Ping check and timeout

[Thomas Marteau]

Hi all,

In order to be complete in my description of the issue, here are two
snapshots of a release/renew.
The first one is OK and you can see the 14 second offset between the
DISCOVER and the OFFER and the second is the problem.
[root at instlinux ~]# tail -f /var/log/messages | grep 00:50:56:86:55:7a
Oct 28 10:09:29 instlinuxdsy dhcpd: DHCPRELEASE of 10.244.143.251 from
00:50:56:86:55:7a (COMP1) via eth0 (found)
Oct 28 10:09:30 instlinuxdsy dhcpd: DHCPDISCOVER from
00:50:56:86:55:7a via 10.244.1.2
Oct 28 10:09:44 instlinuxdsy dhcpd: DHCPOFFER on 10.244.143.251 to
00:50:56:86:55:7a (COMP1) via 10.244.1.2
Oct 28 10:09:44 instlinuxdsy dhcpd: DHCPREQUEST for 10.244.143.251
(10.242.10.74) from 00:50:56:86:55:7a (COMP1) via 10.244.1.2
Oct 28 10:09:44 instlinuxdsy dhcpd: DHCPACK on 10.244.143.251 to
00:50:56:86:55:7a (COMP1) via 10.244.1.2

[root at instlinux ~]# tail -f /var/log/messages | grep 00:14:5e:64:46:b7
Oct 28 10:07:49 instlinuxdsy dhcpd: DHCPRELEASE of 10.244.142.124 from
00:14:5e:64:46:b7 (COMP2) via eth0 (found)
Oct 28 10:07:49 instlinuxdsy dhcpd: DHCPDISCOVER from
00:14:5e:64:46:b7 via 10.244.1.2
Oct 28 10:07:49 instlinuxdsy dhcpd: DHCPOFFER on 10.244.142.124 to
00:14:5e:64:46:b7 (COMP2) via 10.244.1.2
Oct 28 10:07:49 instlinuxdsy dhcpd: DHCPREQUEST for 10.244.142.124
(10.242.10.74) from 00:14:5e:64:46:b7 (COMP2) via 10.244.1.2
Oct 28 10:07:49 instlinuxdsy dhcpd: DHCPACK on 10.244.142.124 to
00:14:5e:64:46:b7 (COMP2) via 10.244.1.2

Any idea where to start digging? Clients are identical in this example
Windows XP 32 bits.

--
Regards, Thomas.



On Tue, Oct 27, 2009 at 4:57 PM, Thomas Marteau <isc at marteau.org> wrote:
> Hi all,
>
> On my bootp/pxe server, I put a timeout with ping-check statement.
> I have a weird behavior because sometimes my DHCP server answers to a
> DHCPDISCOVER without considering the timeout.
> I'm using the ISC dhcp server daemon 3.0.5.
>
> I put it as a general option and as a subnet option and it looks like
> it doesn't solve this issue.
> Any idea to explain this behavior?
>
> --
> Regards, Thomas.
>