[VoiceOps] Strange register attack

Hadriel Kaplan HKaplan at acmepacket.com
Sat Nov 27 06:18:19 EST 2010


If you don't mind sending back a failure response, then you can instead add a route for sip:0.0.0.0 which should cause it to fail.  In newer releases there's an explicit "reject" action which lets you just tell it to reject it more directly, as well as what response code to use when rejecting it, and you can still drop that response so it silently discards it.  Using the reject action is the new recommended way (it's simpler, slightly more efficient, and has specific stats/counters and such).

-hadriel

On Nov 26, 2010, at 7:53 PM, Christian Pena wrote:

I did it by creating a bogus session-agent (say with ip 1.1.1.1) and set it in disabled state. Then routed the packets to that session agent by changing the route header to something like sip:1.1.1.1

anorexicpoodle wrote:
Without getting into the gory details, the recommended method seems to be using HMR to add a route header pointing at a bogus session agent which will be obeyed over any local policy, then you create a response map mapping 503's from that agent back to an unused 6xx code, and then finally setting an option in the sip interface to drop the unused 6xx code, and the effective result is a black hole based on anything you can get at via HMR.

On Sat, 2010-11-27 at 09:38 +1030, Peter Childs wrote:

I'm interested in this.   We use some HMRs to do various manipulations, but is there a method to 'drop' traffic matching HMRs, or cause a DOS/ACL block on the source?

________________________________________
From: voiceops-bounces at voiceops.org<mailto:voiceops-bounces at voiceops.org> [voiceops-bounces at voiceops.org<mailto:voiceops-bounces at voiceops.org>] On Behalf Of anorexicpoodle [anorexicpoodle at gmail.com<mailto:anorexicpoodle at gmail.com>]
Sent: Saturday, 27 November 2010 7:15 AM
To: Christian Pena
Cc: voiceops at voiceops.org<mailto:voiceops at voiceops.org>
Subject: Re: [VoiceOps] Strange register attack

I have a fairly substantial HMR ruleset on our access side doing everything from user agent filtering, to collecting custom values for CDR's, and while there is a CPU hit, its not nearly as bad as you might think.

On Fri, 2010-11-26 at 09:26 -0500, Christian Pena wrote:
We have seen similar things on our network. I never setup the HMRs on the access side of the Acme thinking they would cause a substantial increase in CPU usage. Anyone have any luck in doing this on a production network?

Thanks,

Chris


anorexicpoodle wrote:
This is a well-known attack. If you're running an Acme, contact TAC, they have a HMR ruleset that will blackhole this attack based on the user agent, though if your access side is tuned up well enough it should black-hole itself pretty quickly.

On Thu, 2010-11-25 at 23:44 -0600, Lee Riemer wrote:


Yes, this is normal.  Just figure out how to deal with it.

On 11/25/2010 11:03 PM, Darren Schreiber wrote:
> We've been suffering from this, too. The SIP headers are hacked and
> completely bogus. They seem to be from only a few select IPs from us.
>
> I'm about ready to abandon port 5060 :-)
>
> - Darren
>
>
> On 11/25/10 9:03 PM, "Peter Childs"<PChilds at internode.com.au<mailto:PChilds at internode.com.au><mailto:PChilds at internode.com.au>>  wrote:
>
>> sql>  select count(ua) from sip_trace where ua = 'friendly-scanner';
>> COUNT(UA): 22330
>>
>> We get thousands of these scans from all over the joint all the time.
>>
>> That is in the last 8 hours...
>>
>> sql>  select count(fromip), fromip from sip_trace where ua =
>> 'friendly-scanner' group by fromip;
>> COUNT(FROMIP): 3
>> FROMIP       : 124.195.52.250
>>
>> COUNT(FROMIP): 1
>> FROMIP       : 124.254.44.172
>>
>> COUNT(FROMIP): 13127
>> FROMIP       : 202.101.187.66
>>
>> COUNT(FROMIP): 9199
>> FROMIP       : 74.218.78.29
>> (4 rows, 10201 ms)
>>
>>
>> I occasionally have discussions with others about
>> http://tools.ietf.org/html/rfc5635 using some thresholds to block some of
>> these at the border, with the problem being that one day someone will use
>> some cloud platform and we will take out we shouldn't.
>>
>> The ACME SBCs we use seem to eat this stuff up ok, but some of the issues
>> we encounter
>>     1. Customers with SIP CPE where a high volume of SIP trash causes the
>> CPE to lock
>>     2. Customers running Asterisk implementations getting cracked and
>> owned
>>
>> Cheers,
>>   Peter
>>
>> On 26/11/2010, at 1:32 PM, Colin wrote:
>>
>>> Tonight i'm seeing hundreds of register attempts per second to one of
>>> my SBC's from an IP in china 61.142.250.96.
>>>
>>> the From: and to: line is always  one of these 2 below.
>>>
>>> \"118\"<sip:118 at my SBC IP>;    source port  5063
>>> \"qwerty\"<sip:qwerty at my SBC IP>;  source port 5067
>>>
>>>
>>>
>>> user-agent: friendly-scanner is always.
>>>
>>> Looks like sipvicious default user agent. Anyone seen a register flood
>>> like this before?
>>>
>>>
>>>
>>> Colin
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> VoiceOps mailing list
>>> VoiceOps at voiceops.org<mailto:VoiceOps at voiceops.org><mailto:VoiceOps at voiceops.org>
>>> https://puck.nether.net/mailman/listinfo/voiceops
>>
>> _______________________________________________
>> VoiceOps mailing list
>> VoiceOps at voiceops.org<mailto:VoiceOps at voiceops.org><mailto:VoiceOps at voiceops.org>
>> https://puck.nether.net/mailman/listinfo/voiceops
>
> _______________________________________________
> VoiceOps mailing list
> VoiceOps at voiceops.org<mailto:VoiceOps at voiceops.org><mailto:VoiceOps at voiceops.org>
> https://puck.nether.net/mailman/listinfo/voiceops
_______________________________________________
VoiceOps mailing list
VoiceOps at voiceops.org<mailto:VoiceOps at voiceops.org><mailto:VoiceOps at voiceops.org>
https://puck.nether.net/mailman/listinfo/voiceops





_______________________________________________
VoiceOps mailing list
VoiceOps at voiceops.org<mailto:VoiceOps at voiceops.org><mailto:VoiceOps at voiceops.org>
https://puck.nether.net/mailman/listinfo/voiceops




<ATT00001..c>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/voiceops/attachments/20101127/e78688be/attachment.html>


More information about the VoiceOps mailing list