discrepancy/differences in dhcpd.leases (active leases) on failover pair
Oscar Ricardo Silva
oscars at mail.utexas.edu
Fri May 8 16:39:29 UTC 2009
We've noticed some discrepancies/differences in active leases on a
failover pair and were wondering if it's really an issue or just a side
effect of something else. I should mention that we have changed the
split statement to favor one of the servers over the others. Here's
dhcpd.conf for both servers:
Primary:
option domain-name-servers 128.83.185.41, 128.83.185.40 ;
default-lease-time 86400;
max-lease-time 86400;
one-lease-per-client true;
ddns-update-style ad-hoc;
ddns-updates off;
authoritative;
failover peer "pub-dhcp" {
primary;
address 172.16.200.34;
port 520;
peer address 172.16.201.34;
peer port 520;
max-response-delay 60;
max-unacked-updates 10;
mclt 120;
split 255;
load balance max seconds 5;
}
include "/dhcpd/om.conf";
include "/dhcpd/public.dhcpd.conf";
Secondary:
option domain-name-servers 128.83.185.41, 128.83.185.40 ;
default-lease-time 86400;
max-lease-time 86400;
one-lease-per-client true;
ddns-update-style ad-hoc;
ddns-updates off;
authoritative;
failover peer "pub-dhcp" {
secondary;
address 172.16.201.34;
port 520;
peer address 172.16.200.34;
peer port 520;
max-response-delay 60;
max-unacked-updates 10;
load balance max seconds 5;
}
include "/dhcpd/om.conf";
include "/dhcpd/public.dhcpd.conf";
In looking at the primary's dhcpd.leases I found the following:
lease 128.62.100.104 {
starts 5 2009/05/08 14:01:45;
ends 5 2009/05/08 15:01:45;
tstp 5 2009/05/08 15:31:45;
tsfp 5 2009/05/08 15:31:45;
atsfp 5 2009/05/08 15:31:45;
cltt 5 2009/05/08 14:01:45;
binding state active;
next binding state expired;
hardware ethernet 00:1f:5a:4b:35:5b;
option agent.circuit-id 0:0:0:0;
option agent.remote-id 0:1d:e5:8f:7f:40;
client-hostname "Oscar";
}
lease 128.62.100.118 {
starts 5 2009/05/08 13:45:43;
ends 5 2009/05/08 14:45:43;
tstp 5 2009/05/08 15:15:43;
tsfp 5 2009/05/08 15:15:43;
cltt 5 2009/05/08 13:45:43;
binding state expired;
next binding state free;
hardware ethernet 00:23:13:53:9b:84;
option agent.circuit-id 0:0:0:0:0:0:0:0;
option agent.remote-id 0:19:a9:56:fd:e0:0:0:0:0:0:0;
client-hostname "macbook-2";
}
while the secondary has:
lease 128.62.100.104 {
starts 5 2009/05/08 14:01:45;
ends 5 2009/05/08 14:25:34;
tstp 5 2009/05/08 14:25:34;
tsfp 5 2009/05/08 14:25:34;
atsfp 5 2009/05/08 14:25:34;
cltt 5 2009/05/08 14:01:45;
binding state free;
hardware ethernet 00:1f:5a:4b:35:5b;
}
lease 128.62.100.118 {
starts 5 2009/05/08 13:45:43;
ends 5 2009/05/08 14:45:43;
tstp 5 2009/05/08 14:45:43;
tsfp 5 2009/05/08 15:15:43;
atsfp 5 2009/05/08 15:15:43;
cltt 5 2009/05/08 13:45:43;
binding state free;
hardware ethernet 00:23:13:53:9b:84;
}
Here are the associated logs for these machines and IP addresses. The
secondary's logs show that the request was load balanced to the peer.
128.62.100.118
Primary:
May 8 08:44:42 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 192.168.1.2 from
00:23:13:53:9b:84 via 172.17.32.35: wrong network.
May 8 08:44:42 NOCA-PUB-DHCP dhcpd: DHCPNAK on 192.168.1.2 to
00:23:13:53:9b:84 via 172.17.32.35
May 8 08:44:43 NOCA-PUB-DHCP dhcpd: DHCPDISCOVER from 00:23:13:53:9b:84
via 172.17.32.35
May 8 08:44:44 NOCA-PUB-DHCP dhcpd: DHCPOFFER on 128.62.100.118 to
00:23:13:53:9b:84 (macbook-2) via 172.17.32.35
May 8 08:44:44 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 128.62.100.118
(172.16.200.34) from 00:23:13:53:9b:84 (macbook-2) via 172.17.32.35
May 8 08:44:44 NOCA-PUB-DHCP dhcpd: DHCPACK on 128.62.100.118 to
00:23:13:53:9b:84 (macbook-2) via 172.17.32.35
May 8 08:45:00 NOCA-PUB-DHCP dhcpd: DHCPINFORM from 128.62.100.118 via
172.17.32.35
May 8 08:45:00 NOCA-PUB-DHCP dhcpd: DHCPACK to 128.62.100.118
(00:23:13:53:9b:84) via eth0
May 8 08:45:43 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 128.62.100.118 from
00:23:13:53:9b:84 (macbook-2) via 172.17.32.35
May 8 08:45:43 NOCA-PUB-DHCP dhcpd: DHCPACK on 128.62.100.118 to
00:23:13:53:9b:84 (macbook-2) via 172.17.32.35
Secondary:
May 8 08:44:42 NOCB-PUB-DHCP dhcpd: DHCPREQUEST for 192.168.1.2 from
00:23:13:53:9b:84 via 172.17.32.35: wrong network.
May 8 08:44:42 NOCB-PUB-DHCP dhcpd: DHCPNAK on 192.168.1.2 to
00:23:13:53:9b:84 via 172.17.32.35
May 8 08:44:43 NOCB-PUB-DHCP dhcpd: DHCPDISCOVER from 00:23:13:53:9b:84
via 172.17.32.35: load balance to peer pub-dhcp
128.62.100.104
Primary:
May 8 09:00:35 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 10.0.1.6 from
00:1f:5a:4b:35:5b via 172.17.32.44: wrong network.
May 8 09:00:35 NOCA-PUB-DHCP dhcpd: DHCPNAK on 10.0.1.6 to
00:1f:5a:4b:35:5b via 172.17.32.44
May 8 09:00:35 NOCA-PUB-DHCP dhcpd: DHCPDISCOVER from 00:1f:5a:4b:35:5b
via 172.17.32.44
May 8 09:00:36 NOCA-PUB-DHCP dhcpd: DHCPOFFER on 128.62.100.104 to
00:1f:5a:4b:35:5b (Oscar) via 172.17.32.44
May 8 09:00:37 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 128.62.100.104
(172.16.200.34) from 00:1f:5a:4b:35:5b (Oscar) via 172.17.32.44
May 8 09:00:37 NOCA-PUB-DHCP dhcpd: DHCPACK on 128.62.100.104 to
00:1f:5a:4b:35:5b (Oscar) via 172.17.32.44
May 8 09:01:45 NOCA-PUB-DHCP dhcpd: DHCPREQUEST for 128.62.100.104 from
00:1f:5a:4b:35:5b (Oscar) via 172.17.32.44
May 8 09:01:45 NOCA-PUB-DHCP dhcpd: DHCPACK on 128.62.100.104 to
00:1f:5a:4b:35:5b (Oscar) via 172.17.32.44
Secondary:
May 8 09:00:35 NOCB-PUB-DHCP dhcpd: DHCPREQUEST for 10.0.1.6 from
00:1f:5a:4b:35:5b via 172.17.32.44: wrong network.
May 8 09:00:35 NOCB-PUB-DHCP dhcpd: DHCPNAK on 10.0.1.6 to
00:1f:5a:4b:35:5b via 172.17.32.44
May 8 09:00:35 NOCB-PUB-DHCP dhcpd: DHCPDISCOVER from 00:1f:5a:4b:35:5b
via 172.17.32.44: load balance to peer pub-dhcp
Don't the the peers update each other and consequently update their
dhcpd.leases so that they have a consistent state? For a the range of
times, I found hundreds of differences in the number of active leases
each server in the peer relationship reported.
Oscar
More information about the dhcp-users
mailing list