time out with ping..

[senthil_3]

Hi,

I am having a HP rp2470 server with 11i installed. The problem is that I observ Large amount of packet drops when pinging the server from a different segment. But This doesn't happen with other machines which are in the same segment of the HP server.

I have changed the network cable , and point at switch.. I couldnt make it.
Please help..
Thanks,
Senthil

[dbk]

Hello Senthil,
Do you have the latest network patches installed ?? They have been known to cause issues. Also the server on the other network - did you only try to ping one server or more than one on other segment?
Verify that nslookup and reverse against ip work fine both ways on remote system and local one with the issue. Also, can you ping other machines on same network as suspect box without any issues ??

Regards,
dbk

[Elmar P. Kolkman · Accepted solution]

One other thing: try to set the speed of the network card of the HP (and if possible to port on the switch) to a fixed corresponding speed instead of auto-negotiate. At our site the switches and HP9000 servers have the bad habit of going to different speeds (100HD instead of 100FD, etc.) which could result in this problem.

To do this on a HPUX environment, use lanadmin -X
What number to use for the instance can be found in lanscan, but it differs per OS version/patchlevel if you need the NM ID or the Card Instance number.

[T G Manikandan]

If the machines in the same segment could ping well then it looks like the connectivity/route between the two networks.

check the routing table as whether the route is the way it is.
Also check whether the network is stable with the remote network.

[PVR]

Systems in the same segment are able to ping the server properly, that means server side network card configuration, network patch level etc.. are OK. I suspect the problem in the other segment

1. may be due to high traffic in the other segment.
2. The network device or the port of the network device where the other segment connected may be bad.
3. pl run ping command between 2 systems in the other segment. If that communication is also OK, you can suspect the network device which interconnects the device.

It would be helpful if you can give us a clear picture of how these segments are connected.

[Michael Steele_2]

Regarding "...packet drops when pinging the server from a different segment...". Is this token ring through a router?

More then likely its packet reassembly through the router if not NIC speed.

You can test packet reassembly problems by activating the ICMP 'spray' utility in /etc/inetd.conf on both nodes. 'spray' sends out a shotgun blast of ICMP packets and when used in conjunction with 'netstat':

# netstat -p icmp | grep -i -e collision -e dropp

...isolates the bad node.

Begin with a packet length of 1500 bytes and then reduce until the errors stop. For example, -c = 100 packets, -l = length 1500:

spray hostname -c 100 -l 1500
sending 100 packets of lnth 1502 to hps1 ...
in 0.0 seconds elapsed time,
80% packets dropped by hostname
2167 packets/sec, 1062.5K bytes/sec

Now reduce the -l length to 1000:

spray hostname -c 100 -l 1000
sending 100 packets of lnth 1002 to hps1 ...
in 0.0 seconds elapsed time,
40% packets dropped by hostname
2167 packets/sec, 1062.5K bytes/sec

[senthil_3]

Thnks for all,

There Was the Speed mismatch which After setting it right It worked Fine.

Thanks a Lot.

Senthil