We're a small company but none the less are inundated with firewall logs reporting numerous attempts to find holes in our network; c'est la vie. Seeing as how we are small, we don't have the resources to go through and send emails off to the abuse departments of each network sourcing the probes. Question is: Has there been development of some sort of intelligent unix land app that can understand Cisco syslog output, find the abuse departments of the sourcing networks and send them off a nice little FYI?
try LogDog to act on the syslog data...it sends all syslog log files through a
pipe and scans for specific data...then you can email the complete message to
anyone. It can have a negative performance impact depending on the number of
sustained syslog logs being generated....but I used it on a system receiving
syslog logs from over 200 routers and didn't see any issues. Of course
syslog-ng can also do this....but I found logdog easier to implement. Not
sure how you can automate the abuse email address?? You can specify a perl
script from within the logdog conf file that could do a dig on the ip address
from the source address...but that's just me thinking out loud. I think
you'll find many programs out there that can do this...both commercial and
opensource...but you'll need to do some customization.
if you automate abuse reporting you can basically assume that the reciver
will automate abuse handling. since that has in fact happened as far as i
can tell the probably of you automated asbuse replaies ever reaching a
human who cares or can do something about it is effecetivly zero.
Jason Lixfeld wrote:
...Has there been development of some
sort of intelligent unix land app that can understand Cisco syslog
output, find the abuse departments of the sourcing networks and send
them off a nice little FYI?
With rare exceptions, I'd say don't bother, even if you do come up with
such a thing. I've actually sent off two in the past week, which is my
normal total for the month (any month). One was to a machine that was
agressively testing identd (and starting to annoy me) on every machine in
my netblock (it's little, but it's mine).
The other was more interesting. A tool that had been used to attack imap
servers earlier this year has apparently been modified to hit FTP instead.
The common bond is the user name "lizdy", which is only one of the multiple
of names attempted. If you're curious, hit google with the words (lizdy
ftp), and you'll come up with a few machines already hit by it. One of the
machines that hit was an NT machine in a block that had an actual abuse
dept, and I thought the owner would probably want to know. I got a nice
response back, and I'd bet that it was probably taken care of. The others
were also owned, but out of networks where I know that they just won't
care. Pity there's no way to let the owner of the machine know, but that's
A "nice little FYI" will just be adding to the brownian motion of the
internet as we know it today. On those rare cases where you have the time,
and are sure of the target, of course, send something off. Just please
don't automate it.
Oh, and I no longer have an internet facing FTP server (that tool hits
about 200-400 times in less than 5 seconds...really abusive).
When we get something that looks automated, we send back a reply saying
"We received this, if you'd like us to take action, please have a human
I've been thinking of instead having them send us a cryptographic hash of
their message, saying that we MUST have all such notifications validated.
I'd give them the URL to some page that would provide the hash, of course.
Most likely, automated abuse reports will be treated like abuse reports from
users with those lovely software firewalls that whine all the time that their
ISP's nameserver is trying to hack them on port 53 (IE: thrown in with the
rest of the reports in the round filing cabinet on the floor next to the
I refused to accept automated abuse reports of probes or similar when I was an
Portscans/pingscans/etc are not illegal (and I've seen this sucessfully proven
in court at least once). They are illegal if you use it to bring down
someone's machine though.
Basically, if I were you, I'd turn your firewall's sensitivity WAY down and
only track events that are obviously attempts to hack.
It's difficult to sort out legitimate complaints for port scanning.
Consider that the vast majority of such complaints a provider receieves,
particularly automated ones (groan), are just flat out wrong or stupid (or
For example: "Your web server is hacking my web browser on port 80", or
"Why are you probing me with UDP packets on port 53 from this host named
NS1...", but usually stated with far more capital letters, misspellings,
profanity, and threats to sue or report your web server to the
authorities because it dared to respond to their port 80 connection.
Things only seem to get worse when you actually try to have a halfass team
of people respond to these. Usually the victim is someone who gets a syn
flood from random sourced addresses, correctly responds with RSTs, and
ends up being accused of port scanning due to the backscatter hitting some
random military IP address. Anyone with a reasonable amount of experience
should be able to look at any of the detailed packet logs and clearly see
the very obvious patterns which indicate the differences between
legitimate port scans, backscatter, or classic spoofed source syn floods.
But they never do, even when they claim to be highly experienced and in
positions of power. For many providers, getting a threatening e-mail from
a government agency will result in someone being turned off, even if they
have done nothing wrong.
Recently I saw someone running an online gaming service who experienced
this in the other direction. The attacker set his IP as the source, and
directly fired off millions of packets to random destinations. Not only
was their a direct DoS effect due to all the RST coming in, but over the
course of 48 hours he received THOUSANDS of angry calls, many complaints
to his provider, and even several death threats.
I have, according to my ids around 400pps arriving at my home network that
don't belong there. if I payed attention to all of it I'd be busy, if I
generated abuse reports and fired them off it would generate a lot of
noise... random portscans, dos backsplash and worm traffic don't really
rise to the level that would make me want to invest my time in trying to
identify and deal with the sources.
Not wanting to be ripped to shreds here, I think it's still worthwhile
to alert people to, say, Slammer-infected hosts on their networks.
Sure, the good folks are already monitoring their networks for hosts
sourcing things like that, and they're also the ones that will know how
to deal with automated complaints. The people that don't already
monitor their networks will benefit from being alerted.
Take www.dshield.org for instance. They aggregate logs from various sources and
send complaints to the upstream provider. This is something that would work for
Working for an AUP department at an ISP, we gladly accept automated complaints.
Sending the complaint downstream for investigation should be standard procedure.
Taking action against repeated complaints (differing time stampts of course)
after at least one warning should follow.
Forwarding the complaint either by email or by phone to your downstream
shouldn't be considered a problem. Just don't shoot first and ask questions
later. It's a pretty safe bet to say that something is going wrong on a
downstream network if you are getting complaints from multiple sources.
In fact, reactions seem to be split in 3. The angry ones are the ones we get
logs about their PAT address and they freak out because null routing them would
effectively shut down their entire network. The indifferent ones are typically
used to these problems and rectify the problem, case closed. Finally, we
actually get customers giving us kudos because we advised them of a problem on
[Mon, Dec 29, 2003 at 12:59:09PM -0500]
Daniel Medina Inscribed these words...