|false]
'sc_last' => \@sclst, # last contact
#
# port statistics # all fields B<REQUIRED>
'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)