Ubuntu network problem

[ChrisH]

I have a strange problem witha new install of Ubuntu.

 

Our ip range is 192.168.0.0/23

 

with most servers, switches, printers, WAPs on the 192.168.0 and the clients start at 192.168.1.1 but the default gateway and proxy is 192.168.1.254.

 

Now for some reason the new server just wont connect with that IP no ping no nothing. It will ping all the local servers on the 0.0 range and the random clients on the 1.0 range but just not 192.168.1.254.

 

The default gateway is just a stock smoothie at the moment so nothing exotic there and I have already discounted it being a subnet mask problem by pinging a few clients.

 

So is there some reason it might think that 192.168.1.254 is an invalid ip for that network or something?

[CyberNerd]

whats you routing table say?

netstat -rn

[webman]

Does the smoothie have ICMP replies disabled? Can you ping an external internet address from that box?

[ChrisH]

@ webman: I can ping the smoothie from other machines no problem.

@ CyberNerd: Will check tommorow.

[Geoff]

Assuming your using classless (CIDR) notation 192.168.0.0/23 gives you two class C networks or 512 IPs. Everything from 192.168.0.1 to 192.168.1.254. 192.168.0.0 is your network address. 192.168.1.255 is your broadcast address.

 

One gotcha is that Linux allows you to specify these options seperately. It's perfectly possible to supply a correct IP and netmask and give a rubbish broadcast address.

 

Also, what does the output from 'ifconfig' and 'route' look like?

[ChrisH]

I tried playing with the broadcast address. It had been set to 192.168.1.255 which I assumed was correct but I tried 192.168.0.255 on the off chance.

 

I will check out the routing table tommorow.

 

It just seems it doesnt like that one IP.

 

I thought at first it was the NIC and used a different machine of the same model.

 

I have had stock debian installed fine so not quiete sure whats up with this.

[ChrisH]

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
localnet        *                       255.255.0.0     U      0      0        0 eth0
default         fmproxy.fisherm 0.0.0.0            UG    0      0        0 eth0

 

# The primary network interface
auto eth0
iface eth0 inet static
address 192.168.0.8
netmask 255.255.0.0
network 192.168.0.0
broadcast 192.168.1.255
gateway 192.168.1.254
# dns-* options are implemented by the resolvconf package, if installed
dns-nameservers 192.168.0.6

 

Heres the info above.

[Geoff]

Your netmask is wrong. It should be 255.255.254.0

[SimpleSi]

Maybe just try booting the machine off a LiveCD and see if it picks up an address from your DHCP server

 

regards

 

Simon

[SimpleSi]

As Geoff says your netmask doesn't match your description of using 192.168.0.0/23. (it matches 192.168.0.0/16)

 

What netmask do the other machines on the net use?

 

regards

 

Simon

[ChrisH]

Yeah sorry it was 255.255.254.0 I just changed it to try and get it to see more of the network.

 

It will pick up a dhcp address and it will ping any IP in the 1.0 range apart from 1.254

[ChrisH]

It will ping right upto 192.168.1.253 how very annoying. I would be willing to change the proxy IP to get round this if it wasnt for all the manually configured laptops we have.

[Geoff]

Are there any firewall rules setup on either your proxy or the ubuntu machine that might be causing this?

[ChrisH]

Ubuntu is a lamp install from the cd with only ssh server added. The proxy is a smoothwall box with advanced proxy installed. So as far as I can see there is nothing special configured.

[SimpleSi]

Have you anything clever (VPN) setup on your switches that might be causing problems on that port?

 

Does changing ports make any diff.

 

If you boot off a different distro LiveCD - can you ping the proxy now?

 

regards

 

Simon

[ChrisH]

Ok after having another play with this I have found that any of my windows servers/clients get a response straight away. I have changed the port and network cable and I have swopped the machine completely yesterday when it was displaying these charateristics.

I have now found that if I leave the ping cmd running it starts displaying the replies after 2 minutes.

I ahve replicated this with a VM on my workstation but that doesnt get a response for ten minutes :\ .

I am going do a vanilla debian install now to see whats what.

 

Ps I have verified the disc

[Geoff]

Your switches and/or routers suck. Do they have some sort of QoS or other filtering enabled?

[ChrisH]

I have a full compliment of HP Procurve switches with nothing fancy set on them at the moment. The Ubuntu server it connected to the core switch which connects to every other switch around the school through fibre. The router/proxy is running on an old but decent workstation running smoothwall GPL with the advanced proxy mod.

Like I say everything else gets a response instantly it just doesnt seem to like these Ubuntu installs

[webman]

Connect the Ubuntu box using a crossover cable to the LAN side of the Smoothwall box to eliminate any other cabling and/or switch defects that could be causing this problem.

[ChrisH]

Im starting to think it is the Smoothwall box for some reason as I left the Debian installing when I left and it could not get the security updates but will check properly tommorow.

[ChrisH]

Well I have got it sorted now it was the Smoothwall box and for some reason it didnt want to play nicely with any of Linux box no matter what IP etc it was on!!!! Anyway I have sorted out my CN box and all is well

Strange? I think so

[Geoff]

ECN turned on perhaps?

[ChrisH]

No idea bit I doubt it given the amount of people that use smoothwall.

[Geoff]

http://www.cctec.com/maillists/nanog/historical/0104/msg00714.html