Network rate exceeded threshold

[Abel Berger]

Hello all, I?m working with servers HP-UX
v. 10.20.
ONE these servers when I ping in the same network its OK ( 0 ms), traceroute OK too, but when I use rcp or ftp for trafer files in the same network ( for examples 3 Mbytes ), the transfer is much slow.
Where I can find the problem ? Switch, lan adapter...The only message that I have is :
"Network rate exceeded threshold"
Please, could you help me ?!

Regards,

Abel Berger
Unix Specialist
55-11-7858-2220

[Jaimin Parikh]

Hi,

Try resetting the NIC using lanadmin command.

Hope that will help.

Regards,
Jaimin

[Kevin Wright]

use lanadmin to find out the speed and errors of the nic. find out if the nics are 10/100 base, also, if you are going through a switch, certainly check it for malfunctions. You can do netstat -I lan0 5 5 to check the packets going in and out of your nic.

[Vincenzo Restuccia]

lanscan for see id,lanadmin -M xxxx id.

[Printaporn_1]

Hi,

From lanadmin does MTU is 1500 ?

[Abel Berger]

Yes, MTU=1500, but Ierrs is very high.

Name Mtu Network Address
lan1 1500 10.20.46 besup

Ipkts Ierrs Opkts Oerrs Coll
104488785 9142492 101594462 0 0

[K.Vijayaragavan.]

Hi
Use recive socket buffer size and send socket buffer size options along with rcp command to get better performance

ex.
rcp -R -S

regards

vijay

[Abel Berger]

I think that de reset will not resolve the problem, because de server was rebooted nine
days ago, and the problem appear 30 days ago.
Thanks,
and we go to continue trying!

[rick jones]

a high ierrs rate could be a number of things. one could be a duplex mismatch between the NIC and the switch - make sure that both are set the same. You cannot have one side set to say auto and the other set to say full-duplex.

It is also possible that there are physical problems with that network - see what the specific lanadmin stats say - are there late collisions, or FCS errors or other nasty things happening?

Finally, though I _thought_ it would be an inbound discard, not an "error" it could be that the 10.20 ipintrq is being overflowed for that interface. You could try using nettune to increase the size of the ipintrq to the maximum of 1000. However, that is a bit of a longshot.

[Abel Berger]

For research :
#lanscan :
Hardware Station Crd Hardware Net-Interface NM MAC HP DLPI Mjr
Path Address In# State NameUnit State ID Type Support Num
8/0/1/0 0x001083182E7E 1 UP lan1 UP 4 ETHER Yes 115

#lanadmin :
Network Management ID = 4
Description = lan1 Hewlett-Packard 10/100 TX Full-Duplex Man
ual TT = 1500
Type (value) = ethernet-csmacd(6)
MTU Size = 1500
Speed = 100000000
Station Address = 0x1083182e7e
Administration Status (value) = up(1)
Operation Status (value) = up(1)
Last Change = 56715
Inbound Octets = 610702491
Inbound Unicast Packets = 102047451
Inbound Non-Unicast Packets = 3508709
Inbound Discards = 0
Inbound Errors = 9304291
Inbound Unknown Protocols = 1670534
Outbound Octets = 2921796767
Outbound Unicast Packets = 102609041
Outbound Non-Unicast Packets = 147
Outbound Discards = 0
Outbound Errors = 0
Outbound Queue Length = 0
Specific = 655367

#lanadmin -x 4
Current Speed = 100 Full-Duplex Auto-Negotiation-OFF

[Abel Berger]

# netstat -I 4 4 :
(ni0) -> input output (Total)-> input output
packets errs packets errs colls packets errs packets errs colls
0 0 0 0 0 105921000 9307632 102985096 0
0
0 0 0 0 0 1702 355 1418 0 0
0 0 0 0 0 1478 273 1304 0 0

[Abel Berger]

Continuing laadmi...

Index = 4
Alignment Errors = 18816
FCS Errors = 9328572
Single Collision Frames = 4
Multiple Collision Frames = 0
Deferred Transmissions = 7
Late Collisions = 0
Excessive Collisions = 0
Internal MAC Transmit Errors = 0
Carrier Sense Errors = 0
Frames Too Long = 5007
Internal MAC Receive Errors = 0

[Abel Berger]

Hello Rick Jones !

What is a FCS errors. Do you saw how many FCS erros ?

Regards, and I still to foud the problem.

[Jaimin Parikh]

Hi,

I guess FCS means Frame Check Sequence.

Just a thought.Can you change the NIC to half duplex mode or alter the server's connectivity to other port on switch or hub?

Good luck.
Jaimin Parikh

[rick jones · Accepted solution]

Yep, FCS - frame check sequence - a link-level checksum if you will.

The second half of the lanadmin stats indicate a network in seriously poor physical health. As has been pointed-out, it could be a duplex mismatch between the NIC and the switch port. If could also be a bad NIC or bad switch port or even a bad cable I suppose. Start with the easy stuff first though - make sure the duplex settings match, then try a cable swap with a known good cable, then try a new switch port, then try a new NIC.

[Abel Berger]

Today I will be executing the tests :

Change Switch Port, Change Adapter and
change the cable, OK.

Thus that possivel I will return.

Best regards !
Abel Berger

[Abel Berger]

YYEEESSS !!!, The problem was in the cable ! ! !

I would like to be thankful all, and especially for Rick Jones, my new friend, for helping me in this very important problem for my company.
I wait to see you many other times and to be able also helps them.
Keep in contact ! !

Abel Berger
Unix Specialist
Telef?nica do Brazil
abel_berger@yahoo.com