[SOLVED] Problems after reconfiguring LAN Subnet to 10.10.1.0/24

Sum1 · 09-30-2014, 05:39 PM

Kind of a nagging frustrating day.
Last night I completed an upgrade to my router/firewall/dhcp server box from Slackware 13.37 to Slackware 14.1 and updated everything to Current.

For a long time, I had the DHCP server set to provide addresses to LAN clients on subnet 195.195.195.0/24. I found out this was not an acceptable Private address range, so decided to take the opportunity to change it to 10.10.1.0/24.

This is what my dhcpd.conf looks like:

Code:

ddns-update-style interim;
authoritative;
log-facility local7;

default-lease-time 43200;
max-lease-time 43200;
option subnet-mask 255.255.255.0;
option broadcast-address 10.10.1.255;
option routers 10.10.1.1; ###this is eth1 / the LAN-facing NIC ###
option netbios-name-servers 10.10.1.199; ### Samba Server - "official" WINS server ###
option domain-name-servers 75.75.75.75;
option domain-name-servers 75.75.76.76;

subnet 10.10.1.0 netmask 255.255.255.0 {
  range 10.10.1.50 10.10.1.200;
  }

For the most part everything worked.
I changed the Samba server's ip address and the Backup server too.
Provided static ip's to network document scanners and printers.
Set all the Windows XP and 7 clients to dhcp.

But there's some nagging issues with ip address conflicts, clients getting temporarily unresponsive access to directories on the samba server and spotty internet access. The whole network just seems to stutter a bit after the LAN segment address change.

I also notice my ssh sessions into the router/firewall box get dropped with regularity.

I did a few minutes of tcpdump to try to determine an obvious problem but nothing jumped out except for the amount of packets getting dropped by the kernel. Here's some stats from about 4 - 6 minutes:

Code:

17485 packets captured
72439 packets received by filter
54954 packets dropped by kernel

Doesn't seem right.

The only thing I can figure to ask is whether my subnet netmask is wrong for a LAN segment of 10.10.1.0/24 ??

It works at about 85% right, but the last 15% is killing my time with user complaints.

Any guidance or ideas what to test or log are greatly appreciated.

Sum1 · 10-03-2014, 08:08 AM

1. It appears necessary in newer versions of dhcpd to include "global" configuration parameters in each declared subnet.
There was no subnet mask error in the configuration.

Code:

default-lease-time 43200;
max-lease-time 43200;
option subnet-mask 255.255.255.0;
option broadcast-address 10.10.1.255;
option routers 10.10.1.1;
option netbios-name-servers 10.10.1.100;
option domain-name-servers 75.75.75.75;

subnet 10.10.1.0 netmask 255.255.255.0 {
        range 10.10.1.50 10.10.1.200;
        authoritative;
        option subnet-mask              255.255.255.0;
        option broadcast-address        10.10.1.255;
        option routers                  10.10.1.1;
        option netbios-name-servers     10.10.1.100;
        option domain-name-servers      75.75.75.75;
        option ntp-servers              10.10.1.1;
        option domain-name-servers      75.75.76.76;
        option time-offset              -18000; # Eastern Standard Time
  }

2. I was not able to determine why certain ip address conflicts were occurring.
I did not want to delete the /var/state/dhcp/dhcpd.leases file.
I gave in and assigned different/updated addresses on the subnet to those workstations experiencing the problem.

3. The remote ssh session disconnect problem was solved by adding the following to /etc/ssh/sshd_config:

Code:

TCPKeepAlive yes

and

Code:

UseDNS no

jefro · 10-03-2014, 02:46 PM

Thanks for the update and solution.