cannot ping one host on network - all others are ok

[Ron Kunzi]

Hi,

I have a HP J282 with a pinging problem. This box has HP-UX 10.20 loaded. Sometimes right after I logon root I can ping another box with 11.0 then the ping will stop. If I ping other hosts in my network -all in the same subnet- ping works fine. The other 11.0 box is a J6000. The J282 has been behaving strangely before this as well. I have used the ping hostname and IP address with both results the same. The behavior before this is things like it will not allow rlogins from the other hosts if we are not running our software on it. Also, I have swapped out the integrated module with the keyboard, mouse, and AUI ports on the back of the unit with the same results. How can one tell if the backplane is bad?

Any suggestions would be highly appreciated.

Ron

[Steve Labar]

What happens when you ping from the J6000 to other stations on the network? You might want to verify your network settings on the J282 and the J6000 for things like half-duplex vs full-duplex.

Good Luck.

Steve

[Shannon Petry]

Also, you may need to clear the arp table, and look for overlapping IP's. Most of the time I have seen this, 2 systems have the same IP. Sometimes 2 systems will have different ARP entries for both hosts, so some hosts reach 1, others reach the other.

man arp for more information.

Regards,
Shannon

[Tom Jackson]

Hi Ron:

If you can ping addresses within your network and can sometime ping addresses outside your network, sounds like a router/hub problem. Can you ping your default gateway? Can you try another port on your hub?

Tom

[Tom Jackson]

Hi again:

I agree with Shannon. You could be getting some inconsistant responses if you have another device, like a printer, with the same IP.

Look at and clear your arp tables.

Tom

[John Bolene]

was this machine built with an ignite image?

you may have duplicate mac addresses

check /etc/rc.config.d/hp***conf files

where the *** is ether, or base100, or whatever lan cards you are running

if you have anything listed in the station address field, this value overrides the hardware mac

[rick jones]

Assuming the two machines are not separated by a router, run lanscan on both machines to list the MAC addresses. Then on each machine, run the linkloop command, giving the other system's MAC address, using the PPA of the interface that should be connected to the same network.

If link-loop works in each direction, it implies (but not conclusively) that there are no duplicate MACs on the network, but perhaps there are some duplicate IP addresses as suggested.

As you ping the system with ping, the inbound ICMP echo request and outbound ICMP echo reply fields of the output of netstat -p icmp should increment on the "target." On the source, flip that around - the outbound ICMP echo request and the inbound ICMP echo replies should (iirc) increment.
Another thing to consider is running tcpdump while you run ping. Run tcpdump on both machines and tell it to only look at ICMP packets. Then on the target of the ping, if you see the ICMP echo request you know that the sender has the right IP to MAC mapping.

If you run tcpdump with the -e option, you can see the MAC address being used for the ICMP echo reply. That should match what you see in lanscan on the system sending the pings.