honeyd(8) Honeypot Daemon


honeyd [-dP ] [-l logfile ] [-s servicelog ] [-p fingerprints ] [-0 p0f-file ] [-x xprobe ] [-a assoc ] [-f file ] [-i interface ] [-u uid ] [-g gid ] [-c host:port:username:password ] [--webserver-address address ] [--webserver-port port ] [--webserver-root path ] [--rrdtool-path path ] [--disable-webserver ] [--disable-update ] [--verify-config ] [--fix-webserver-permissions ] [-V|--version ] [-h|--help ] [--include-dir ] [--data-dir ] [net ... ]


Honeyd creates virtual hosts for IP addresses matching the specified net The daemon simulates the networking stack of the configured hosts and can simulate any TCP and UDP service. ICMP is fully supported, too. By default, all UDP ports are closed and honeyd will generate an ICMP unreachable port message if the configured personality permits that.

Honeyd enables a single host to claim unused addresses on a LAN for network simulation. The net argument may contain multiple addresses and network ranges. For configured templates (see below) to work all the configured addreses need to be included in net Notice that honeyd will answer to ICMP requests for all addresses defined in net if there is no configured specific host or default template.

In order for honeyd to receive network traffic for IP addresses that it should simulate, it is necessary to either explicitly route traffic to it, use proxy arp or run farpd(8) for unassigned IP addresses on a shared network. By assigning an ethernet address to a template via

  set template ethernet "<vendor|mac address>"

it is possible to integrate virtual honeypots into a production network without running farpd(8). Futhermore, Honeyd supports acquiring IP addresses via DHCP. You can ask a template to be assigned to a DHCP IP address via

  dhcp template on fxp0 [ethernet "<vendor|mac address>"]

Honeyd exits on an interrupt or termination signal and rotates logfiles on SIGUSR1.

The options are as follows:

Do not daemonize, and enable verbose debugging messages.
On some operating systems, it is not possible to get event notifications for pcap via select(2). In that case, honeyd needs to run in polling mode. This flag enables polling.
-l logfile
Log packets and connections to the logfile specified by logfile
-s servicelog
Logs information from service scripts to the log file specified by servicelog
-p fingerprints
Read nmap style fingerprints. The names defined after the Fingerprint token are stored as personalities. The personalities can be used in the configuration file to modify the behaviour of the simulated TCP stack.
-x xprobe
Read xprobe style fingerprints. This file determines how honeyd reacts to ICMP fingerprinting tools.
-a assoc
Read the file that associates nmap style fingerprints with xprobe style fingerprints.
-0 p0f-file
Read the database for passive fingerprinting. The names of the operating systems specified in this file are recognized by Honeyd's parser and can be used for dynamic templates.
-u uid
Set the UID that Honeyd is running as.
-g gid
Set the GID that Honeyd is running as.
-f file
Read the configuration in file It is possible to create host templates with the configuration file that specify which servers should run and which scripts should be started to simulate them. honeyd will reread the configuration file when sent a SIGHUP signal.

The syntax is as follows:

config  = creation | addition | delete | binding | set |
          annotate | route [config] | option
creation= "create" template-name | "create" "default" |
  "dynamic" template-name
addition= "add" template-name proto "port" port-number action |
  "add" template-name "subsystem" cmd-string ["shared"] |
  "add" template-name "use" template-name "if" condition
delete= "delete" template-name |
  "delete" template-name proto "port" port-number
binding = "bind" ip-address template-name |
  "bind" condition ip-address template-name |
  "bind" ip-address "to" interface-name |
  "dhcp" template-name "on" interface-name ["ethernet" cmd-string] |
  "clone" template-name template-name
set     = "set" template-name "default" proto "action" action |
  "set" template-name "personality" personality-name |
  "set" template-name "personality" "random"
  "set" template-name "ethernet" cmd-string
  "set" template-name "uptime" seconds
  "set" template-name "droprate" "in" percent
  "set" template-name "uid" number ["gid" number]
  "set" template-name "spoof" ["from" ip-address] ["to" ip-address]
  "set" template-name "maxfds" number
  "set" ip-address "uptime" seconds
annotate= "annotate" personality-name [no] finscan |
  "annotate" personality-name "fragment" ("drop" | "old" | "new")
route   = "route" "entry" ipaddr |
  "route" "entry" ipaddr "network" ipnetwork |
  "route" ipaddr "link" ipnetwork |
  "route" ipaddr "unreach" ipnetwork |
  "route" ipaddr "add" "net" ipnetwork \
                 "tunnel" ipaddr(src) ipaddr(dst) |
  "route" ipaddr "add" "net" ipnetwork ipaddr \
                ["latency" number"ms"] ["loss" percent] \
                ["bandwidth" number["Mbps"|"Kbps"] \
                ["drop" "between" number "ms" "-" number "ms" ]
proto   = "tcp" | "udp" | "icmp"
action  = ["tarpit"] ("block" | "open" | "reset" | cmd-string | \
  "internal" cmd-string \
  "proxy" ipaddr":"port )
condition = "source os =" cmd-string |
  "source ip =" ipaddr | "source ip =" ipnetwork |
  "time " timecondition | "proto" protocol | "otherwise"
timecondition = "between" time "-" time
option  = "option" plugin option value

The cmd-string and the personality-name are arbitrary strings enclosed with quotation marks. Variable expansion on the tokens $ipsrc $ipdst $sport $dport and $date is performed when executing the command string or when resolving the proxy address. Additionally, the environment variables HONEYD_IP_SRC HONEYD_IP_DST HONEYD_DST_PORT HONEYD_SRC_PORT HONEYD_PERSONALITY and HONEYD_REMOTE_OS are available, too.

If the internal key word is use, honeyd interprets the command string as Python module. These modules are executed within honeyd without forking a new process. As a result, internal scripts are very fast and cheap to execute.

The special keyword tarpit is used to slow down the progress of a TCP connection. This is used to hold network resources of the connecting computer.

If an IP address is not bound to a template, the actions specified in the default template are executed.

Personalities need to be annotated before they are assigned to a template or an IP address.

The default fragment policy is to accept fragment and resolve overlaps in favor of old data. If the personality returns TCP timestamps, the default uptime is a randomly chosen between zero and twenty days.

The special include directive may be used to include other configuration files, for example to keep all personality annotations separate from the main configuration file.

All honeyd plugins can be configured via the configuration file. Each configuration option goes in one line, indicated by the option keyword. It is followed by three items: the name of the plugin, the name of the configuration option, and a value. The value can be either an integer, a float, or a character string. The options are picked up when honeyd reads the configuration file and can then be queried by the plugins.

-i interface
Listen on interface It is possible to specify multiple interfaces.
-c hostname:port:username:password
Using this flag, Honeyd functions as a traffic statistic collector. Collected statistics get propagated upstream to an aggregator running at the specified hostname and port. The username and password is used to create a signature on the data packet that can be used to verify the integrity of the data. The statistics can be used to automatically detect anomalies like worm propagation.
--webserver-address address
Specifies the address on which the web server should listen. By default, this is so that only local requests are served. By specifying the webserver is going to answer to external requests, too.
--webserver-port port
Specifies the port on which the web server should listen.
--webserver-root path
The path to the document tree of the webserver. This is usually {prefix}/shared/honeyd/webserver/htdocs/
--rrdtool-path path
Specifies the path for rrdtool(1). Without rrdtool no traffic graphs can be generated.
Disables the builtin webserver.
Prevents Honeyd from checking if there are any security problems with the current version of the application.
Verifies that Honeyd can parse the configuration correctly. This does not require any special permissions, although some configurations that require direct access to interfaces might fail to validate.
Changes the ownership of the web server files to the user, Honeyd is going to run as.
Print version information and exit.
Print summary of command line options and exit.
For plugin development. Reports the directory in which honeyd stores its header files.
Reports the directory in which honeyd stores data files like Python modules.
The IP address or network (specified in CIDR notation) or IP address ranges to claim (e.g. ``'', ``'' or ``''). If unspecified, honeyd will attempt to claim any IP address it sees traffic for.


honeyd supports the creation of a complete network topology including routing. In order to enable the simulation of a network topology, a router entry point has to be configured with
  route entry <IP address>

By adding a network to a router entry point, honeyd is told about the network addresses this entry point is responsible for. This enables multiple entry points into the routing topology.

Every route add net directive creates the specified gateway as a new router. In the case of tunneling, no new router is created, instead packets are gre(4) encapsulated and sent to the tunnel destination address. It is assumed that the tunnel destination address routes the encapsulated packets to a honeyd machine.

The virtual machines that can be directly accessed by a router are defined as network range in the route link command.

A link may carry attributes like latency loss and bandwidth The latency specifies a constant delay for packets travelling across the link. The bandwidth on the other hand tracks the bandwidth related queuing delay for each link. If a packet is still being transmitted on the link then the queue delay for another packet is the propagating delay depending on the bandwidth plus the time for the previous packet to clear the link.

Unless the link is configured to drop packets between a configurable delay threshold, Honeyd currently assumes infinite buffer space, so use this option with care.

An address space can be made unrouteable via the route unreach command.

The router entry point is the first address that inspects a packet. The packet follows a path defined by the network topology until the current router has the destination IP address on its local network.

It is possible to integrate real machines into the routing topology. honeyd takes care of ARP requests and replies and encapsulates packets that go to external machines into ethernet packets.

External machines can be configured with the following command:

  bind <IP address> to <interface name>


Subsystem virtualization allows you to run regular network applications under a virtual IP address controlled by honeyd The application's network calls are intercepted and virtualized to the honeypot that they are configured to. As a result, all network calls that subsytem applications make appear to originate from the virtual IP address of a honeypot. This includes binding ports, accepting and initiating UDP and TCP connections. Raw sockets are not supported.

Subsystem are configured as follows

    set template subsystem "/usr/sbin/httpd"

and are started as a separate process for every bound template. Applications started as a honeyd subsystem need to be dynamically linked in order to work under Honeyd

It is possible to shared subsystems across different addresses if they are created with the shared flag. This allows a subsystem to bind to several virtual IP addresses and may also be used to increase the performance of subsystems across addresses.


Dynamic templates give Honeyd the ability to change networking behavior based on several different conditions:

source address
The source address of the network connection determines which template is going to be used.
operating system
The operating system as determined by passive fingerprinting needs to be matched for the template to be activated.
The template is only being used between a certain time interval. This allows Honeyd to simulate machines being turned on and off.

A dynamic template can be created with the following command:

  dynamic magichost
  add magichost use windowsxp if source os = windows
  add magichost use linux if source ip =
  add magichost use invisible if time between 12:00am - 5:00am
  add magichost otherwise use default 

As an alternative, it is possible to use a short cut in the bind command to create dynamic templates:

  bind source ip = cisco
  bind source ip = juniper

In this example, the host on behaves like a cisco router if it is contacted from IP addresses in the 192.168 network. If it is contacted from IP addresses in the 10 network, it behaves like a juniper router.


The honeydctl(1) command allows the dynamic configuration of Honeyd while it is running; see honeydctl(1) for more information.


Honeyd has two different logging modes. The syslog facility is used to log connection establishment and termination including other relevant packet events. Most messages can be disabled when configuring syslog.conf5 to drop all messages for the LOG_DAEMON facility if the log level is below LOG_NOTICE

Services started by honeyd can cause the daemon to log data by sending information to stderr

The second way of logging network activity is by using the -l flag. This causes honeyd to log all received packets in a human readable format. For UDP and TCP connections, honeyd logs the start and end of a flow including the amount of data transfered.

For logging any other information, it is suggested to run a separate intrusion detection system.


Honeyd supports internal service scripts that have been written in Python. To improve the performance of these services, Honeyd provides an event-driven model. The services need to indiciate when they are ready to read and when they are ready to write data. Honeyd keeps track of state that is provided to the Python scripts on every invocation.

The folowing example uses a Python script to implement a simple echo server:

  import honeyd
  import sys
  def honeyd_init(data):
    mydata = {}
    return mydata
  def honeyd_readdata(mydata, data):
    mydata["write"] = data
    return 0
  def honeyd_writedata(mydata):
    data = mydata["write"]
    del mydata["write"]
    return data
  def honeyd_end(mydata):
    del mydata
    return 0


A sample configuration file looks as follows:
# Example of a simple host template and its binding
include annotations
# Set up the hosts
create template
set template personality "OpenBSD 2.6-2.7"
add template tcp port 80 "sh scripts/web.sh"
add template tcp port 22 "sh scripts/test.sh $ipsrc $dport"
add template udp port 53 proxy yournameserver:53
set template default tcp action reset
set template uid 32767 gid 32767
bind template
set uptime 1327650

A simple example of a routing topology:

route entry
route link
route add net latency 10ms loss 3.4
route link

For this topology to work the net value in the command line has to be


The PID of the current daemon.
Python modules and web server documents used by the internal webserver.
A shared library that can be preloaded to virtualize applications within honeyd
An association file to match xprobe2 fingerprints against nmap
Nmap fingerprints used by honeyd to impersonate operating system stacks.
Xprobe fingerprints used by honeyd to impersonsate the ICMP section of operating system stacks.


Niels Provos Aq [email protected]