Nova is a software application for preventing and detecting hostile network reconnaissance (such as nmap scans). It does this by first creating the Haystack: a large collection of low interaction honeypots using an updated version of Honeyd.
Finding real machines on the network becomes like finding a needle in a haystack of fake machines. Second, Nova uses machine learning algorithms to automatically detect and classify attempts at hostile reconnaissance, so there's no need to go searching manually through your honeypot's log files. It provides an easy to use Web-based interface powered by Node.js to configure itself and Honeyd instances.
Nova is a software application for preventing and detecting hostile Network
Reconnaissance. It does this by first creating the Haystack: a large array of
thin virtual machines on the target network. These VMs are created using an
updated Honeyd to be extremely lightweight. They're not your typical VMs that
you might see from VirtualBox or VMWare. They just appear to be real from the
perspective of the network, and run network "services", which are just shell
- Second, Nova uses machine learning algorithms to automatically detect and
classify attempts at hostile reconnaissance, so there's no need to go
searching manually through your honeypot's log files.
The Installation Guide
- The first thing to note is that Nova is currently only compatible with Linux.
All of our development is done on Ubuntu 12.04, so we suggest using that to
make installation easiest. We provide a helper script which should get all
dependencies and download, build, and install Nova and Honeyd.
sudo bash novaInstallHelper.sh
- This script has only been tested on the most recent stable version of Ubuntu.
Any other distributions or versions should manually compile using the
Getting the newest code
- Nova and Honeyd are stored as seperate Git repositories on github. Go to the
directory you wish to download the code to and run the following commands,
git clone git://github.com/DataSoft/Honeyd.git
git clone git://github.com/DataSoft/Nova.git Nova
# You also need to get the git submodules where training data is stored
# in it's own repo
git submodule init
git submodule update
- This will create a "honeyd" and "Nova" folder with the source located inside.
From this point on they will be referred to as $HONEYD_SOURCE and
- This will default to the "master" branch, which is the latest stable release.
If you want to use the latest unstable version, cd to the $NOVA_SOURCE and
$HONEYD_SOURCE and run the following,
git checkout integration
- Beware that the integration branch changes on a daily basis and may be
Getting Dependencies on Ubuntu
- Install required dependencies with the following command:
sudo apt-get install git build-essential libcap2-bin libann-dev libpcap0.8-dev libboost-program-options-dev libboost-serialization-dev libnotify-dev sqlite3 libsqlite3-dev libcurl3 libcurl4-gnutls-dev iptables libxml2-dev libboost-system-dev libboost-filesystem-dev
- Now, we'll have to take a quick detour to get another integral component of
Nova: Honeyd. We will have to download some extra libraries for Honeyd as
well; you can get them using this command:
sudo apt-get install libevent-dev libdumbnet-dev libpcap-dev libpcre3-dev libedit-dev bison flex libtool automake
- For the Honeyd Autoconfiguration tool, we require Nmap 6.00 or higher. The
current version in the apt repository is 5.21, so you'll have to go to the
Nmap website and get 6.00. It can be found here:
- To get the dependencies for the Quasar web UI (nodejs 0.8.5, npm's forever,
and cvv8) you can either install them manually or get them by running the
sudo bash Quasar/getDependencies.sh
- There are instructions for manual install on the same page (but are just the
standard ./configure, make, and sudo make install commands).
- NOTE: Honeyd requires libevent version 2.x. If you are running Ubuntu 10.10 or
lower, the version of libevent available in the repos is only 1.x. So you will
need to either find a backport or build libevent 2.x from source.
- If you wish to optionally generate Debian packages for Nova, you will also
sudo apt-get install dpkg-dev
- Change directories to the $HONEYD_SOURCE folder where all of the Honeyd source
code should be on your machine. When inside the source directory, follow the
next steps to build and install.
Step 1: ./autogen.sh
Step 2: automake
Step 3: ./configure
Step 4: make
Step 5: sudo make install
- Change into the $NOVAD_SOURCE folder where the novad source code resides.
To build and install Nova run the commands,
Step 1: autoconf
Step 2: ./configure
Step 3: make
Step 4: sudo make install
- Note: If building fails for some reason, make sure you run 'make clean' before
- Finally, while logged in as the user you plan to run Novad with, run the
following command to add your user to the 'nova' permission group and to set
up database tables for the web interface.
Step 5: sudo nova_init
- Your user will have to be in the "nova" group in order for nova and Honeyd to
run properly. The nova_init script will do this, but you must log in and back
out for the change to take effect.
Step 6: Log out and log back in
Refer to the Nova wiki on github for more information.
Daemonizing with Upstart
- If you want to start Quasar, novad, and the haystack when the machine boots
and have them restart if they crash, you can use the upstart service by copying
the files in $NOVAD_SOURCE/Installer/miscFiles/upstart/* to /etc/init. This is
assuming that upstart is already installed and configured on your system (it
comes by default on newer versions of Ubuntu).
High Level Nova Components
Haystack: Active honeypots
- The Haystack is the collection of honeypots which emulate machines on
the network. The haystack is created using the Honeyd daemon and runs
in it's own executable. Configuration for Honeyd is auto generated at
Novad: Classification tool
- The Novad executable is the daemon that monitors and classifies
network traffic to identify hostile looking traffic. Novad will listen
promiscuously on the configured network interfaces and keep track of
various statistics such as IPs contacted, ports contacted, honeypots
contacted, and other details. Novad is can be configured manually via
the configuration file ~/.config/nova/config/NOVAConfig.txt, but it is
recommended that you use the GUI (Quasar) unless you know what you're
NovaCLI: Nova Command line Interface
- NovaCLI provides a simple interface for accessing some of the Novad
functionality. Usage for the tool can be gotten by running "novacli
Quasar: Nova Web Interface
- Nova's main GUI, Quasar, is a web interface run with a nodejs web
- To start the web interface, run the command "quasar" and go to
https://localhost:8080 in a web browser.
Default username: nova
Default password: toor
- "quasar --debug" may provide more information if there are problems.
Quasar launches the nodejs server with the "forever" daemon so it will
be restarted if it crashes. The command "forever list" can be useful
for seeing the current status, and it can be stopped with "forever
stop index (usually 0)". See the forever documentation for more
Haystack Auto Configuration Tool: Generates honeyd configurations based off of
- This tool can scan your network with nmap and then generate honeypot
configurations that are based on the operating systems and ethernet
vendors that it finds.
NovaTest: Unit Tests
- If you're a developer interested in using the unit tests in NovaTest, you can
find instructions at,
A set of example TLS keys are provided, but because of their public nature
provide no real security. Paths to the TLS keys are in the Nova configuration
file at ~/.config/nova/config/NOVAConfig.txt
To generate a self signed certificate and key for the Quasar or Pulsar https
# Generate a private key
openssl genrsa -des3 -out ui.key 1024
# Create a request for a certificate
openssl req -new -key ui.key -out ui.csr
# Generate a self signed certificate
openssl x509 -req -days 365 -in ui.csr -signkey ui.key -out ui.crt
# Creating keys for the Pulsar/Quasar connection is a bit more complicated.
# Pulsar authenticates clients by using TLS client certificates signed by a
# certificate authority.
# Create a new certificate athority
openssl genrsa -des3 -out ca.key 1024
openssl req -new -key ca.key -out ca.csr
openssl x509 -req -days 365 -in ca.csr -out ca.crt -signkey ca.key
# Create and sign the Pulsar key
openssl genrsa -des3 -out pulsarTether.key 1024
openssl req -new -key server.key -out pulsarTether.csr
openssl x509 -req -in pulsarTether.csr -out pulsarTether.crt -CA ca.crt -CAkey ca.key -CAcreateserial -days 365
# Create and sign the Quasar keys. For each quasar instance,
openssl genrsa -des3 -out quasarTether.key 1024
openssl req -new -key server.key -out quasarTether.csr
openssl x509 -req -in quasarTether.csr -out quasarTether.crt -CA ca.crt -CAkey ca.key -CAcreateserial -days 365
# Transfer this key to the Quasar instance
Remember to make sure that all paths and passphrases are updated correctly in
~/.config/nova/config/NOVAConfig.txt to use the new keys you created.
- To generate a Debian package, simply checkout what version of the software you
like (or make what changes to it that you want) and run the generateDebs
script (as a normal user).
./generateDebs <version number>
Common Problems and solutions
Haystack Autoconfig nmap fails on large networks
Nmap will often fail when scanning networks of size greater than 1024
IPs with the error "nexthost: failed to determine route" or "Strange
connect error(105): No buffer space available". This is usually
caused by the kernel ARP table running out of space and not being
garbaged collected fast enough to handle all of the ARP requests nmap
is doing. The solution is to increase the size by adding the following
lines to /etc/sysctl.conf,
net.ipv4.neigh.default.gc_thresh1 = 1024
net.ipv4.neigh.default.gc_thresh2 = 4096
net.ipv4.neigh.default.gc_thresh3 = 65536
Then run the command,
And try running the Haystack autoconfig tool again.
Tips for debugging problems
To enable verbose debug log messages, run the command,
novacli writesetting SERVICE_PREFERENCES 0:0+\;1:5+\;2:6+\;
If the above fails for some reason, you can also change the logging
settings manually in the ~/.config/nova/config/NOVAConfig.txt file
under the SERVICE_PREFERENCES setting.
You should be able to run quasar/novad/honeyd without needing explicit
root permissions. One requirement for this is that the user you're
running with is in the "nova" group and has run the nova_init script
located in Installer/. This script adds the user to the group and also
configures sudo (via adding a file to /etc/sudoers.d). Logging out and
back in is required for the group addition to work.
If you're seeing permission related errors, you can try the following
sudo chmod -R g+rw /usr/share/nova
sudo chgrp -R nova /usr/share/nova
Web interface problems
If you can't access the web interface, try stopping it if it's running
in 'forever' and manually running it as a foreground process with the
This should provide more verbose output and show if it is crashing
rather than running it as a background daemon process.
If Novad appears to be having problems, try to start it manually
instead of as a background process with the command
novacli start nova debug
If the Haystack appears to be having problems, try to start it
manually instead of as a background process with the command,
novacli start haystack debug
If something gets messed up to the point you want to start over, you
can do so with the commands,
cd $NOVA_SOURCE sudo make reinstall
Note that this will remove any configuration changes that you made.
Building with debugging symbols
If you're seeing novad crash, it might be helpful to compile with
debugging symbols and get a stack trace.
cd $NOVA_SOURCE make clean make debug make reinstall
gdb novad run backtrace
There is an option for designating a target Rsyslog instance electing to receive
messages exposed within the Advanced Options page of the Quasar Web UI. Some
-Make sure that whatever IP is pointed is given in the format IP_ADDRESS:PORT.
-Make sure the designated port is both open and listening on the receiving
machine. The easiest way to do this is to uncomment the InputTCPServer lines in
/etc/rsyslog.conf and change the port number away from 514 (because rsyslog now
drops permissions, using port 514 is no longer an option, as it's < 1024). To
test that rsyslog is listening, run
netstat -tlnup | grep PORT
as root and check that the PID/Name combination for rsyslogd is listed under
the results. The port may also need to be registered into the /etc/services
file, if changed from the normal port 514.
-Within /etc/rsyslog.d/ lie the configuration files; in one of these files, a
rule MUST be created similar to the following:
where YOUR_ACTION_CHOICE represents the action (most likely a write to a
destination) to take upon receipt of messages from a client server that have
those program names. This is to help organize the logs, as Nova can
potentially send many log messages that would otherwise pollute the normal
syslog file. Note that there will be three rules like this total, one each
for the strings "Nova", "novad" and "honeyd"
Testing that these changes worked is a good idea as well. Simply start and stop
novad on the client with the novacli command line interface and check that the
log messages sent at startup arrived at the right place
Pulsar does not install with the standard Nova ./configure, make, make
install process. Instead, the user must change directory into the Nova
directory (most commonly located in the /home/$USER/Code/ directory) and
run make install- pulsar with superuser permissions. This will place the
Pulsar files within the proper directories and allow for the user to
use the alias 'pulsar' on the command line to start the forever process
for Pulsar. To access the Pulsar interface, the user must first
have configured Nova such that it has the MASTER_UI_ENABLED
configuration variable set to 1, as well as properly configuring the
MASTER_UI_IP and MASTER_UI_CLIENT_ID variables to match the location and
naming requirements for the user's network.