Rob
DNS service for my clients is provided by my gateway server, the same
machine as the DHCPD server. I think that's what the "option
domain-name-servers" line does. This allows me to provide 192.168
addresses to them when they try to access anything inside the house
with a name. If it's not a locally defined name, BIND forwards the
request to the internet.
I'm not sure I understand about dhcpd log and dns log. I scan
/var/log/messages, using the service name as the key. Looking at
'named' entries, all I see are messages of the form
"clients-per-query increased to XX".
I'm still mystified by the fact that only the i-devices (iphone,
ipad) exhibit this behavior of rapid dhcpd renewals. Mac's and PC's don't.
David
At 06:48 AM 2/17/2016, you wrote:
On 16/02/16 16:59, david wrote:
Folks
This might be the wrong place to ask, but I don't know where to turn.
My internal home network, including wireless, is controlled by a
Centos6 server, which provides dhcpd services, along with NAT. I
have DHCPD configured with the addresses 192.168.155.200 through
192.168.155.254 as the range for dynamic allocations. The
default-lease time is 1800 seconds, the maximum is 3600 seconds.
My windows clients, and even an ipad-mini behave nicely, asking for
DHCP renewals once ever five minutes, or at about 80% of the
default lease time, a behavior I can understand. However, several
of my guests, with their own iPads, I-watches, iPhones, connect to
my network (via a wireless access point which does not do routing
functions) and they're renewing once every 20-30 seconds. In
addition, these devices also loose connectivity for brief
intervals, which seems to be roughly synchronized with dhcp
renewal. This last fact I deduce by doing "tail -f
/etc/log/messages" and hearing them say "lost connection" at just
about the same moment the DHCPREQUEST and DHCPACK statements show up.
It's difficult to believe that Apple IOS devices (all of which are
running apple's latest) have a dhcp client problem not shared by
windows or even linux hosts.
Does anyone have any clues?
does your dhcpd update the dns? name resolution for devices seems to
be required for some applications and thus the dns needs to know
about the leases. Have you checked your dhcpd log entries and your
dns log entries? I have had situations where the dhcpd lease is
dropped due to not being able to complete dns update of the info -
thus the client retries again and again - they do get onto the
internet but the connection drops and a new lease is requested,
HTH
David Kurn
San Francisco
DHCPD.CONF file is excerpted below:
----------------------------------------
ddns-update-style none;
subnet 192.168.155.0 netmask 255.255.255.0 {
authoritative;
option routers 192.168.155.2;
option subnet-mask 255.255.255.0;
option broadcast-address 192.168.155.255;
option domain-name "daku.org";
option domain-name-servers 192.168.155.2;
option netbios-name-servers 192.168.155.2;
option time-offset -28800; # Pacific standard time
range dynamic-bootp 192.168.155.200 192.168.155.254;
default-lease-time 1800;
max-lease-time 3600;
}
--------------------------------------------
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
https://lists.centos.org/mailman/listinfo/centos
