Sunday, September 29, 2019

Datasets with Suricata

Suricata recently introduced datasets so I thought I would take a stab at using them and seeing what could maybe be done with them.

From the 5.0.0rc1 release announcement:

"Still experimental at this time, the initial work to support datasets is part of this release. It allows matching on large amounts of data. It is controlled from the rule language and will work with any ‘sticky buffer’.

A lot of my recent research/work has been around DNS so I figured I would start there.

Step 1:

Get suricata master or 5.0.0rc1 installed, the installation from source instructions are here:

Source tarballs can be retrieved from:

If you install from github be sure to clone libhtp into the suricata repo directory, libhtp can be found here:

On a Fedora 30 system installing the build time requirements should be something like:'sudo dnf install gcc gcc-c++ rust cargo libyaml-devel python3-pyyaml libnfnetlink-devel libnetfilter_queue-devel libnet-devel zlib-devel pcre-devel libcap-ng-devel lz4-devel libpcap-devel nspr-devel nss-devel nss-softokn-devel file-devel  jansson-devel GeoIP-devel python3-devel lua-devel autoconf automake libtool'

My particular install steps for Suricata from source (Fedora 30) with all the build requirements installed:

mkdir git
cd git
git clone
cd suricata
git clone
./; ./configure --enable-gccprotect --enable-pie --disable-coccinelle --enable-nfqueue --enable-af-packet --with-libnspr-includes=/usr/include/nspr4 --enable-jansson --enable-geoip --enable-lua --enable-rust --enable-debug --enable-profiling --enable-rust-debug --enable-lzma
sudo make install

Step 2:

I chose to put part of the configuration in the suricata.yaml file since setting up the datasets seems like something that would be easier to do in the yaml with a configuration management setup (chef, ansible, puppet, etc.).

So I tossed the following in the suricata.yaml just under the yaml header/boilerplate stuff:

 - dns-seen:
     type: sha256
     state: dns-seen.lst

This is telling Suricata that I want to set up a dataset name 'dns-seen' and that it will contain sha256 values and running and saved information about the dataset will be stored in a file named dns-seen.lst.

So far so good.

Step 3:

Time to write a rule that will use the dataset. Since we are going to look for DNS queries, it only makes sense of course to use the DNS sticky buffer. Time to be creative... :)

alert dns any any -> any any (msg: "dns list test"; dns.query; to_sha256; dataset:isset,dns-seen; sid:123; rev:1;)

What this rule will do is write an alert (in our case to alert.json) for any traffic that the Suricata protocol parser(s) determines is a DNS query for any domain in our dns-seen.lst file.

One thing worth noting here is the to_sha256 keyword, this is what Suricata calls a transform. This keyword will tell suricata to take whatever is in the dns buffer and calculate a sha256 hash. To say the least transforms are quite useful in rule writing!

For more on transforms:

Step 4:

Okay so we have all the prep work done..well not quite. So generally speaking domain IOCs don't come to us as sha256 hashes. Soo what do we do?

We write some bad python:

This simple script takes a file with domain names and writes out a file with the corresponding sha256 hash values.

In my case I selected,, and to use in the dns-seen.lst file.

Step 5:

Now that we have our hashes, we just need a pcap with DNS traffic that contains DNS queries for our domains. Using dig and tcpdump we can generate the traffic and pcap.

something like 'tcpdump -nn -i $activenetworkcard -w dnslisttest.pcap' and then


Step 6:

Now we can replay our pcap through suricata to see what happens. I ran the following:

jason@dinosaur suri]$ rm *.json *.log; $(which suricata) -k none -c suricata.yaml -r dnslisttest.pcap

This just makes sure I don't have an old logs laying around and then runs suricata with my configuration file and replays the pcap we captured in Step 5.

If I look at the alert.json file (eve log configured for alerts):
{"timestamp":"2019-09-28T20:38:06.536624-0400","flow_id":2206638181068848,"pcap_cnt":11,"event_type":"alert","src_ip":"","src_port":38966,"dest_ip":"","dest_port":53,"proto":"UDP","tx_id":0,"alert":{"action":"allowed","gid":1,"signature_id":123,"rev":1,"signature":"dns list test","category":"","severity":3},"dns":{"query":[{"type":"query","id":16741,"rrname":"","rrtype":"A","tx_id":0}]},"app_proto":"dns","flow":{"pkts_toserver":1,"pkts_toclient":0,"bytes_toserver":93,"bytes_toclient":0,"start":"2019-09-28T20:38:06.536624-0400"},"payload":"QWUBIAABAAAAAAABBmdvb2dsZQNjb20AAAEAAQAAKRAAAAAAAAAMAAoACAI9cKIVTyJM","stream":0,"packet":"8nUnCb+QtLZ2CFMnCABFAABPXUkAAEARcSysECoHrBAqAZg2ADUAO6PpQWUBIAABAAAAAAABBmdvb2dsZQNjb20AAAEAAQAAKRAAAAAAAAAMAAoACAI9cKIVTyJM","packet_info":{"linktype":1}}
{"timestamp":"2019-09-28T20:38:01.959896-0400","flow_id":2191197773342104,"pcap_cnt":5,"event_type":"alert","src_ip":"","src_port":38358,"dest_ip":"","dest_port":53,"proto":"UDP","tx_id":0,"alert":{"action":"allowed","gid":1,"signature_id":123,"rev":1,"signature":"dns list test","category":"","severity":3},"dns":{"query":[{"type":"query","id":22527,"rrname":"","rrtype":"A","tx_id":0}]},"app_proto":"dns","flow":{"pkts_toserver":1,"pkts_toclient":0,"bytes_toserver":95,"bytes_toclient":0,"start":"2019-09-28T20:38:01.959896-0400"},"payload":"V\/8BIAABAAAAAAABCHNsYXNoZG90A29yZwAAAQABAAApEAAAAAAAAAwACgAIrWxDkiq\/JNg=","stream":0,"packet":"8nUnCb+QtLZ2CFMnCABFAABRTS4AAEARgUWsECoHrBAqAZXWADUAPe+oV\/8BIAABAAAAAAABCHNsYXNoZG90A29yZwAAAQABAAApEAAAAAAAAAwACgAIrWxDkiq\/JNg=","packet_info":{"linktype":1}}
{"timestamp":"2019-09-28T20:38:10.967322-0400","flow_id":781430593602202,"pcap_cnt":13,"event_type":"alert","src_ip":"","src_port":39980,"dest_ip":"","dest_port":53,"proto":"UDP","tx_id":0,"alert":{"action":"allowed","gid":1,"signature_id":123,"rev":1,"signature":"dns list test","category":"","severity":3},"dns":{"query":[{"type":"query","id":11178,"rrname":"","rrtype":"A","tx_id":0}]},"app_proto":"dns","flow":{"pkts_toserver":1,"pkts_toclient":0,"bytes_toserver":93,"bytes_toclient":0,"start":"2019-09-28T20:38:10.967322-0400"},"payload":"K6oBIAABAAAAAAABBnJlZGRpdANjb20AAAEAAQAAKRAAAAAAAAAMAAoACHszHT2Q7UUp","stream":0,"packet":"8nUnCb+QtLZ2CFMnCABFAABPaqMAAEARY9KsECoHrBAqAZwsADUAO6btK6oBIAABAAAAAAABBnJlZGRpdANjb20AAAEAAQAAKRAAAAAAAAAMAAoACHszHT2Q7UUp","packet_info":{"linktype":1}}

It fired alerts!

So this is pretty interesting functionality and I expect to test a lot more with it in the upcoming months.

No comments:

Post a Comment