problems with netconf file

[Donald Lalonde]

Hi,

We installed a new Gig fibre card in our RP5470 server running HPUX 11i.

I tried to change the netconf file so that the gig card would be lan0 (see the attached file).
But, when I restarted the network (/sbin/init.d/net stop/start), it complained about a duplicate IP for lan0 and did not complete the restart of the network.

I restarted the server and the boot process locked up when starting the SNMB Sub Agent.

Is there something wrong with the new netconf file?

Thanks


->lanscan
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
0/0/0/0 0x00306E1C0063 0 UP lan0 snap0 1 ETHER Yes 119
0/1/0/0 0x00306E4A8834 1 UP lan1 snap1 2 ETHER Yes 119
0/5/0/0 0x00306EF56524 2 UP lan2 snap2 3 ETHER Yes 119


->ioscan -fnC lan
Class I H/W Path Driver S/W State H/W Type Description
===================================================================
lan 0 0/0/0/0 btlan CLAIMED INTERFACE HP PCI 10/100Base-TX Core
/dev/diag/lan0 /dev/ether0 /dev/lan0
lan 1 0/1/0/0 btlan CLAIMED INTERFACE HP A5230A/B5509BA PCI 10/
100Base-TX Addon
/dev/diag/lan1 /dev/ether1 /dev/lan1
lan 2 0/5/0/0 igelan CLAIMED INTERFACE HP A6847-60101 PCI 1000Ba
se-SX Adapter

[Pete Randall]

Why do you want it to be lan0? Just leave it as lan2 and it should work fine.


Pete

[melvyn burnard · Accepted solution]

Well from what I can see, your lan0 card is your built in Core interface, and the gigabit interface is lan2.

So if you wish to assign the address that was on lan0 to the gigabit interface, i.e. lan2, that is the change yiou need to make.
Also, from the netconf file, you have one setup using array [0], and another using array [2].
You need to have no missing array numbers, so set the [2] to [1] in these lines:

INTERFACE_NAME[2]=lan2
IP_ADDRESS[2]=142.53.157.236
SUBNET_MASK[2]=255.255.255.0
BROADCAST_ADDRESS[2]=142.53.157.255
INTERFACE_STATE[2]=up
DHCP_ENABLE[2]=0

[Noel Miranda]

What does ifconfig show?

[Donald Lalonde]

I wanted to make the Gig card the default route.
I was told by a tech at HP that only lan0 can use the word default for it's routing.

I did not know about the sequential array numbers. So, I have to use 0 then 1 and then 2, ...

OK thanks

[Pete Randall]

I think your tech at HP is confused. Have a look at this from one of my systems:

# lanscan
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
1/0/1/0/0/4/0 0x00306EA72B18 1 UP lan1 snap1 1 ETHER Yes
119

It's obviously lan1 and if you take a look at the attached netconf, you can see the default route.


Pete

[Donald Lalonde]

Sorry, I meant that the default statement for the route has to be with the lan in array 0.

[Pete Randall]

Ahh! That may well be - I wasn't aware of that restriction but it makes sense, I think.


Pete

[Ron Kinner]

I'm not so sure I agree with the HP Tech.

Looks to me like what you need to do is move things around in your netconf. First define both interfaces then set your default route to what ever you want it to be. Currently you define lan0 then the default route and then lan2. It's no wonder that you can't use the default route you want since its network hasn't been defined at the time you try to use it. If you read the remarks in netconf you will see they really want you to finish defining networks before going on to the route definitions.

Lan0 is always the builtin lan so don't try and change it. Don't know that the [ ] numbers have to correspond with lanID but I know it will work with gaps in the [ ] sequence so using [0] for the lan0 and [2] for lan2 shouldn't hurt and is easier to keep straight.

Remember there is no connection between the [0] in the lan definitions and the [0] in the route definitions. These are two separate lists and the [numbers] are only used to group commands together.

Finally your second route statement makes no sense. Since the 142.53.22 network is on lan0 this traffic is local and needs no statement and certainly not one that attempts to send it to another router. You are essentially telling it that traffic to the default gateway really has to go to a different place which it doesn't understand.

Ron

[Donald Lalonde]

OK, so I did some test on a development server.

First of all, the default route does not have to be on the [0] interface.

I am not sure about the sequential numbering of the interfaces. But, for now, I decided to use interface [0] and [1].

I rebuilt my netconf file.
I put the internal 10/100 Mbps card as lan0, interface [0], and the gig card as lan1, interface [1].

I also configured the default route to use the Gig Card. But, it uses interface [0]. Does this matter?
Will this new file work?

[melvyn burnard]

Looks ok to me.
Just remember, although the ROUT statements use [0], this does not mean they are tide to the lan you configured with [0], the important thing is what address you specify, and what route type etc.
The array is used simply to allow the start up scripts to go through each set of parameters, in an easy way.

[Pete Randall]

Yeah, that should work. If you look at my netconf file (previously attached) I think you'll see much the same thing: lan1, yet the default route is instance [0].


Pete

[Ron Kinner]

You have a typo in netconf:

INTERFACE_STATE[1]=up
DHCP_ENABLE[2]=0


You want DHCP_ENABLE[1]=0

But that won't hurt anything.

What do you mean it still uses lan0?

What does

netstat -rn

show?

(You can't trust traceroute. It's been broken on HPUX since 11.0. Always uses lan0 and ignores the routing table. There is a ping option that works better. ping -o should show the first 9 hops.)

Ron

[Jim Keeble]

I'm not sure traceroute is broken, but the syntax did change in 11.11. You can use the -i switch to specify which lan interface the packets are sent on, and leaving it off is usually the reason it fails for me.


root@dolan> /usr/contrib/bin/traceroute
Version 1.4a5 with HP mods
Usage: traceroute [-dFInrvx] [-6] [-f first_ttl] [-g gateway] [-i iface]
[-m max_ttl] [ -p port] [-q nqueries]
[-s src_addr] [-t tos] [-w waittime]
host [packetsize]

Sorry, I know this is off the real subject, but just in case it helps anyone ...

[Jim Keeble]

I think I see what the problem is. Or was.

If you look at the "stop" section of /sbin/init.d/net, the "stop" function doesn't really remove IP's from any interface, so when you re-ran it, you really were trying to give an IP to lan0 that what already assigned to a different interface.

As others have alluded to, changing netconf can't alter the PPA number assigned to an interface, it can only change which interface name you assign the IP to. So, you assigned the IP to lan0 which was probably still plugged into the wrong subnet you an invalid IP for that subnet. It hung because it couldn't reach any hosts using that subnet, like your DNS server or anyone else.

So, using lan2 for your gigabit card's name should work ok. If you feel you HAVE to change it, you may be able to use the "ioinit -f" command to re-arrange the "instance numbers" of the cards, which is where the PPA number comes from.

Good luck, hope this helps.

[Ron Kinner]

JIm,

I use traceroute to find out where the system would naturally send a packet. If I have to tell it which interface to use then that defeats the purpose so I still maintain it's broken. Hey, Microsoft can get it right even if they spell traceroute wrong, why can't HPUX?

Ron

[Jim Keeble]

Ron,

Your point is well taken. I'm just trying to point out that it's possible to use traceroute, it's not completely "broken".

At any rate, I hope BP is able to get his problem straightened out.

[Donald Lalonde]

Thanks guys.
Last night, I was finally able to play with the server again.

With my new netconf file, everything works.