Blocking of domain strings in iptables

Mailman List Mirror (Read Only)
1

Hello everyone

I am trying to figure out the way to drop a domain name DNS resolution
before it hits application server. I do not want to do domain to IP mapping
and block destination IP (and source IP blocking is also not an option).

I can see that a string like this:

iptables -A INPUT -p udp -m udp --dport 53 -m string --string "domain"
--algo kmp --to 65535 -j DROP

this can block "domain" which includes domain.com/domain.net and everything
in that pattern. I tried using hexadecimal string for value like domaincom
(hexa equivalent) and firewall doesn't pics that at all.

The only other option which I found to be working nicely is u32 based
string as something suggested on DNS amplification blog post here -
http://dnsamplificationattacks.blogspot.in/2013/12/domain-dnsamplificationattackscc.html

A string like this as suggested on above link works exactly for that domain

iptables --insert INPUT -p udp --dport 53 -m u32 --u32
"0x28&0xFFDFDFDF=0x17444e53 && 0x2c&0xDFDFDFDF=0x414d504c &&
0x30&0xDFDFDFDF=0x49464943 && 0x34&0xDFDFDFDF=0x4154494f &&
0x38&0xDFDFDFDF=0x4e415454 && 0x3c&0xDFDFDFDF=0x41434b53 &&
0x40&0xFFDFDFFF=0x02434300" -j DROP -m comment --comment "DROP DNS Q
dnsamplificationattacks.cc"

but here I am not sure how to create such string out and script them for
automation.

Can someone suggest a way out for this within IPTables or may be some other
open source firewall?

Thanks.

2

This is going to be tricky to do, as DNS packets don't necessarily contain
entire query values or FQDNs as complete strings due to packet label
compression (remember, original DNS only has 512 bytes to work with).

You can use those u32 module matches to find some known-bad packets if
they're sufficiently unique, but it simply lacks enough logic to fully
parse DNS queries.
Here's an interesting example to visualize what's happening:
http://dnsamplificationattacks.blogspot.com/p/iptables-block-list.html

One quick thing that would work would be to match a single label (e.g.
"google", but not "google.com"), but this will end up blocking any frames
with that substring in it (e.g. you want to block "evil.com", but this also
blocks "evil.example.com").

If you find yourself needing to parse and block DNS packets based on their
content in a more flexible way, I would look into either making an iptables
module that does the DNS parsing (
http://inai.de/documents/Netfilter_Modules.pdf), or using a userspace
library like with NFQUEUE (e.g. https://pypi.python.org/pypi/NetfilterQueue)
or l7-filter (http://l7-filter.sourceforge.net/).

Best of luck and happy hacking!

Cheers,
jof

3

Howdy,

The DNS query essentially always contains the full string in a
sequence. It doesn't *have* to per the protocol but you'll be hard
pressed to find a real-world example where it doesn't.

The catch is, the dots aren't encoded. The components of the name
being queried are separated by a byte indicating the length of the
next piece. So, instead of www.google.com the query packet contains
www 0x06 google 0x03 com.

You can implement this with --hex-string instead of --string but
you'll have to convert the entire thing to hex first

Regards,
Bill Herrin

4

Have you looked at perhaps using DNS RPZ (Response Policy Zones)?

https://dnsrpz.info/

- - ferg

5

You could use RPZ but wouldn't something as simple as putting these two entries in a host files meet the mail?

Tom

6

I implemented this easily some time ago due to a situation where product
development was unable or unwilling to disable open resolvers.

i'll post my ruleset then describe it then describe it since it contains
multiple functions.

    Chain INPUT (policy ACCEPT 68M packets, 4377M bytes)
     pkts bytes target prot opt in out
    source destination
      22M 1423M ACCEPT all -- lo *
    0.0.0.0/0 0.0.0.0/0
        0 0 REJECT all -- * *
    0.0.0.0/0 0.0.0.0/0 recent: CHECK name:
    blacklist side: source reject-with icmp-admin-prohibited
      34M 2463M find_dnsany udp -- * *
    0.0.0.0/0 0.0.0.0/0 udp dpt:53

    Chain FORWARD (policy ACCEPT 460M packets, 298G bytes)
     pkts bytes target prot opt in out
    source destination
        0 0 REJECT all -- * *
    0.0.0.0/0 0.0.0.0/0 recent: CHECK name:
    blacklist side: source reject-with icmp-admin-prohibited
        0 0 irc tcp -- * eth0
    0.0.0.0/0 0.0.0.0/0 multiport dports
    6660:6669,6670
    1826M 1144G local_ips all -- * *
    0.0.0.0/0 0.0.0.0/0
    35387 2569K find_dnsany udp -- * *
    0.0.0.0/0 0.0.0.0/0 udp dpt:53

    Chain OUTPUT (policy ACCEPT 39M packets, 316G bytes)
     pkts bytes target prot opt in out
    source destination
        0 0 irc tcp -- * eth0
    0.0.0.0/0 0.0.0.0/0 multiport dports
    6660:6669,6670
      22M 1423M ACCEPT all -- * lo
    0.0.0.0/0 0.0.0.0/0
     310M 1637G local_ips all -- * *
    0.0.0.0/0 0.0.0.0/0
      13M 1056M CONNMARK udp -- * *
    0.0.0.0/0 0.0.0.0/0 udp dpt:53 owner UID match
    25 CONNMARK set 0x35
      13M 1056M find_dnsany udp -- * *
    0.0.0.0/0 0.0.0.0/0 udp dpt:53

    Chain find_dnsany (3 references)
     pkts bytes target prot opt in out
    source destination
     302K 19M limit_dnsany all -- * *
    0.0.0.0/0 0.0.0.0/0 u32
    "0x0>>0x16&0x3c@0x8>>0xf&0x1=0x0&&0x0>>0x18&0x1=0x1" STRING match
    "|0000ff0001|" ALGO name bm FROM 36 TO 70 /* match ANY? queries */

    Chain irc (2 references)
     pkts bytes target prot opt in out
    source destination
        0 0 ULOG all -- * *
    0.0.0.0/0 0.0.0.0/0 ULOG copy_range 0 nlgroup
    30 queue_threshold 1
        0 0 LOG all -- * *
    0.0.0.0/0 0.0.0.0/0 LOG flags 8 level 4 prefix
    "[IRC] "
        0 0 REJECT all -- * *
    0.0.0.0/0 0.0.0.0/0 reject-with
    icmp-admin-prohibited

    Chain limit_dnsany (1 references)
     pkts bytes target prot opt in out
    source destination
      827 53727 ACCEPT all -- * * 1.2.3.4
       0.0.0.0/0 limit: avg 20/min burst 60
        0 0 limit_venet all -- * * 1.2.3.4
            0.0.0.0/0
     4297 302K ACCEPT all -- * *
    0.0.0.0/0 0.0.0.0/0 CONNMARK match 0x35
    limit: avg 10/min burst 30
    22798 1475K ACCEPT all -- * *
    0.0.0.0/0 0.0.0.0/0 limit: avg 4/min burst 10
     7277 468K LOG all -- * *
    0.0.0.0/0 0.0.0.0/0 limit: avg 1/min burst 5
    LOG flags 0 level 4 prefix "DNSANY: "
     279K 18M DROP all -- * *
    0.0.0.0/0 0.0.0.0/0

    Chain limit_venet (1 references)
     pkts bytes target prot opt in out
    source destination
        0 0 LOG all -- * *
    0.0.0.0/0 0.0.0.0/0 limit: avg 1/min burst 5
    LOG flags 0 level 4 prefix "DNSANYint: "
        0 0 REJECT all -- * *
    0.0.0.0/0 0.0.0.0/0 reject-with
    icmp-admin-prohibited

    Chain local_ips (2 references)
     pkts bytes target prot opt in out
    source destination
    2136M 2782G RETURN all -- * !eth0
    0.0.0.0/0 0.0.0.0/0 /* only check outgoing
    packets */
        0 0 RETURN all -- * *
    0.0.0.0/0 0.0.0.0/0 ADDRTYPE match src-type
    LOCAL /* accept packet generated from any locally bound IP */
        0 0 RETURN all -- * *
    0.0.0.0/0 0.0.0.0/0 recent: CHECK name:
    local_ips side: source
        0 0 ULOG all -- * *
    0.0.0.0/0 0.0.0.0/0 ULOG copy_range 0 nlgroup
    30 queue_threshold 1
        0 0 LOG all -- * *
    0.0.0.0/0 0.0.0.0/0 limit: avg 1/min burst 5
    /* block non-local IPs from exiting */ LOG flags 8 level 4 prefix
    "SPOOF: "
        0 0 REJECT all -- * *
    0.0.0.0/0 0.0.0.0/0 reject-with
    icmp-admin-prohibited

First, INPUT, FORWARD, and OUTPUT are very similar.
1) INPUT accepts all /lo/ traffic as does OUTPUT
2) INPUT and FORWARD auto block any IPs admins put into the
//proc/net/ipt_recent/blacklist/ (or //proc/net/xt_recent/blacklist/
depending on kernel version)
3) common IRC port traffic is blocked (insane number of bots use these
ports)
4) outgoing IPs are matched to interface IPs. this set of rules are used
for when the kernel is too old to employ
//proc/sys/net/ipv4/conf/all/rp_filter/ mode against forwarded IPs too
(prevents spoofing)
5) OUTPUT tags DNS traffic owned by uid 25 (usually for sendmail,
postfix, etc, change accordingly)
6) in order to prevent killing our box with syslog when an attack
happens, logging is strictly rate limited

now on to the DNS specific stuff
1) each of INPUT, FORWARD, and OUTPUT call /find_dnsany/ to identify our
suspect traffic
2) the rule in /find_dnsany/ uses a u32 match rule to first identify DNS
traffic that is a query, and a string match to identify the ANY flag,
further, we try to make this efficient by limiting our match to the byte
range of 36 to 70. we've found that 100% of our DNS amplification
attacks request zone data that fits within this. it certainly may be a
longer zone but is unlikely. change accordingly
3) once DNS ANY queries have been identified, jump to /limit_dnsany/
4) /limit_dnsany/ is designed to accept packets until a limit is
reached. there are three limits employed:
4.1) customers in a VZ or bound to a specific local IP that may have
higher than normal rates of legitimate DNS ANY queries
4.2) our local smtp initiated lookups. qmail is horrible in that it
employs DNS ANY (broken design, discussion out of scope)
4.3) all other DNS ANY traffic
5) incoming DNS ANY that is rate limited is DROPped on the floor, rate
limited outgoing is REJECted so internal customers get a friendly icmp
reject message

additional notes:
1) outgoing should also be checked against the blacklist, somehow our QA
dropped that rule before disting
2) IRC is checked before local traffic to ensure two infected customers
aren't communicating with each other

this set of rules reliably filters:
1) incoming and outgoing DNS QUERY floods. incoming is rate limited to
4/min per IP source. forwarded and outgoing has higher limits
2) majority of bot traffic on IRC ports
3) spoofed and blacklisted packets

-david

7

a message of 54 lines which said:

but here I am not sure how to create such string out and script them
for automation.

Use this program:

http://www.bortzmeyer.org/files/generate-netfilter-u32-dns-rule.py

8

a message of 88 lines which said:

This is going to be tricky to do, as DNS packets don't necessarily
contain entire query values or FQDNs as complete strings due to
packet label compression

Apprently, the OP wanted to match the *question* in a *query* and
these are never compressed (they could, in theory, but are not).

You can use those u32 module matches to find some known-bad packets
if they're sufficiently unique, but it simply lacks enough logic to
fully parse DNS queries.

u32's language is not Turing-complete but It is sufficient in the case
presented here.

9

This is going to be tricky to do, as DNS packets don't necessarily contain
entire query values or FQDNs as complete strings due to packet label
compression (remember, original DNS only has 512 bytes to work with).

Howdy,

The DNS query essentially always contains the full string in a
sequence. It doesn't *have* to per the protocol but you'll be hard
pressed to find a real-world example where it doesn't.

The catch is, the dots aren't encoded. The components of the name
being queried are separated by a byte indicating the length of the
next piece. So, instead of www.google.com the query packet contains
www 0x06 google 0x03 com.

For the completeness of the archives, the length of the first token is
also encoded and final terminator is 0.

0x03 www 0x06 google 0x03 com 0x00

-DMM

10

Thanks everyone for useful responses. I almost used script mentioned by
Stephane (http://www.bortzmeyer.org/files/generate-netfilter-u32-dns-rule.py)
but I realized that for a rule for "domain.com" it blocks "domain.com" only
and their was no easy way out to block subdomains as well. In last few days
after my post, I noticed traffic in pattern of sub1.sub2.domain.com where
sub1 and sub2 are randomly generated strings.

I tried creating .domain.com and other rules in u32 but didn't help for
subdomain. Also since there were very high number of subdomains (but
limited domains), possibility to generate u32 rule for each sub didn't made
sense. I re-visited Hexadecimal string with 03 and 00 for dot was actually
able to help.

RPZ and some other option I am still exploring.

Thanks.