why routing table different from netconf?

[Jenny C]

I have been having some network connection problem. When I checked the routing table, I found that it has a different subnet mask from my /etc/rc.config.d/netconf file. Would this has been the problem of my network connection? Why is this and how can I fix it?

Here is my abbreviated netconf entries:

IP_ADDRESS[0]=15.x.x.106
SUBNET_MASK[0]=255.255.248.0
....
ROUTE_DESTINATION[0]=default
ROUTE_MASK[0]="255.255.248.0"
ROUTE_GATEWAY[0]=15.x.x.1

But 'netstat -rvn' showed the following:
127.0.0.1/255.255.255.255 127.0.0.1 UH 0 lo0 4136
15.x.x.106/255.255.255.255 15.x.x.106 UH 0 lan0 4136
15.x.x.0/255.255.248.0 15.x.x.106 U 2 lan0 1500
127.0.0.0/255.0.0.0 127.0.0.1 U 0 lo0 0
default/0.0.0.0 15.x.x.1 UG 0 lan0 0

[Steven E. Protter]

But 'netstat -rvn' showed the following:

# ???????
127.0.0.1/255.255.255.255 127.0.0.1 UH 0 lo0 4136

#actual ip address
15.x.x.106/255.255.255.255 15.x.x.106 UH 0 lan0 4136

#the network
15.x.x.0/255.255.248.0 15.x.x.106 U 2 lan0 1500

#loopback has to be there
127.0.0.0/255.0.0.0 127.0.0.1 U 0 lo0 0

default/0.0.0.0 15.x.x.1 UG 0 lan0 0

If you look at the routing table, you'll see it does relate back to either the loopback, the ip addresss, or its network or any network routes that have been added with the route command.

SEP

[Ron Kinner]

Steven is correct. There is nothing wrong with your netstat or your masks.

What version of HPUX are you running? Some of the later version had dead gateway detection on by default. This used a ping to test that a gateway was alive and if the ping failed the gateway was declared dead and could not be used until it responded to a ping. Sometimes everything works fine until the router admin decides to harden his router and stops it from responding to pings.

Have you tried a traceroute?

Ron

[Bharat Katkar]

I suspect you used "ifconfig" to configure the lan interfaces which do not update the /etc/rc.config.d/netconf file and during which you may have entered 255.255.255.255 as subnet mask.

What you can do is if possible reconfigure the lan interfaces and Gatewayes thr' SAM or on the other hand try rebooting the server.

Regards,

[Jenny C]

I am using HP-UX 11.11. I don't quite understand why Ron and Steve said there is no problem in my routing table, since the netstat said that the subnet mask was 255.255.255.255, but my netconf file was 255.255.248.0. Did I miss something here?
When I lost the connection, I could ping 127.0.0.1, but could not ping the actual IP address of my box or the default gateway. syslog said inetd and sshd was terminated by signal 15, for which I didn't know why. I manually edited the netconf to make sure everything is fine, then rebooted the server. everything is fine. But I am afraid the box will disconnect again.

[Ron Kinner]

A mask of 255.255.255.255 refers to a single address, in this case, that of your HPUX's lan0 NIC. You will note that they use the same mask for the localhost address of 127.0.0.1 Your 255.255.248.0 still shows up on the network entry which the the one right below. You need to check your layer two stuff before picking on layer three.

While it is working do:

lanadmin
lan
display

and note the results. Especially the first couple of lines where it tells you if the circuit is up and what speed it is running at. The second page of the display will show you any errors. If errors are building up it may indicate a duplex mismatch, bad cable or bad NIC or bad port on the hub/switch.

Also look up the command:
linkloop
in the man and learn how to use it. If you can find a device on your local LAN which responds to this command then you you have a powerful tool for testing your layer two connectivity.

Finally you might want to check that your HPUX does not have dead gateway detection turned on. This is on by default in 11.0 and causes no end of problems when the gateway router stops answering pings - quite common these days that the net admin will turn them off to prevent worm attacks.

ndd -get /dev/ip ip_ire_gw_probe

if it comes back 1 then you should consider turning it off.

ndd -set /dev/ip ip_ire_gw_probe 0

and then edit /etc/rc.conf.d/nddconf

to make the change permanent.

Ron

[Jeff Schussele]

Hi,

If you couldn't ping the local IP, then you either had a HW problem with the NIC or the bus in which it resides or you lost network services altogether. inetd & sshd going down on signal 15 could indicate that they were killed manually by the kill command or they could have been killed when the NIC failed or network services died.
I'd check the GSP for HW errors to see if the NIC or bus failed.

Rgds,
Jeff

[Jenny C]

Thanks Ron for the detailed reply. It was helpful. Lanadmin and linkloop tests seemed ok. Gateway probe value was 1, I changed to 0. I'd like to check the console HW messages also, but the machine is remote. I will go back to the server for the local console.