strange Ping problem

[Nobody's Hero]

Hello all, I have a strange problem going on but only between 2 systems. Every 5 minutes or so, box 'a' cant ping box'b' but 'a' can ping all other systems just fine.
# while true
> do
> ping s1dfa62
> sleep 5
> done
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive
s1dfa62 is alive

no answer from s1dfa62
no answer from s1dfa62
no answer from s1dfa62
no answer from s1dfa62
no answer from s1dfa62
s1dfa62 is alive

Any idea what the problem may be? Routing?
Any ideas how I can trouble shoot this?

Thanks.

[Torsten.]

Is this really hp-ux?

I would first use traceroute (or something similar) to see what is going on and who is involved.

[Nobody's Hero]

Actually it's our on;y Sun system we have here but I figured you knew that.

I did run a trace route yesterday and it never really finishes, just '*******'.

I may try and reset the nic by taking it up and down. Does that reset errors?

[TTr · Accepted solution]

You need to narrow down the problem. It seems that you have verified that there is nothing wrong with box a and when a can not ping b, a can ping other hosts. It that correct?

You need to use box c and box d to ping box b at the same time and see if the black-out to box b can be confirmed from boxes c and d and that the blackout occurs at exactly the same time. If it is confirmed then you move on to box b and start pinging multiple servers. This will tell you if it is within box b. To verify if it is on box b itself or on the network, you need to do some testing on other boxes that are connected in the same segment/switch as box b.

Possible causes can be missing patches on box b, network contention on box b, flaky network hardware on box b, flaky switch port on the network...

[Fabien GUTIERREZ]

i don't know exactly the time intervall in which your server is unreachable but it could comes from an ignoring from the distant host if it considers a "ping flood" from your first system
hp-ux stack has this kind of intelligent behaviour

[Prasanth Thomas]

Hi

Do one thing check the physical connectivity,,like cable and card....

and also restart the server and ping the server.....

is the server's lan is configured by apa method then just reconfigure it by manual mode.....

revert back

Regards

Prasanth Thomas

[Steven E. Protter]

Shalom,

I'd throw a traceroute in the code if the return code is non-zero.

ping <1 time timeout 10 seconds>
rc=$?
if [ $rc -ne 0 ]
tnen
traceroute
fi


Its not a true diagnostic tool but you might get closer to an answer.

Several possible causes.

Bad network hardware
duplicate ip address
network congestion

The traceroute will give you an idea where to look.

SEP