Linux DHCP clients dropping off

["Adam Williams" <awilliam@xxxxxxxxxxxxx>]

I have RedHat 4.2 machines running as DHCP clients, and a Redhat 5.0 machine
(192.168.1.9) running as the DHCP server (and NIS and DNS as well).  Most
clients are seperated by two routers hops,  the first router (on their subnet)
is configured as a dhcp relay to 192.168.1.9.  The problem is quite infrequent,
say once or twice a week a client drops out,  out of 75 - 100 machines on the
network 24/7.  The ethernet interface on the router (far side) shows no
significant errors, and WAN circuit utilization is fairly low.  Clients on the
local subnet (connected to the backbone via a switch) have yet to exhibit this
behavior.  ypbind seems to be the first thing to notice the failure, and I will
be unable to ping the machine.

This is the end of the log on the client before rebooting (/var/log/messages):
Oct 21 01:02:12 eds01333 PAM_pwdb[20169]:  (su) session opened for user nobody
by (uid=99)~
Oct 21 01:02:44 eds01333 PAM_pwdb[20169]: (su) session closed for user nobody~
Oct 21 03:21:01 eds01333 dhcpcd[102]: got in RENEWING state~
Oct 21 03:21:01 eds01333 dhcpcd[102]: got in BOUND state~
Oct 21 21:19:21 eds01333 dhcpcd[102]: got in RENEWING state~
Oct 21 21:19:22 eds01333 dhcpcd[102]: got in BOUND state~
Oct 22 01:02:12 eds01333 PAM_pwdb[24661]:  (su) session opened for user nobody
by (uid=99)~
Oct 22 01:02:43 eds01333 PAM_pwdb[24661]: (su) session closed for user nobody~
Oct 22 15:17:42 eds01333 dhcpcd[102]: got in RENEWING state~
Oct 22 15:17:42 eds01333 dhcpcd[102]: got in BOUND state~
Oct 22 15:45:56 eds01333 ypbind[203]: 192.168.1.9: RPC: Timed out~
Oct 22 15:46:56 eds01333 ypbind[203]:  clnt_create for server 192.168.1.9
failed~
Oct 22 15:48:57 eds01333 ypbind[203]:  clnt_create for server 192.168.1.9
failed~
Oct 22 15:50:57 eds01333 ypbind[203]:  clnt_create for server 192.168.1.9
failed~
Oct 22 15:52:57 eds01333 ypbind[203]:  clnt_create for server 192.168.1.9
failed~
Oct 22 15:54:57 eds01333 ypbind[203]:  clnt_create for server 192.168.1.9
failed~
Oct 22 15:54:37 eds01333 syslogd 1.3-3: restart.~



The following command on the server produces the subsequent output:~
grep dhcp /var/log/messages | grep "11\.102"~

Oct 20 09:18:47 littleboy dhcpd:  DHCPREQUEST for 192.168.11.102 from
0:40:5:40:3f:1e via eth0~
Oct 20 09:18:47 littleboy dhcpd:  DHCPACK on 192.168.11.102 to 0:40:5:40:3f:1e
via eth0~
Oct 21 03:16:49 littleboy dhcpd:  DHCPREQUEST for 192.168.11.102 from
0:40:5:40:3f:1e via eth0~
Oct 21 03:16:50 littleboy dhcpd:  DHCPACK on 192.168.11.102 to 0:40:5:40:3f:1e
via eth0~
Oct 21 21:14:52 littleboy dhcpd:  DHCPREQUEST for 192.168.11.102 from
0:40:5:40:3f:1e via eth0~
Oct 21 21:14:52 littleboy dhcpd:  DHCPACK on 192.168.11.102 to 0:40:5:40:3f:1e
via eth0~
Oct 22 15:13:02 littleboy dhcpd:  DHCPREQUEST for 192.168.11.102 from
0:40:5:40:3f:1e via eth0~
Oct 22 15:13:02 littleboy dhcpd:  DHCPACK on 192.168.11.102 to 0:40:5:40:3f:1e
via eth0~
Oct 22 15:52:02 littleboy dhcpd:  DHCPREQUEST for 192.168.11.102 from
0:40:5:40:3f:1e via 192.168.11.1~
Oct 22 15:52:02 littleboy dhcpd:  DHCPACK on 192.168.11.102 to 0:40:5:40:3f:1e
via 192.168.11.1

There is about a five minute diffrence in the machines clocks.