Ethernet duplex problems

The link between the cisco and the MT router is performing very badly. I am seeing many errors on the cisco. MT does not seem to have the capability to report errors as far as I can tell.

See below how the interfaces are set up.

Notice how MT reports full-duplex=no and the cisco reports full-duplex

When I attempt to turn off auto-negotiation on the MT in order to match settings on the 2 devices the link is lost.

Any insight welcome.



Hardware:
Cisco 7204VXR w/ FastEthernet card.

interface FastEthernet2/0
 ip address xxx.xxx.xxx.xxx xxx.xxx.xxx.xxx
 no ip proxy-arp
 duplex full
!

Speed 100 is not an option here for some reason. maybe the interface does not even know how to do 10BaseT


ROS 3.11 on an Intel MB. Intel NICs (on board 10/100/1000)

/interface ethernet> print detail 
Flags: X - disabled, R - running, S - slave 
 0 R  ;;; to WiSpring
      name="ether1" mtu=1500 mac-address=00:0E:0C:xx:xx:15 arp=enabled disable-running-check=yes 
      auto-negotiation=no full-duplex=yes cable-settings=standard speed=100Mbps 

 1 R  ;;; to GBLX
      name="ether2" mtu=1500 mac-address=00:0E:0C:xx:xx:14 arp=enabled disable-running-check=yes 
      auto-negotiation=yes full-duplex=yes cable-settings=default speed=100Mbps



/interface ethernet> monitor ether2
               status: link-ok
       auto-negotiation: done
                   rate: 100Mbps
            full-duplex: no
  default-cable-setting: standard



#sh interfaces FastEthernet 2/0
FastEthernet2/0 is up, line protocol is up
  Hardware is DEC21140A, address is 0012.xxxx.xxxx (bia 0012.xxxx.xxxx)
  Internet address is xxx.xxx.xxx.xxx/30
  MTU 1500 bytes, BW 100000 Kbit/sec, DLY 100 usec,
     reliability 255/255, txload 11/255, rxload 1/255
  Encapsulation ARPA, loopback not set
  Keepalive set (10 sec)
  Full-duplex, 100Mb/s, 100BaseTX/FX
  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 1d12h
  Input queue: 1/75/817/0 (size/max/drops/flushes); Total output drops: 260
  Queueing strategy: fifo
  Output queue: 0/40 (size/max)
  5 minute input rate 677000 bits/sec, 579 packets/sec
  5 minute output rate 4437000 bits/sec, 692 packets/sec
     73264142 packets input, 12223147555 bytes
     Received 68875 broadcasts, 0 runts, 265 giants, 66 throttles
     535192 input errors, 535033 CRC, 0 frame, 0 overrun, 159 ignored
     0 watchdog
     0 input packets with dribble condition detected
     88429926 packets output, 62854707655 bytes, 0 underruns
     7 output errors, 75 collisions, 2 interface resets
     81 unknown protocol drops
     0 babbles, 0 late collision, 0 deferred
     3 lost carrier, 0 no carrier
     0 output buffer failures, 0 output buffers swapped out

start from layer 1.

Change the cable. Change another one and if both cannot resolve it then its not cable.

Check if the cable placed near AC power lines and is it properly shielded.

Change the port for both routers if possible.

If cannot, try placing a L2 switch in between MT and Cisco Router.

The wire is only 5ft long. It does not run by an AC line. It is a UTP cable. I did replace the cable twice, no change.

Since this is a critical link I would rather not put a layer 2 switch and have an additional point of failure.

There seems to be some sore of mismatch between the cisco and the MT when it comes to autonegotiation. The cisco is running at 100 full and the MT is running at 100 half. This is probably because the link drops at 100 full.

Is there a known problem with cisco or MT that would cause this? Should I try getting a different interface?

As a side note: This MT router also connects to a cisco 2950 via the other on board NIC and there are no problems there even that duplex and speed are hard coded on both sides.

You can’t force speed and/or duplex on one side of the link, and not on the other - that’s asking for trouble.

Force both side to 100/full, and you’ll notice you will more than likely loose connectivity. Leave both on auto neogotiate, and they will more than likely run at half duplex.

From what I can see, and from experience, there is nothing wrong on the switch, but the mikrotik is still doing auto neogotiation. As others have said, different cable, different NIC. Yes this is stupid, yes, it does happen, and yes, chances are more than likely it is the cables…

We have to force 100/F on our X86 machines when connected to Cisco gear. I always took it as Cisco had a wierd auto-negotiate issue. http://en.wikipedia.org/wiki/Autonegotiation Interoperability section.

Also to show Ethernet errors in Winbox you will need to upgrade to at least 3.23.

-Louis

We used to have a couple Cisco 7206VXR terminating DS3’s. The FastEthernet ports on them did not like to play nice with anything (including other Cisco products). I also seem to remember that the onboard interface could not back down to 10; as it was only Fast Ethernet. It looks like you’re using an expansion card from the interface name, which I do remember were E(10) or FE(100) only, depending on the PA card you have.

Try a longer cable, as there is a minimum cable length. Most devices don’t seem to care, but some do.

100BaseT - Also known as fast ethernet. Uses RJ-45 connectors. Topology is star. Uses CSMA/CD media access. Minimum length between nodes is 2.5 meters. Maximum number of connected segments is 1024. Maximum number of nodes per segment is 1 (star topology). IEEE802.3 specification.

Excerpt from here:
http://www.comptechdoc.org/independent/networking/guide/nethwethernet.html

You didn’t say which Cisco ethernet card you had in your 7204vxr. Probably you have an older Cisco PA-FE-TX card which only supports 100Mb ethernet. It does not support auto-negotiation. You can set it to either full duplex or half duplex.

You must force the Mikrotik ethernet port to 100Mbps and the correct duplex. You cannot let it auto-negotiate or you will see lots of dropped packets.

Tom

In some situations you need to use a cross-cable. Auto-MDI is usually disabled when using forced duplex settings.