HP switch with Cisco 1605

[Pascal Le Pesant]

I Have a cisco 1605 router linking two differents networks. Each ethernet interface of the router is plugged in an HP switch (a Procurve 4000 and a 2524).

The problem we have is that the ports of the switch plugged to the router are having a lot of Rx errors and Excessive CRC errors.

I have tried to manually change the duplex to from Auto to Auto-10, 10Half or 10Full. But the errors have not stopped to happen.

Thank you !
Pascal

[Arimo Laine_2]

Hi.

What you have is most likely one of the 2 things: Old firmwares and/or Duplex Mismatch. So, first of all make sure that you have the latest FW on all devices.

Note that both Procurves are capabple of 100FDx. You should always set the link the same way on both ends, so if you set the Procurve port to Autonegotiating, accordingly router port must also be Auto.

I'd advise that you set the ports on both devices to 100FDx - or any fixed speed as long as the setting is the same on both ends of the link.

HTH,
Arimo

[Ron Kinner]

The 1605 only supports 2 builtin 10BaseT half duplex Ethernet connections and a WAN card. If setting the switch to 10 half did not help the problem then something else is wrong.

How long are the cables? Are we talking simple patch cables between the router and the switches or do we have to worry about housewiring? Did you make them yourself or did you buy them ready made? If you made them yourself (or are using preinstalled housewiring) did you use the right kind of wire? Is it less than 100 meters long? Have you tried new cables or tested the cables with a cable tester? Do the cables run near any possible source of interference?

What does
sh int e0
and
sh int e1

show on the 1605?

Do a
clear count
and confirm
then wait a few minutes and repeat
sh int e0
sh int e1
then post the results. (Verify that it says Encapsulation ARPA - I'm not sure the 1605 is capable of anything else but it never hurts to check.)

You can test the port on the switch by connecting a PC to the port and then checking to see if you are still getting the errors. Do the same to the router but remember to use a crossover cable. If you don't have a cable tester you can check the cable using the same PC. If it works connected to the switch with a short patch cable but doesn't work when connected in place of the router then the cable is bad.

Ron

[Pascal Le Pesant]

Thank you Ron for the informations,

To answer your questions, the 1605 is connected directly to the switches with one 4 feet and one 6 feet cables. To be sure I have plugged the router using brand new Cables in a sealed bag and tested them with fluke nettool.

I have cleared the counters then transfered a 25 Mb file. Here are the results after:

Router1605#clear counters e0
Router1605#clear counters e1

Transfert of a 25Mb file between the two networks.

Router1605#sh int e0
Ethernet0 is up, line protocol is up
Hardware is QUICC Ethernet, address is 00d0.58a9.0e6e (bia 00d0.58a9.0e6e)
Internet address is yyy.yyy.yyy.yyy/25
MTU 1500 bytes, BW 10000 Kbit, DLY 1000 usec, rely 255/255, load 12/255
Encapsulation ARPA, loopback not set, keepalive set (10 sec)
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:06, output 00:00:03, output hang never
Last clearing of "show interface" counters 00:02:25
Queueing strategy: fifo
Output queue 0/40, 0 drops; input queue 0/75, 0 drops
5 minute input rate 44000 bits/sec, 31 packets/sec
5 minute output rate 498000 bits/sec, 46 packets/sec
11358 packets input, 1589888 bytes, 0 no buffer
Received 161 broadcasts, 0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored, 0 abort
0 input packets with dribble condition detected
20679 packets output, 27012403 bytes, 0 underruns
0 output errors, 10251 collisions, 0 interface resets
0 babbles, 0 late collision, 1119 deferred
0 lost carrier, 0 no carrier
0 output buffer failures, 0 output buffers swapped out

Router1605#sh int e1
Ethernet1 is up, line protocol is up
Hardware is QUICC Ethernet, address is 00d0.58a9.0e6f (bia 00d0.58a9.0e6f)
Internet address is xxx.xxx.xxx.xxx/24
MTU 1500 bytes, BW 10000 Kbit, DLY 1000 usec, rely 255/255, load 1/255
Encapsulation ARPA, loopback not set, keepalive set (10 sec)
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:00, output 00:00:00, output hang never
Last clearing of "show interface" counters 00:03:03
Queueing strategy: fifo
Output queue 0/40, 0 drops; input queue 1/75, 0 drops
5 minute input rate 438000 bits/sec, 40 packets/sec
5 minute output rate 36000 bits/sec, 23 packets/sec
20769 packets input, 27021528 bytes, 0 no buffer
Received 27 broadcasts, 0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored, 0 abort
0 input packets with dribble condition detected
11275 packets output, 1580595 bytes, 0 underruns
0 output errors, 12087 collisions, 0 interface resets
0 babbles, 0 late collision, 149 deferred
0 lost carrier, 0 no carrier
0 output buffer failures, 0 output buffers swapped out


Hope it will help us to find the trouble.

Pascal

[Ron Kinner]

Since you have ruled out cabling and judging from the fact that you have more collisions than you have packets out I would say that the switch is stuck in full duplex mode. Otherwise the router appears to be working normally.

You are getting CRC errors on the switch because when the router (thinking it is on a half duplex circuit) gets an input from the switch (which thinks it is a full duplex circuit so why should it wait for the router to stop talking before sending?) while it is sending it stops sending immediately and drops back and waits a random amount of time before trying again. When it stops sending a packet the resulting fragment is often interpreted by the switch as a full packet (if it is long enough and it maybe if it ends on a byte boundary) and the last few bytes are assumed to be the CRC. Obviously this is not going to match the bit pattern in the fragment so it claims it has found a CRC error. Too short packets are called runts by Cisco.

Since the router can only do 10 half you will have to convince the switches to run 10-half in order to fix your problem. Once you make your changes look at the status of the port and verify that it thinks it is 10 half. Make sure you save your changes. You might even want to reboot it then verify that it stayed at 10-half.

Ron

[Anthony deRito]

It would help to show counters from your switch ports. Are you able to get this information.

[Pascal Le Pesant]

here is the port counter showed by the switch for the port attached to the router. These counters are since the switch is up (About 80 days).

Link Status : Up

Bytes Rx : 1,221,649,664
Bytes Tx : 1,331,992,425
Unicast Rx : 274,121,002
Unicast Tx : 300,386,617
Bcast/Mcast Rx : 118,664
Bcast/Mcast Tx : 15,659,952

FCS Rx : 41
Drops Tx : 0
Alignment Rx : 40
Collisions Tx : 37,787,009
Runts Rx : 0
Late Colln Tx : 0
Giants Rx : 0
Excessive Colln : 2
Total Rx Errors : 81
Deferred Tx : 12,944,622