two cards on same subnet

[H_16]

I have two btlan cards on the same subnet

and carried out the following:

#ifconfig lan1 ip_addr netmask_addr up
#ifconfig lan5 ip_addr netmaks_addr up
#lanadmin -x 1
Current Config = 100 Full-Duplex AUTONEG

# lanadmin -x 100FD 5
btlan: ERROR: Setting 100FD is unsupported
Current Config = NO LINK AUTONEG

I am having problems with modifying the second card to 100 Full duplex

Is this due to both cards on the same subnet is there are work around?

[Matti_Kurkela]

Having two or more cards on the same subnet should not cause any problems with duplex/speed settings, but it *will* cause problems with IP routing, unless you use APA and/or ServiceGuard.

(With "ndd -set /dev/ip ip_strong_es_model 1" you can make it work somewhat, but having two NICs simultaneously active on the same network segment without APA is an unsupported configuration and is not recommended.)

With APA you can get both cards active together. ServiceGuard can use APA if both are installed on the same server.

With just ServiceGuard only one of the cards can be active at a time. If it fails, SG will switch the traffic (and the IP addresses) to another card.

Without knowing the hardware type of your NICs and your HP-UX version & Quality Pack level, it's hard to guess what might be wrong.

The output of commands "ioscan -fnkClan" and "lanscan -v" would be very informative. If one of the cards is built-in to the system board of your server, the output of "model" command might be useful too.

MK

[F Verschuren]

You have to use the capital X to set the speet..
lanadmin -X 100FD 5
if it does not work please check the cable (linkloop.)

[F Verschuren]

explanation for linkloop
use lanscan to find out the mac adress
to check the cable in lan 5
linkloop -i [mac of lan5] 5
to check the cable in lan 1
linkloop -i [mac of lan1] 1
to check if lan 1 and 5 are in the same vlan.
linkloop -i [mac of lan1] 5

[Ganesan R]

Hi,

Could you provide the lan5 card details?
Is this card supports Full duplex?

Look at the errors.
>>btlan: ERROR: Setting 100FD is unsupported
Current Config = NO LINK AUTONEG<<

[F Verschuren · Accepted solution]

Just tested the ERROR it is the capital X verses the low letter x that is causing the error...
strange error code for this typo...

:nl11588> lanadmin -x 3
Current Config = 100 Full-Duplex MANUAL

nlxsms01:nl11588> lanadmin -x 100DF 3
btlan: ERROR: Setting 100DF is unsupported
Current Config = NO LINK

:nl11588> lanadmin -X 100DF 3
Must be super-user
:nl11588>

[UVK]

As said did you try
#lanadmin -X 100DF 5
You should be set with that, If not check for a couple of things
1) Cabling
2) Duplex setting on your switch
3) You NIC card spec's

[Jeroen Peereboom]

Did you both really type 100DF? Should be 100FD.

JP

[rick jones]

Frankly, in this day and age, unless one is playing with ancient networking kit, there should be no need to hardcode duplex settings. Your own output already shows that you were getting 100FD, which says the other side was doing autoneg, so unless you change the other side too, all setting a hardcode of FD will do is (not so subtly) break the link.

[Steven E. Protter]

Shalom,

In early versions of 11.11 this plan was strictly forbidden. I tried it once and both NIC cards dropped of the network instantly.

Later releases of 11.11 supposedly supported it.

I think you need APA, Auto Port Aggregation to get any real benefit of increased band width out of this configuration. That is the tool designed to do the job.

SEP

[rick jones]

If by "same subnet" you mean "same IP subnet" then there are issues to be aware of involving configuring two NICs in the same system with IP addresses in the same subnet. It may or may not do what you expect depending on what you expect.

If by "same subnet" you simply meant the same set of switches, that issue with IP's wouldn't come into play, and in future it might be better to say "same broadcast domain"

[Jeroen Peereboom]

Indeed I used to set interfaces to 100FD because many times it ended up as 10HD (sometimes only at one end of the connection).
Nowadays (with 10/100/1000Mbps) I usually have autoneg with good results.

JP

[H_16]

Rick,

At the moment I have two cards going to different switchs

one card ip is for example:
172.168.3.144

# lanadmin -x 1
Current Config = 100 Full-Duplex AUTONEG


The other card is
172.168.3.145

# lanadmin -x 5
Current Config = 100 Full-Duplex MANUAL

My aim is to have the primary card .144 and then use the lan5 card as a failover secondary card only.

Im also concerned when I rebooted I had to re-run the lanadmin -x 100FD 1 command.

I did update my hpblanconf file with:

HP_BTLAN_INTERFACE_NAME[0]=lan1
HP_BTLAN_STATION_ADDRESS[0]=0x001083F7D908
HP_BTLAN_SPEED[0]=100FD
HP_BTLAN_MTU[0]=""


Is this correct??

[F Verschuren]

If this is correct is depening on your lancart, please post a
ioscan -funC lan

Also it is better not to use
HP_BTLAN_STATION_ADDRESS[0]=0x001083F7D908
This wil fix the mac adress, this is dangaros if you use ignite to reinstall servers. After creating a clone whit this setting you will get a duplicated mac adress.

Beter is to never configger the mac and let the system deside what the mac adress is.

[H_16]

I have attached ioscan output

[likid0]

Don't you have a service guard license ?, or and media os that includes it ?

As said above, for failover you can use a one node service guard cluster, you have one card as primary and the other on standy, when you loos the primary, the IP on the primary nic siwtches to the standby card. So you only need one IP to configure it.

APA also works great, you need to purchase it do.

[H_16]

I have ServiceGuard installed..

So if I create a cluster package I only need to specify just my primary IP address within the package or do I use a virtual IP??

[F Verschuren]

to see if the hpblanconf is correct please do a grep -v ^# /etc/rc.congig.d/hpblanconf

questions
- The extra IP is that a cluster IP?
- are the lancarts already in the cluster?
cmviewcl -v
- do you want (for input trafic) more performance ore higher availebility(for in and out trafic)?

[H_16]

root@nmesito1:/
# grep -v ^# /etc/rc.config.d/hpbtlanconf

HP_BTLAN_INTERFACE_NAME[0]=lan1
HP_BTLAN_STATION_ADDRESS[0]=0x001083F7D908
HP_BTLAN_SPEED[0]=100FD
HP_BTLAN_MTU[0]=""
HP_BTLAN_INTERFACE_NAME[0]=lan5
HP_BTLAN_STATION_ADDRESS[0]=0x001083F74A4D
HP_BTLAN_SPEED[0]=100FD
HP_BTLAN_MTU[0]=""


HP_BTLAN_INIT_ARGS="HP_BTLAN_STATION_ADDRESS HP_BTLAN_SPEED HP_BTLAN_MTU"

[H_16]

questions
- The extra IP is that a cluster IP?

I have three IPs

The first ip is for the first card
The second ip is for the second card
And I have a Virtual IP address

- are the lancarts already in the cluster?
cmviewcl -v

No....the cluster hasnt been built yet, just installed new OS and MC/SG

- do you want (for input trafic) more performance ore higher availebility(for in and out trafic)?

These servers are going to used for network node manager, so I want high network performance

[Laurent Menase]

HP_BTLAN_INTERFACE_NAME[0]=lan5

shouldn't it be
HP_BTLAN_INTERFACE_NAME[1]=lan5
....[1]...

[F Verschuren]

This is the correct configuration:
to make sure that after a reboot you still have the same settings,
I removed the HP_BTLAN_STATION_ADDRESS to prevent problems in the future...

HP_BTLAN_INTERFACE_NAME[0]=lan1
HP_BTLAN_SPEED[0]=100FD
HP_BTLAN_INTERFACE_NAME[1]=lan5
HP_BTLAN_SPEED[1]=100FD

[F Verschuren]

The last line must be in the file also (never change this one)

[F Verschuren]

to get the higher availebility:
ad lan 5 (als failback of lan1) in the cluster
and ad the IP adresses in the cluser,
IF a cable (ore lancart ) fails it will automatic switch form lan 1 to lan 5.