Open vms 2nd nic not pinging

[PRABHAT KUMAR GUPTA]

I have Alpha ds20E , we have alpha 200 running ovms 7.1 , we have upgraded ovms 7.1 to 7.3-2 and then 8.2 to test our application running on alpha ds20e . after upgrading we found our 2nd nic not pinging to their n/w.Both card ip are in same class and net .we are using de500 nic card . and tcp/ip ver is 5.5

[Duncan Morris]

Prabhat,

can you give us some more details please?

Try

tcpip show inter/full
tcpip show route

and

mc lancp show dev/fu
mc lancp show dev/cou

for starters.

Duncan

[Duncan Morris]

Oops,

that first lancp command should be

mc lancp show dev/char

Which version of tcpip were you using on the original 7.1 system - UCX 4.x, or TCPIP V5.0?

Duncan

[PRABHAT KUMAR GUPTA]

The outputs asked are attached .pls advice

regds

Prabhat

[Oswald Knoppers_1]

Duncan asked for LANCP informantion not LATCP :-).

My guess is some mismatch in full duplex / autonegotiate settings on the device.

Compare the console setings ewa*mode with the settings on your switch/router.

Oswald

[Duncan Morris]

Prabhat,

your stats show that both NICs are being used to send/receive packets.

What address are you trying to ping from and to?

Try using traceroute to check the route.

Duncan

[Hoff]

I see a private Class C with active controllers and network traffic.

Could this be a case of misunderstanding IP subnet routing and IP default gateways, and how same works?

(IP hosts don't automatically choose the path that the packet arrived upon as the path for the response, for instance.)

Or is this a case of an intervening firewall?

Or...

ICMP ping can itself be misunderstood, it's a low-level piece of IP itself, and not layer 3 traffic. (Subnet routing is weird enough, when you first run into IP.)

Based on what I see in the text file, the configuration appears to be working, we1 is the current path for most traffic, which in aggregate which implies there's some sort of routing or other unstated issue here -- nobody typically discusses nor even points to ping without having something else going on -- and that detail is lacking.

Some background on the particular problem and the particular environment and the particular commands used here may well be (will be?) useful.

Depending on the configuration and the goal, if you want all the controllers active akin to NetRAIN, that's not AFAIK available with TCP/IP Services; OpenVMS TCP/IP does not offer link aggregation. There are ways to have an address be more reliable (eg: FailSAFE), and there are ways to use a managed switch here.

Or get somebody in to look at your IP network.

[Jur van der Burg]

Using two nics with ip addresses in the same subnet is asking for trouble. If you send a packet to your subnet which interface will it be sent to? You don't know. And it will not go out on both interfaces for certain. This setup is already broken.

Jur.

[Khalique]

Two IP address from the same class is the problem here as the hostname is mapped to only one interface that actually disable the second nic with no mapping to host database.

[Khalique]

Both the NICs are configured in the same ip class due to which only one NIC is mapped to host database leave the second one unable to communicate on the network. So, you have to assign ip address to the second NIC of different IP class or subnet.