Highly Available DHCP Server on FreeBSD

Today I would like to share a highly available DHCP server setup on FreeBSD system, but it should be similarly simple on other UNIX and Unix-like systems. I will use the most obvious choice here – the Internet Systems Consortium implementation – ISC DHCP server – available in the FreeBSD Ports and packages as well.

ISC

Since some time ISC is developing a new DHCP server – Kea – with which they intend to eventually replace the ISC DHCP in most server implementations. They also recommend that new implementers consider using Kea instead ISC DHCP and implement ISC DHCP only if Kea does not meet their needs. Kea currently does not include either client or relay for example. Maybe I will make an UPDATE to this post or a separate article some time.

Also Kea got high availability mode just a month ago so if I would be writing this article little earlier then such setup would not be possible with Kea. It also shows how young Kea implementation is thus I would stick to ISC DHCP server for now and ‘watch’ Kea development for the future.

Architecture

Below is the POOR MAN’S ASCII ARCHITECT diagram showing our ISC DHCP setup.

  +-------------+              +-------------+
  | {primary}   |              | {secondary} |
  | DHCPs1      | ==== HA ==== | DHCPs2      |
  | 10.0.10.251 |              | 10.0.10.252 |
  +-------------+              +-------------+
                 \            /
  +------------------------------------------+
  | ADDRESS POOL  10.0.10.x/24  ADDRESS POOL |
  +------------------------------------------+
              \                  /
               +----------------+
               | {DHCP CLIENTS} |
               +----------------+

The setup of each DHCP server node is very simple. Its FreeBSD 11.2-RELEASE installed on a 4 GB GPT partition using UFS for the / filesystem and only 666 MB are used as shown below.

root@DHCPs1:/ # uname -v
FreeBSD 11.2-RELEASE #0 r335510: Fri Jun 22 04:32:14 UTC 2018     root@releng2.nyi.freebsd.org:/usr/obj/usr/src/sys/GENERIC 

root@DHCPs1:/ # gpart show
=>     40  8388528  ada0  GPT  (4.0G)
       40     1024     1  freebsd-boot  (512K)
     1064  8386560     2  freebsd-ufs  (4.0G)
  8387624      944        - free -  (472K)

root@DHCPs1:/ # du -smc * | sort -n
0       sys
1       COPYRIGHT
1       dev
1       entropy
1       libexec
1       media
1       mnt
1       net
1       proc
1       root
1       tmp
2       bin
4       etc
7       sbin
8       var
10      rescue
12      lib
128     boot
499     usr
666     total

The 128 MB of RAM is enough for small amount of clients. There is still 32 MB free memory along with 32 MB of Inactive and Buffered memory that can be swapped out. Not to mention that each getty process takes about 2 MB ram and instead of 8 you just only need 1 of them. In other words you would be able to run it even with as low as 64 MB of RAM.

root@DHCPs1:~ # top -b -o res
last pid: 15205;  load averages:  0.13,  0.25,  0.29  up 0+07:39:11    20:03:48
16 processes:  2 running, 14 sleeping

Mem: 1688K Active, 30M Inact, 26M Wired, 3800K Buf, 32M Free
Swap:


  PID USERNAME    THR PRI NICE   SIZE    RES STATE    TIME    WCPU COMMAND
38897 dhcpd         1  20    0 16424K 10724K select   0:00   0.00% dhcpd
30199 root          1  20    0 13160K  8036K RUN      0:00   0.00% sshd
15106 root          1  28    0 12848K  7136K select   0:00   0.00% sshd
53100 root          1  20    0  9180K  5040K select   0:02   0.00% devd
31079 root          1  20    0  7412K  3640K pause    0:00   0.00% csh
15205 root          1  20    0  7916K  3060K RUN      0:00   0.00% top
15960 root          1  20    0  6464K  2480K nanslp   0:00   0.00% cron
69084 root          1  20    0  6412K  2364K select   0:01   0.00% syslogd
28412 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
28188 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
28504 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
28972 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
29736 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
29080 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
30106 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty
29392 root          1  52    0  6408K  2124K ttyin    0:00   0.00% getty



The /etc/rc.conf file for DHCP nodes DHCPs1 and DHCPs2 is the same (besides hostname and address).

root@DHCPs1:/ # cat /etc/rc.conf
hostname=DHCPs1
ifconfig_em0="inet 10.0.10.251/24 up"
sshd_enable=YES
sendmail_enable=NONE
clear_tmp_enable=YES
syslogd_flags="-ss"
dumpdev=NO

The /etc/sysctl.conf and /boot/loader.conf files modifications are not needed.

Now you will have to install the ISC DHCP server, as the current version is 4.4.x the package will be named accordingly – isc-dhcp44-server – lets add it using the pkg(8) command.

root@DHCPs1:/ # pkg update -f -y
The package management tool is not yet installed on your system.
Bootstrapping pkg from pkg+http://pkg.FreeBSD.org/FreeBSD:11:amd64//quarterly, please wait...
Verifying signature with trusted certificate pkg.freebsd.org.2013102301... done
[nextcloud] Installing pkg-1.10.5...
[nextcloud] Extracting pkg-1.10.5: 100%
Updating FreeBSD repository catalogue...
pkg: Repository FreeBSD load error: access repo file(/var/db/pkg/repo-FreeBSD.sqlite) failed: No such file or directory
[nextcloud] Fetching meta.txz: 100%    944 B   0.9kB/s    00:01
[nextcloud] Fetching packagesite.txz: 100%    6 MiB 530.8kB/s    00:12
Processing entries: 100%
FreeBSD repository update completed. 31134 packages processed.
All repositories are up to date.
root@DHCPs1:/ # echo ?
0
root@DHCPs1:/ #

Now lets install isc-dhcp44-server package.

root@DHCPs1:/ # pkg install isc-dhcp44-server
Updating FreeBSD repository catalogue...
FreeBSD repository is up to date.
All repositories are up to date.
Checking integrity... done (0 conflicting)
The following 1 package(s) will be affected (of 0 checked):

New packages to be INSTALLED:
        isc-dhcp44-server: 4.4.1_3 [FreeBSD]

Number of packages to be installed: 1

The process will require 6 MiB more space.

Proceed with this action? [y/N]: y
[1/1] Installing isc-dhcp44-server-4.4.1_3...
===> Creating groups.
Creating group 'dhcpd' with gid '136'.
===> Creating users
Creating user 'dhcpd' with uid '136'.
[1/1] Extracting isc-dhcp44-server-4.4.1_3: 100%
Message from isc-dhcp44-server-4.4.1_3:

****  To setup dhcpd, please edit /usr/local/etc/dhcpd.conf.

****  This port installs the dhcp daemon, but doesn't invoke dhcpd by default.
      If you want to invoke dhcpd at startup, add these lines to /etc/rc.conf:

            dhcpd_enable="YES"                          # dhcpd enabled?
            dhcpd_flags="-q"                            # command option(s)
            dhcpd_conf="/usr/local/etc/dhcpd.conf"      # configuration file
            dhcpd_ifaces=""                             # ethernet interface(s)
            dhcpd_withumask="022"                       # file creation mask

****  If compiled with paranoia support (the default), the following rc.conf
      options are also supported:

            dhcpd_chuser_enable="YES"           # runs w/o privileges?
            dhcpd_withuser="dhcpd"              # user name to run as
            dhcpd_withgroup="dhcpd"             # group name to run as
            dhcpd_chroot_enable="YES"           # runs chrooted?
            dhcpd_devfs_enable="YES"            # use devfs if available?
            dhcpd_rootdir="/var/db/dhcpd"       # directory to run in
            dhcpd_includedir=""       # directory with config-
                                                  files to include

****  WARNING: never edit the chrooted or jailed dhcpd.conf file but
      /usr/local/etc/dhcpd.conf instead which is always copied where
      needed upon startup.

Now update the pkg(8) repository data and install the isc-dhcp44-server package on DHCPs2 node.

The configuration uses single network segment 10.0.10.0/24 for the clients in the range of 10-250 values in the last octet. The parameter split 128 will split the load equally between DHCP server nodes. As this is just example, we will use 1.1.1.1 and 9.9.9.9 DNS servers and ‘domain.com‘ domain. For the record, the split 128 parameter is set only on the primary node – DHCPs1 in our case. As the man dhcpd.conf page suggests we will “use the same master configuration file for both servers, and have a separate file that contains the peer declaration and includes the master file.” as “This will help you to avoid configuration mismatches.”

root@DHCPs1:/ # cat /usr/local/etc/dhcpd.conf
# CORE
failover peer "ha-dhcp" {
  primary;
  address 10.0.10.251;
  port 678;
  peer address 10.0.10.252;
  peer port 678;
  max-response-delay 60;
  max-unacked-updates 10;
  mclt 3600;
  split 128;
  load balance max seconds 3;
}

include "/usr/local/etc/dhcpd.conf.SHARED";
root@DHCPs1:/ # cat /usr/local/etc/dhcpd.conf.SHARED
# CLIENTS
subnet 10.0.10.0 netmask 255.255.255.0 {
  default-lease-time         604800;
  max-lease-time             604800;
  option routers             10.0.10.254;
  option broadcast-address   10.0.10.255;
  option subnet-mask         255.255.255.0;
  option domain-search       "domain.com";
  option domain-name-servers 1.1.1.1,9.9.9.9;

  pool {
    failover peer "ha-dhcp";
    range 10.0.10.10 10.0.10.250;
  }
}

… and the secondary node.

root@DHCPs2:~ # cat /usr/local/etc/dhcpd.conf
# CORE
failover peer "ha-dhcp" {
  secondary;
  address 10.0.10.252;
  port 678;
  peer address 10.0.10.251;
  peer port 678;
  max-response-delay 60;
  max-unacked-updates 10;
  mclt 3600;
  load balance max seconds 3;
}

include "/usr/local/etc/dhcpd.conf.SHARED";
root@DHCPs2:/ # cat /usr/local/etc/dhcpd.conf.SHARED
# CLIENTS
subnet 10.0.10.0 netmask 255.255.255.0 {
  default-lease-time         604800;
  max-lease-time             604800;
  option routers             10.0.10.254;
  option broadcast-address   10.0.10.255;
  option subnet-mask         255.255.255.0;
  option domain-search       "domain.com";
  option domain-name-servers 1.1.1.1,9.9.9.9;

  pool {
    failover peer "ha-dhcp";
    range 10.0.10.10 10.0.10.250;
  }
}

The /usr/local/etc/dhcpd.conf.SHARED file is identical on both nodes.

Now lets start the DHCP server on both nodes.

root@DHCPs1:~ # sysrc dhcpd_enable=YES
dhcpd_enable:  -> YES
root@DHCPs1:~ # service isc-dhcpd start
Starting dhcpd.
Internet Systems Consortium DHCP Server 4.4.1
Copyright 2004-2018 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/
Config file: /usr/local/etc/dhcpd.conf
Database file: /var/db/dhcpd/dhcpd.leases
PID file: /var/run/dhcpd/dhcpd.pid
Wrote 122 leases to leases file.
Listening on BPF/em0/08:00:27:3c:ab:c8/10.0.10.0/24
Sending on   BPF/em0/08:00:27:3c:ab:c8/10.0.10.0/24
Sending on   Socket/fallback/fallback-net
failover peer ha-dhcp: I move from normal to startup

… and the same on secondary node.

root@DHCPs2:~ # sysrc dhcpd_enable=YES
dhcpd_enable:  -> YES
root@DHCPs2:~ # service isc-dhcpd onestart
Starting dhcpd.
Internet Systems Consortium DHCP Server 4.4.1
Copyright 2004-2018 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/
Config file: /usr/local/etc/dhcpd.conf
Database file: /var/db/dhcpd/dhcpd.leases
PID file: /var/run/dhcpd/dhcpd.pid
Wrote 122 leases to leases file.
Listening on BPF/em0/08:00:27:de:9b:3d/10.0.10.0/24
Sending on   BPF/em0/08:00:27:de:9b:3d/10.0.10.0/24
Sending on   Socket/fallback/fallback-net
failover peer ha-dhcp: I move from communications-interrupted to startup

Now, as the both nodes for the highly available DHCP server are started, lets try to get some DHCP lease on the DHCP client – DHCPc in our example.

root@DHCPc:~ # dhclient em0
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPACK from 10.0.10.251
bound to 10.0.10.131 -- renewal in 302119 seconds.
root@DHCPc:~ # ifconfig em0
em0: flags=8843 metric 0 mtu 1500
        options=9b
        ether 08:00:27:d9:45:96
        hwaddr 08:00:27:d9:45:96
        inet 10.0.10.131 netmask 0xffffff00 broadcast 10.0.10.255
        nd6 options=29
        media: Ethernet autoselect (1000baseT )
        status: active

We can see that the DHCP client –Β DHCPc – got the 10.0.10.131 address.

We can of course set permanent address for it with the host option in the /usr/local/etc/dhcpd.conf.SHARED config file as show below.

The needed ‘addon’ is shown below.

  group
  {
    host DHCPc {
      hardware ethernet 08:00:27:d9:45:96;
      fixed-address 10.0.10.9;
    }
  }

It needs to be added on both nodes in the /usr/local/etc/dhcpd.conf.SHARED config file, here is how the new shared config file would look like.

root@DHCPs1:~ # cat /usr/local/etc/dhcpd.conf.SHARED
# CLIENTS
subnet 10.0.10.0 netmask 255.255.255.0 {
  default-lease-time         604800;
  max-lease-time             604800;
  option routers             10.0.10.254;
  option broadcast-address   10.0.10.255;
  option subnet-mask         255.255.255.0;
  option domain-search       "domain.com";
  option domain-name-servers 1.1.1.1,9.9.9.9;

  group
  {
    host DHCPc {
      hardware ethernet 08:00:27:d9:45:96;
      fixed-address 10.0.10.9;
    }
  }

  pool {
    failover peer "ha-dhcp";
    range 10.0.10.10 10.0.10.250;
  }
}

Now copy the /usr/local/etc/dhcpd.conf.SHARED file to the second node.

Lets try again to get the address from the same DHCP client.

root@DHCPc:~ # pkill dhclient
root@DHCPc:~ # service netif restart
root@DHCPc:~ # dhclient em0
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPACK from 10.0.10.252
bound to 10.0.10.131 -- renewal in 1665 seconds.
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPNAK from 10.0.10.252
DHCPDISCOVER on em0 to 255.255.255.255 port 67 interval 3
DHCPOFFER from 10.0.10.251
DHCPOFFER from 10.0.10.252
DHCPOFFER already seen.
DHCPREQUEST on em0 to 255.255.255.255 port 67
DHCPACK from 10.0.10.252
bound to 10.0.10.9 -- renewal in 302400 seconds.
root@DHCPc:~ # ifconfig em0
em0: flags=8843 metric 0 mtu 1500
        options=9b
        ether 08:00:27:d9:45:96
        hwaddr 08:00:27:d9:45:96
        inet 10.0.10.9 netmask 0xffffff00 broadcast 10.0.10.255
        nd6 options=29
        media: Ethernet autoselect (1000baseT )
        status: active

Now we got the permanent 10.0.10.9 address.

You can now experiment with these values in the /etc/rc.conf file:

  • dhcpd_flags
  • dhcpd_ifaces
  • dhcpd_withumask
  • dhcpd_chuser_enable
  • dhcpd_withuser
  • dhcpd_withgroup
  • dhcpd_chroot_enable
  • dhcpd_devfs_enable
  • dhcpd_rootdir
  • dhcpd_includedirnclude

… with the all other possible options from the man dhcpd.conf page πŸ™‚

EOF
Advertisements

2 thoughts on “Highly Available DHCP Server on FreeBSD

  1. Pingback: Highly available DHCP server (FreeBSD) | 0ddn1x: tricks with *nix

  2. Pingback: In Other BSDs for 2018/09/15 – DragonFly BSD Digest

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s