28

using CentOS with static IP, Is there any way to determinate if a DHCP server running on the network using bash ?

49

nmap does this easily:

sudo nmap --script broadcast-dhcp-discover -e eth0

will show:

Starting Nmap 6.40 ( http://nmap.org ) at 2016-08-16 09:25 UTC
Pre-scan script results:
| broadcast-dhcp-discover: 
|   IP Offered: 192.168.0.67
|   DHCP Message Type: DHCPOFFER
|   Server Identifier: 192.168.0.1
|   IP Address Lease Time: 0 days, 0:05:00
|   Subnet Mask: 255.255.255.0
|   Router: 192.168.0.1
|   Domain Name Server: 8.8.8.8
|   Domain Name: maas
|   Broadcast Address: 192.168.0.255
|_  NTP Servers: 91.189.91.157, 91.189.89.199, 91.189.94.4, 91.189.89.198
WARNING: No targets were specified, so 0 hosts scanned.
Nmap done: 0 IP addresses (0 hosts up) scanned in 0.27 seconds

Note: there is a similar script for dhcpv6

sudo nmap --script broadcast-dhcp6-discover -e eth0
3
  • 2
    Although existing answers were fairly good, I much prefer your solution! Aug 19 '16 at 0:44
  • 4
    Good command but its worth noting that this just outputs the first DHCP server that responds. If multiple DHCP servers exists, this command won't find them.
    – poke
    Jul 11 '19 at 17:24
  • I prefer this variant, but it seems that the machine on which you're executing this, really gets an offered IP assigned to the queried interface. Any way to prevent that?
    – sgohl
    May 29 '20 at 18:44
6

If available in the repository there is dhcpdump

from man page:

SYNOPSIS
       dhcpdump [-h regular-expression] -i interface

DESCRIPTION
       This command parses the output of tcpdump to display the dhcp-packets for easier checking and debugging.

USAGE
       dhcpdump -i /dev/fxp0

       If you want to filter a specific Client Hardware Address (CHADDR), then you can specifiy it as a regular expressions:

       dhcpdump -i /dev/fxp0 -h ^00:c0:4f

       This will display only the packets with Client Hardware Addresses which start with 00:c0:4f.
5

If you have tcpdump available to you, invoking the program as root with the following parameters might assist you in finding the server:

tcpdump -i [interface id] -nev udp port 68

Unfortunately, due to my network's layout, I can't get a full DHCP handshake captured right away. However, I do see a DHCP Request from my iPad:

22:16:44.767371 30:10:e4:8f:02:14 > ff:ff:ff:ff:ff:ff, ethertype IPv4 (0x0800), length 342: (tos 0x0, ttl 255, id 15652, offset 0, flags [none], proto UDP (17), length 328)
    0.0.0.0.68 > 255.255.255.255.67: BOOTP/DHCP, Request from 30:10:e4:8f:02:14, length 300, xid 0x42448eb6, Flags [none]
      Client-Ethernet-Address 30:10:e4:8f:02:14
      Vendor-rfc1048 Extensions
        Magic Cookie 0x63825363
        DHCP-Message Option 53, length 1: Request
        Parameter-Request Option 55, length 6: 
          Subnet-Mask, Default-Gateway, Domain-Name-Server, Domain-Name
          Option 119, Option 252
        MSZ Option 57, length 2: 1500
        Client-ID Option 61, length 7: ether 30:10:e4:8f:02:14
        Requested-IP Option 50, length 4: 192.168.2.222
        Lease-Time Option 51, length 4: 7776000
        Hostname Option 12, length 15: "NevinWiamssiPad"

After letting `tcpdump' run overnight, I did eventually see this ACK:

07:46:40.049423 a8:39:44:96:fa:b8 > 68:a8:6d:58:5b:f3, ethertype IPv4 (0x0800), length 320: (tos 0x0, ttl 64, id 0, offset 0, flags [none], proto UDP (17), length 306)
    192.168.2.1.67 > 192.168.2.22.68: BOOTP/DHCP, Reply, length 278, xid 0x5e7944f, Flags [none]
      Client-IP 192.168.2.22
      Your-IP 192.168.2.22
      Client-Ethernet-Address 68:a8:6d:58:5b:f3
      Vendor-rfc1048 Extensions
        Magic Cookie 0x63825363
        DHCP-Message Option 53, length 1: ACK
        Server-ID Option 54, length 4: 192.168.2.1
        Lease-Time Option 51, length 4: 86400
        Subnet-Mask Option 1, length 4: 255.255.255.0
        Default-Gateway Option 3, length 4: 192.168.2.1
        Domain-Name-Server Option 6, length 8: 192.168.2.1,142.166.166.166

If when running that tcpdump command, and you see a BOOTP/DHCP Offer or Ack(Nack), that will be from a DHCP server, and the server's MAC address will be right after the timestamp on the first line.

So the (valid) DHCP server here has MAC address a8:39:44:96:fa:b8`.

Using one of many MAC address lookup tools on the web I see this MAC belongs to A8:39:44 Actiontec Electronics, Inc which is my router.

In order to catch rogue DHCP server packets as they happen, I would have to leave this tcpdump process running in terminal window:

tcpdump -i en0 -nev udp src port 67 and not ether host a8:39:44:96:fa:b8

This will only show me DHCP server responses from hosts other than my valid DHCP server, as long as the process is running in its own window.

The following command will run in the background until 100 packets are captured, appending any rogue DHCP server messages to the file /tmp/rogue. Again, the MAC address of your valid DHCP server has to be used in the appropriate place, as well as the interface descriptor on your system.

tcpdump -U -i en0 -c 100 -nev udp src port 67 and not ether host a8:39:44:96:fa:b8 >> /tmp/rogue 2>&1 &

`

4
  • +1 used to use this all the time on customer's VoIP networks to help them spot rogue DHCP servers that were messing up their service.
    – MaQleod
    May 7 '14 at 1:44
  • Yeah, I had to use it for finding rogue DHCP servers on early deployments of cable modem networks until we figured out how to filter outbound UDP port 67 on each kind of pre-DOCSIS cable modems on our network. Even after we identified the rogue, it wasn't always possible to identify what modem it was behind. In severe cases, we had to shut off each node until the rogue stopped, and then isolate what amp on that node to narrow the number of truck rolls the MSO would make. Fun times. Long live DOCSIS. May 7 '14 at 15:33
  • any simple solution, because I want to check frequently with bash
    – Steve
    May 9 '14 at 12:41
  • If you leave this tcpdump command running, it will display DHCP server packets sent out by a MAC address that is different than your DHCP server's. Of course, you'll need to supply the address in the indicated place... I'll place it in my answer presently, to get better formatting. May 9 '14 at 20:55
0

Given enough time, it might be possible to do the detection passively: Recall that an initializing client sends out a DHCPDISCOVER broadcast. If a dhcpserver is available, it will optain an offer and then emit (again as broadcast!) a DHCPREQUEST. Thus

  • if you receive a DHCPREQUEST broadcast, there is a dhcp server
  • if you receive DHCPDISCOVER, but no DHCPREQUEST within a second or two (to account for remote dhcp servers via dhcp relay), there is no dhcp server
  • if you do not even receive DHCPDISCOVER, you have to wait longer - or try to actively find a dhcp server (e.g. as per Kenned's method)

The success of the first two points deopends on the number of other hosts newly attached / booted to the network and also the lease times.

0

You could try creating an alias device and using a dhcp client in testing mode, where it prints out any response, without actually reconfiguring the interface:

ifconfig eth0:1 up
dhclient -w -n eth0:1

I only have access to a debian box, so if you have another dhcp implementation, the dry-run option might be different, like for dhcpcd:

dhcpcd -T eth0:1

I used to have a cron script run with something like this, that would alert the admin (me!) about rogue dhcp servers.

1
  • 1
    when call ifconfig eth0:1 up the output SIOCSIFFLAGS: Cannot assign requested address
    – Steve
    May 13 '14 at 11:37
0

If you have scapy available, it is a (even scriptable) command-line simple solution. Just this code (run inside scapy interface or script it via .py file, as desired) should perform the DHCP request and return the number of answers (equals to DHCP different servers):

conf.checkIPaddr = False
fam,hw = get_if_raw_hwaddr(conf.iface)
dhcp_discover = Ether(dst="ff:ff:ff:ff:ff:ff")/IP(src="0.0.0.0",dst="255.255.255.255")/UDP(sport=68,dport=67)/BOOTP(chaddr=hw)/DHCP(options=[("message-type","discover"),"end"])
ans, unans = srp(dhcp_discover, multi=True)      # Press CTRL-C after several seconds
ans.summary()

The Scapy way is a good one since it is available on many platforms and embedded devices (like OpenWRT) and accepts scripting.

The downside: you need python installed (well... widely available too).

No need for Bash on this solution.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.