Load Balanced DHCP Servers
Jan Hugo Prins
jhp at jhprins.org
Mon Nov 6 15:15:37 UTC 2006
Hi,
I'm currently looking into the option of creating 2 loadbalanced DHCP
servers for our production environment and I have setup a little test
network. At first sight it looks as if everything works ok, but after
checking some logs etc I run into some strange things:
The first thing I see is that in the RFC they are talking about several
different states that either one of the servers can be in, and in the
logs I only see about 3 different states and both server sometimes go
from normal to normal etc. A little strange.
A little bit of my logs:
Nov 3 10:40:50 dhcp1 dhcpd: Internet Systems Consortium DHCP Server
V3.0.3-RedHat
Nov 3 10:40:50 dhcp1 dhcpd: Copyright 2004-2005 Internet Systems
Consortium.
Nov 3 10:40:50 dhcp1 dhcpd: All rights reserved.
Nov 3 10:40:50 dhcp1 dhcpd: For info, please visit
http://www.isc.org/sw/dhcp/
Nov 3 10:40:50 dhcp1 dhcpd: Wrote 100 leases to leases file.
Nov 3 10:40:50 dhcp1 dhcpd: Listening on
LPF/eth1/00:0c:29:18:cf:ed/192.168.102/24
Nov 3 10:40:50 dhcp1 dhcpd: Sending on
LPF/eth1/00:0c:29:18:cf:ed/192.168.102/24
Nov 3 10:40:50 dhcp1 dhcpd: Sending on Socket/fallback/fallback-net
Nov 3 10:40:51 dhcp1 dhcpd: failover peer dhcp: I move from normal to
startup
Nov 3 10:40:51 dhcp1 dhcpd: failover peer dhcp: peer moves from normal
to communications-interrupted
Nov 3 10:40:51 dhcp1 dhcpd: failover peer dhcp: I move from startup to
normal
Nov 3 10:40:51 dhcp1 dhcpd: failover peer dhcp: peer moves from
communications-interrupted to normal
Nov 3 10:40:51 dhcp1 dhcpd: pool 9e9aa20 192.168.102/24 total 191 free
104 backup 85 lts -9
Nov 3 10:43:16 dhcp1 dhcpd: peer dhcp: disconnected
Nov 3 10:43:16 dhcp1 dhcpd: failover peer dhcp: I move from normal to
communications-interrupted
Nov 3 10:48:32 dhcp1 dhcpd: DHCPREQUEST for 192.168.102.101 from
00:17:31:4d:36:9b via eth1
Nov 3 10:48:32 dhcp1 dhcpd: DHCPACK on 192.168.102.101 to
00:17:31:4d:36:9b via eth1
Nov 3 10:49:04 dhcp1 dhcpd: failover peer dhcp: peer moves from normal
to normal
Nov 3 10:49:04 dhcp1 dhcpd: failover peer dhcp: I move from
communications-interrupted to normal
Nov 3 10:49:04 dhcp1 dhcpd: pool 9e9aa20 192.168.102/24 total 191 free
104 backup 85 lts -9
Nov 3 11:12:41 dhcp1 dhcpd: peer dhcp: disconnected
Nov 3 11:12:41 dhcp1 dhcpd: failover peer dhcp: I move from normal to
communications-interrupted
And on host 2:
Nov 3 11:43:42 dhcp2 dhcpd: Internet Systems Consortium DHCP Server
V3.0.4-RedHat
Nov 3 11:43:42 dhcp2 dhcpd: Copyright 2004-2006 Internet Systems
Consortium.
Nov 3 11:43:42 dhcp2 dhcpd: All rights reserved.
Nov 3 11:43:42 dhcp2 dhcpd: For info, please visit
http://www.isc.org/sw/dhcp/
Nov 3 11:43:42 dhcp2 dhcpd: Wrote 100 leases to leases file.
Nov 3 11:43:42 dhcp2 dhcpd: Listening on
LPF/eth1/00:0c:29:a9:44:66/192.168.102/24
Nov 3 11:43:42 dhcp2 dhcpd: Sending on
LPF/eth1/00:0c:29:a9:44:66/192.168.102/24
Nov 3 11:43:42 dhcp2 dhcpd: Sending on Socket/fallback/fallback-net
Nov 3 11:43:42 dhcp2 dhcpd: failover peer dhcp: I move from
communications-interrupted to startup
Nov 3 11:43:50 dhcp2 dhcpd: Internet Systems Consortium DHCP Server
V3.0.4-RedHat
Nov 3 11:43:50 dhcp2 dhcpd: Copyright 2004-2006 Internet Systems
Consortium.
Nov 3 11:43:50 dhcp2 dhcpd: All rights reserved.
Nov 3 11:43:50 dhcp2 dhcpd: For info, please visit
http://www.isc.org/sw/dhcp/
Nov 3 11:43:50 dhcp2 dhcpd: Wrote 100 leases to leases file.
Nov 3 11:43:50 dhcp2 dhcpd: Listening on
LPF/eth1/00:0c:29:a9:44:66/192.168.102/24
Nov 3 11:43:50 dhcp2 dhcpd: Sending on
LPF/eth1/00:0c:29:a9:44:66/192.168.102/24
Nov 3 11:43:50 dhcp2 dhcpd: Sending on Socket/fallback/fallback-net
Nov 3 11:43:50 dhcp2 dhcpd: failover peer dhcp: I move from
communications-interrupted to startup
Nov 3 11:43:56 dhcp2 dhcpd: failover peer dhcp: I move from startup to
communications-interrupted
Nov 3 11:43:56 dhcp2 dhcpd: failover peer dhcp: peer moves from normal
to normal
Nov 3 11:43:56 dhcp2 dhcpd: failover peer dhcp: I move from
communications-interrupted to normal
Nov 3 11:43:56 dhcp2 dhcpd: pool 952f9c0 192.168.102/24 total 191 free
104 backup 85 lts 9
Then I see in the leases file something strange. On host 1 I see the
following:
failover peer "dhcp" state {
my state communications-interrupted at 5 2006/11/03 10:14:53;
partner state normal at 1 2006/10/30 17:13:45;
mclt 600;
}
failover peer "dhcp" state {
my state normal at 5 2006/11/03 10:43:56;
partner state normal at 1 2006/10/30 17:13:45;
mclt 600;
}
On host 2 I see the following:
failover peer "dhcp" state {
my state normal at 5 2006/11/03 10:14:43;
partner state normal at 1 2006/10/30 17:11:41;
}
The "partner state normal at" time never seems to change. Whenever I
look in the leases file of either one of the servers, they allways
report the partner state normal time as being the time that the server
got started the first time.
This seems a little strange to me.
Finally my config files that I put on the servers:
Server1:
ddns-update-style none;
one-lease-per-client true;
option domain-name "testlan.local";
option domain-name-servers 192.168.102.1;
option subnet-mask 255.255.255.0;
authoritative;
failover peer "dhcp" {
primary;
address 192.168.102.1;
port 647;
peer address 192.168.102.2;
peer port 847;
max-response-delay 60;
max-unacked-updates 10;
mclt 600;
split 128;
load balance max seconds 3;
}
include "/etc/dhcpd.master";
Server2:
ddns-update-style none;
one-lease-per-client true;
option domain-name "testlan.local";
option domain-name-servers 192.168.102.1;
option subnet-mask 255.255.255.0;
authoritative;
failover peer "dhcp" {
secondary;
address 192.168.102.2;
port 847;
peer address 192.168.102.1;
peer port 647;
max-response-delay 60;
max-unacked-updates 10;
}
include "/etc/dhcpd.master";
The dhcpd.master file is the same on both servers:
subnet 192.168.102.0 netmask 255.255.255.0 {
pool {
failover peer "dhcp";
range 192.168.102.10 192.168.102.200;
deny dynamic bootp clients;
}
option routers 192.168.102.1;
max-lease-time 86400;
default-lease-time 86400;
}
I hope someone can declare the things that I see for me. Am I doing
something completly wrong? Is the behaviour normal or did I run into
some strange bugs?
Please enlighten me.
Greetings,
Jan Hugo Prins
More information about the dhcp-users
mailing list