Outlook Web Access Log Analysis
Hello Reader,
In this entry I’d like to discuss log analysis on Outlook Web Access servers. I’ve successfully used OWA log analysis in the past to quickly determine who has been reading mailboxes other than their own. Two pieces of information in the logs that exist by default in the OWA creation process allow this to occur. The first is that OWA uses NTLM authentication for web mail users who log in and the domain and username authenticated is stored in the logs in the cs-username field with format “domain\username”, remember this field will only be populated if the user successfully authenticated otherwise it will be filled with “-“. The second is that the mailbox accessed is stored in the cs-uri-query field in the logs and will look something like “isnewwindow=0&mailbox=username”. By comparing the authenticated NTLM username to the username of the mailbox requested we can write some pretty easy code to determine who has been accessing the mailboxes of other users, or attempting to.
First things first, we need the OWA logs themselves. They should be located in the “%systemroot%\system32\logfiles” directory usually in W3SVC1 if it’s the first default web created. Once we have them we need to either copy or export the log files in that directory from the image. Our first bit of code reads the content of the directory:
Next we want to see if the username they have authenticated with matches the username of the mailbox they have requested. If it does, move on and print a . to the screen so we can see some activity. If it does not print a ! to the screen and write the resulting access to a separate file.
We can also store a unique list of these users in a hash so we can get a list of offenders to review. Additionally you could store all of this in a database table in larger cases so you can begin to run queries in time periods, users affected and start breaking out what messages, attachments, tasks and calendar items have been accessed.
I have posted the raw perl code and the windows compiled executable for this. For the windows executable I made it just search the same directory the executable is in, so just copy it into the directory with the logs and run it. Load up the report.csv file and find out who your suspects are.
Raw Code
Windows Executable
In this entry I’d like to discuss log analysis on Outlook Web Access servers. I’ve successfully used OWA log analysis in the past to quickly determine who has been reading mailboxes other than their own. Two pieces of information in the logs that exist by default in the OWA creation process allow this to occur. The first is that OWA uses NTLM authentication for web mail users who log in and the domain and username authenticated is stored in the logs in the cs-username field with format “domain\username”, remember this field will only be populated if the user successfully authenticated otherwise it will be filled with “-“. The second is that the mailbox accessed is stored in the cs-uri-query field in the logs and will look something like “isnewwindow=0&mailbox=username”. By comparing the authenticated NTLM username to the username of the mailbox requested we can write some pretty easy code to determine who has been accessing the mailboxes of other users, or attempting to.
First things first, we need the OWA logs themselves. They should be located in the “%systemroot%\system32\logfiles” directory usually in W3SVC1 if it’s the first default web created. Once we have them we need to either copy or export the log files in that directory from the image. Our first bit of code reads the content of the directory:
opendir(IMD, $dirtoget) die("Cannot open directory");
@thefiles=
readdir(IMD);
closedir(IMD);
foreach $file
(@thefiles)
{
print "my file: $file\n";
open(FILE, "$file");
}
Next we need to do something with these files. We want to parse each line looking for people accessing mailboxes:while(FILEHANDLE)
{
if ($_ =~ m/^([0-9\-]+
[0-9:]+) ([0-9.]+) ([^ ]+) [^ ]+ [^ ]+ [0-9.]+
[0-9]+ (GETPOST) ([^ ]+)
isnewwindow=0&mailbox=([^ ]+) ([1-3][0-9][0-9]) [0-9] [0-9]+ [0-9]+
[0-
9]+ HTTP.+ [^ ]+ ([^ ]+) ([^ ]+) (.+)$/i)
{
my ($access, $ip, $username, $method, $url, $query,
$status,
$useragent, $cookie, $referer) = ($1, $2, $3, $4, $5, $6, $7, $8,
$9,
$10);
}
Next we want to see if the username they have authenticated with matches the username of the mailbox they have requested. If it does, move on and print a . to the screen so we can see some activity. If it does not print a ! to the screen and write the resulting access to a separate file.
if ($username !~ m/$query/i )
{
print OUTFILE "$access, $ip, $username, $query,
$status\n";
print "!";
}
else
{
print
".";
}
We can also store a unique list of these users in a hash so we can get a list of offenders to review. Additionally you could store all of this in a database table in larger cases so you can begin to run queries in time periods, users affected and start breaking out what messages, attachments, tasks and calendar items have been accessed.
I have posted the raw perl code and the windows compiled executable for this. For the windows executable I made it just search the same directory the executable is in, so just copy it into the directory with the logs and run it. Load up the report.csv file and find out who your suspects are.
Raw Code
Windows Executable
