Multiple Network Interfaces in 11i

[Michael Simone]

I have an L2000 running 11i with 2 4-port network cards. Its primary LAN interface is lan1, 192.100.100.2, 255.255.252.0. I want to configure lan5 on a secondary network (10.10.10.2, 255.0.0.0), connected to a totally separate private switch with a Windows server on the other side (10.10.10.108). I'm able to configure the lan5 interface and everything looks right, but it's not being used when I try to ping 10.10.10.108. Ping and traceroute appear to try going out through 192.100.100.2. And when I force traceroute (-s 10.10.10.2), it times out. Ping doesn't work either, so I don't think I'm making a connection.

I've successfully implemented this on another HP-UX machine, running 11.0, with the same Windows server, using the same cables and lines. And I've also tried it on lan6 of this node's card. So I think I've isolated the problem down to something with the networking config on this 11i host.

It appears to match the 11.0 machine in settings, but I must be missing something.

Any help is appreciated.

[Ross Zubritski]

Hi,

What is your default gateway set to?

Regards.

RZ

[Steven E. Protter]

One of your network cards is designated as primary. ping requests go out that interface first.

ping can probably be told to use different lan interfaces.


You need to look for conflicts in your /etc/rc.config.d/netconf file

A single extra period or typo in the configuration of one of the lan's can cause these types of problems.

I'm pasting in a partial example for you to use as a reference.


HOSTNAME="tzfat.juf.net"
OPERATING_SYSTEM=HP-UX
LOOPBACK_ADDRESS=127.0.0.1

INTERFACE_NAME[0]="lan1"
IP_ADDRESS[0]="10.1.11.60"
SUBNET_MASK[0]="0xffff0000"
BROADCAST_ADDRESS[0]="10.1.255.255"
INTERFACE_STATE[0]=""
DHCP_ENABLE[0]=0


INTERFACE_NAME[1]="lan1"
IP_ADDRESS[1]="10.1.11.61"
SUBNET_MASK[1]="0xffff0000"
INTERFACE_NAME[2]="lan1:1"
INTERFACE_STATE[1]=""
DHCP_ENABLE[1]=0

ROUTE_DESTINATION[0]="default"
ROUTE_MASK[0]=""
ROUTE_GATEWAY[0]="10.1.10.1"
ROUTE_COUNT[0]="1"
ROUTE_ARGS[0]=""

INTERFACE_NAME[1]="lan0"
IP_ADDRESS[1]="192.168.0.4"
SUBNET_MASK[1]="255.255.255.0"


note that I've got a virtual address on lan1, two addresses one port.

lan0 notation is working by the way.


You've got great NIC cards btw, but you might want to try a different setup.

You can do port aggregation and get 1 IP address across all four ports on each card. That gives you a built in load balance and less IP addresses to deal with. It is an added cost add in software product.

If I had your NIC card I'd look into it.


SEP

[Michael Simone]

I don't have one set now. I DID have one before (on my primary network), but I did a netstat -f to reset everything, and made sure there was no default route set.

[Michael Simone]

I've reviewed my netconf file, and everything appears to be in order. Does that require a restart of network services to take effect (this machine is in production, so I have to be careful messing with the primary interface).

I like the idea of using APA for port aggregation, but part of the reason for the private network is to isolate traffic from the Windows server (web calls through the 192.100, db calls through the 10.10).

Thanks for your help, and any future assistance.

[Sridhar Bhaskarla]

Hi,

The IPs do seem to fall under the same broadcast domain. However, check the subnet mask of the windows server. They should match. On 11i, traceroute without any options is goign to try the first interface. Use -s option with traceroute.

Since, you don't have a default route set, the possible reason is that the subnet masks of your windows server and your HP-Server are not matching.

Also, to find out the physical connectivity, get the physical address of the windows server and use linkloop on hp system to check the physical connectivity.

-Sri

[Ron Kinner]

Pings being a subset of IP should follow the routing table. The concept of primary and secondary only apply when you have two interfaces on the same subnet.

Post your /etc/rc.config/netconf

A restart is your best bet on getting it to work. I don't think it constantly reads that file looking for a new change. There may be a way to reread with something like lanconfig but I'm at home and don't have access to my hpux systems or docs right now.

First verify that the interface is up and open for business:

lanadmin
lan
ppa x
display

Where x is the index for your interface. If you are not sure just start with 1 and increment until it shows up. Do you have two UPs and do you see any packets received or sent?

Run a linkloop test with the -v option to make sure your level 2 connectivity is OK. This will rule out things like a duplicate MAC address.

Post your netstat -rnv

Do you see the 10.0.0.0 network in there?
Do you see the 10.10.10.2 interface?

Verify that you do not have a duplicate IP address by disconnecting the network cable and having another device on the 10 LAN ping 10.10.10.2. If you get a response you have a duplicate. Try a different number.

Ron

[Michael Simone]

I've confirmed in lanadmin that the interface (lan5) is up/up, and is showing incoming and outgoing packets when I ping from either side, but I still see no actual replies. The cable is the same one I used testing with my 11.0 machine, though I will try another one just to eliminate that. Linkloop is next as well.

Thanks!

Here's my netconf:
HOSTNAME="w3"
OPERATING_SYSTEM=HP-UX
LOOPBACK_ADDRESS=127.0.0.1

INTERFACE_NAME[0]=lan0
IP_ADDRESS[0]=192.100.100.2
SUBNET_MASK[0]=255.255.252.0
BROADCAST_ADDRESS[0]=192.100.103.255
INTERFACE_STATE[0]=up
DHCP_ENABLE[0]=0

GATED=0
GATED_ARGS=""
RDPD=0
RARP=0

IP_ADDRESS[5]=10.10.10.2
SUBNET_MASK[5]=255.0.0.0
INTERFACE_NAME[5]=lan5
BROADCAST_ADDRESS[5]=10.255.255.255
INTERFACE_STATE[5]=up

[Ron Kinner]

You should probably use [1] instead of [5] in netconf unless you are leaving out a few entries but I don't think that would cause your problem since you are getting an UP/UP in LANADMIN.

What does
netstat -rnv
show?

Are you sure the IP address is unique?

Have you tried a linkloop test? It's pretty easy to do. You just need the mac of another device on the same LAN. Check out the man on linkloop.

Could the switch have you in the wrong VLAN? Try sticking a Windows PC on the same port and assign the same IP address and mask to it and see if it can ping.

The switch could also be tagging your packets with VLAN info. If you are not set up for it that will kill your connection.

Ron

[Michael Simone]

Thanks everyone. It's working now. The numbering on my ports seems out of whack (lan6 to the left of lan5), but I've got it working just fine on lan8 now.

Thanks for all your pointers. Lots of good info in there.