Configuring adittional lan card

[radi_1]

Hi,
I have 10.20 server with a built-in 10mbs lan card which has ip addr. 192.5.5.2.
When I checked exisiting lan card by SAM I found the connected 100mbs card not configured so I cofigured it within sam as Fast Ether card type and card name btlan01(its EISA card).
I gave it IP addrr. 192.5.5.4 and aliase name ,and this has updated /etc/hosts with the new ip and host name.
But when I ran (ioscan -fnC lan)command the lan0 was claimed whith its device file is displayed and another lan instance 1 with driver name (btlan0)is claimed but no device file created for it.I ran insf -eC lan but still no device file. I ran lanadmin command, the speed of the card was 10000000 ,so do I need to edit /etc/rc.config.d/netconf for the new card and then reboot?What about the hpbase100conf?
thanks

[Uday_S_Ankolekar]

If you configured with sam then there is no need to update anything manualy in netconf file.

check with lanscan command , this will give you path , state of the card and many more details, also try netstat -in command.
Then see if you can ping the ipaddress from the server itself

-USA..

[Tim D Fulford]

I think everything is actually fine with the networking. To double check simply check the lan1 (I assume)
ifconfig lan1
If this retutns UP statement in the first line & the correct IP & netmask (in HEX) in the subsequent then all is well.

Couple of things additional to the question you asked this...

Because you have two devices on the same network (192.5.5.0, & I assume Class C netmask 255.255.255.0) the OUTBOUND traffic will go out of the lowest lan card, lan0. Thus by configuring the new lan on 100BaseT you are probably not increasing the network bandwidth much as the outbound bandwidth is simply 10BaseT!!!

Regards

Tim

[Ron Kinner]

I'm not all that thrilled with your choice of an IP address in the same subnet as the builtin card. I've seen a lot of posts saying the machine gets confused if you do that.

Anyway. First check your patches:

"The following patch is required in order to use the EISA 10/100Base-TX card on HP-UX 10.20 systems:
PHCO_9895 -- a patch for SAM to support EISA 10/100Base-TX.


PHNE_11507 -- a patch for the MAC address change.
The following patches are optional except if your system uses DLPI or Service Guard. If your system uses either of those, then please install the required patch in order to use the EISA 10/100Base-TX card on HP-UX 10.20 systems:

PHNE_9277 and PHNE_9278-- patches for DLPI to support EISA 10/100Base-TX.


PHNE_10145 -- a patch for Service Guard to support 10/100Base-TX.
The patch numbers listed above are current as of this release note. Please contact the Worldwide Enterprise Response Center if you need to ensure that you have the latest patches."

If your patches are uptodate then see the troubleshooting routine at:

http://docs.hp.com/hpux/onlinedocs/J2782-90009/J2782-90009.html

Note that pinging the server's new address from the server will not prove anything since it doesn't get sent out on the wire. The server knows its own ip addresses and will respond without bothering to transmit. Ping another device on the same subnet to see if the card is working.

Ron

[Tim D Fulford]

Rons advice...
.. Ping another device on the same subnet to see if the card is working...

As I mentioned before, this will go out of lan0, so will prove nothing. You CAN ping the new address 192.5.5.4 from ANOTHER device though.

You could use linkloop -i 1 , this will check physical connectivity out of lan1.

Tim

[Ron Kinner]

oops, I originally had that as "ping it from another device on the same subnet" then forgot why and changed it. Thanks for pointing that out Tim.

Ron

[Kevin Wright]

not all drivers actually create a device file. Not to worry about that. Got to another server and ping the IP to ensure connectivity.

[rick jones]

Indeed, only those drivers which were around when LLA (Link Level Access) was still a going concern would have device files. Those drivers that switched to DLPI (Data-Link Provider Interface) would not. The announcements of the impending demies of LLA went-out on or before the release of 10.20.

Multiple physical (in the eye of the transport) NICs in the same IP subnet on 10.20 is "difficult" - you have to do things with explicit route statements and such.

Life is much better in that area on 11.X, where you can often use ndd to set ip_strong_es_model to get the behaviour most folks expect. However, APA (Auto Port Aggregation) is still the best way to go to increase the bandwidth/availability of a system on a given subnet.

Finally, don't expect very much from an EISA 100BT NIC in terms of performance. If you get better than half link-rate you are doing well indeed. Also, the EISA 100BT NIC will _not_ autonegotiate, and only suports half-duplex operation.