Troubleshooting Interfaces and Cabling

Written By FightTheWest

Excessive Noise

This is a problem that can cause transmission errors. Noise errors are caused by a physical media problem. A damaged cable or wrong cable type could cause excessive noise errors to occur.

A high number of Cyclic Redundancy Checks (CRC) errors along with a low number of collisions can indicate noise issues.

Collisions

Too many collisions can cause network congestion due to packets being retransmitted as a result of the collisions.

Malfunctioning NICs in hosts can cause jabber on the network which causes collisions. Too many devices on one network can also cause collisions and duplex mismatch errors between devices can also cause collisions.

Resolving collision errors may involve:

  • Replacing NICs in client computers

  • Creating additional network segments

  • Reconfiguring duplex settings

LATE COLLISIONS

A late collision occurs after the 512th bit (64th byte) of a frame that has been transmitted. The amount of time it takes to send the first 512 bits of a frame is dependent on the network technology:

  • 51.2 microseconds to send over 10Mbps Ethernet segment

  • 5.12 microseconds to send over 100Mbps Ethernet segment

Late collisions can occur as a result of;

  • Duplex mismatch errors

  • Network segment that extends farther than the cable length supports

Mismatches

Duplex mismatch errors can cause a number of problems:

  • Intermittent connectivity

  • Performance problems

  • High number of collisions

  • Late collisions

Duplex mismatch occurs when the ends of a network link are configured with different duplex settings - both ends need to be configured with the same duplex setting.

One symptom of mismatch is the half-duplex side will report late collisions. The full duplex side will report runs, frame check sequences (FCS) and alignment errors.

Duplex mismatches can sometimes be difficult to diagnose. If you suspect that a duplex mismatch error is causing network problems, use the "show interfaces status" command - this verifies the duplex settings for all interfaces on a device.

The "a-" indicates autonegotiation. Autonegotiation is a method of electrical signalling between interfaces to enable the automatic configuration of speed and duplex settings on an interface.

If one side of the link is statically configured, the autonegotiation-enabled side will attempt to operate at the fastest speed supported - the speed field for autonegotiated port will display auto.

Setting auto-negotiation on only side of a link can cause configuration problems on the link - it is not possible to statically configure the duplex settings of a port unless the port speed is statically configured first.

Speed Mismatch

Speed mismatch errors can prevent an interface from sending and receiving traffic. A speed mismatch occurs when one end of a network link is configured to use a different speed than the other end - the link between the 2 would not be able to be established and remain in a down state.

You can explicitly configure the speed setting on an interface or autonegotiate it - you are allowed to have one end autonegotiation and one static.

In the case where one is static and one is autonegotiated, the autonegotiating port can identify the other port link's speed by the electrical signal sent by the port.

Prevent a speed mismatch from occurring by ensuring that at least one end of a link is configured to autonegotiate the speed settings.

Using OSI Model to Troubleshoot

There are 3 main techniques for troubleshooting using the OSI Model:

  • Bottom Up

  • Top Down

  • Divide and Conquer

The Bottom Up technique starts at the physical layer and works up. Typically you start by checking the cable is plugged in correctly, then check and verify the network card, then IP address etc...

The Top Down technique starts at the application layer and works down. Typically you start by examining or restarting network apps.

The Divide and Conquer technique starts at the network layer and works either up or down depending on the outcome of different tests.

TROUBLESHOOTING PHYSICAL LAYER CONNECTIVITY

To troubleshoot the physical layer, begin by verifying that physical connectivity exists between the router and ISP. There are several ways to verify physical connectivity:

  • Examine the cable

  • Connect to the router and issue the "show interfaces" command

When issued without parameters the show interfaces command displays info about each interface. An interface status of up indicates that the physical interface is working properly. An interface status of down indicate the presence of a Layer 1 issue.

Examples of L1 issues include:

  • A faulty interface

  • Broken cable

  • Incorrect cable

Should use a crossover cable to connect the Ethernet interfaces of two similar devices and a straight through to connect two dissimilar devices.

If the interface is in the administrative down state, issue the "no shut" command.

The "show interfaces" command provides stats that can help diagnose other L1 problems. Many CRC errors on an interface could be indicative of a bad cable. A high number of input/output queue drops could indicate that the router hardware is unable to efficiently process the volume of traffic being sent.

Some Cisco configuration mistakes can create problems. The DCE end of a serial connection provides clocking information to the DTE end. If the correct clock rate is not set on the DTE interface, physical connectivity cannot be established.

The "show controllers serial" command will determine which end is which.

Ethernet interfaces require that the duplex configuration matches on each end of the link - full-duplex or half-duplex. Full duplex means data is being sent from one pair of wires and received by using a different pair of wires.

This prevents collisions from occurring and enables both ends to transmit and receive at the same time.

Most modern devices automatically negotiate the duplex settings for Ethernet, FastEthernet and GigabitEthernet. Duplex mismatches can still occur if the duplex command is manually configured with different nodes on each end. If a high number of collisions are displayed in the output of "show interfaces", a duplex mismatch could be it.

The speed of an interface is automatically negotiated on a modern device - a full duplex Gig interface that is connected to a full duplex Fast interface will negotiate a speed of 100Mbps. If the speed command is issued on each side of the link and speeds don't match, no link is established.

A Fast interface that is manually configured to a speed of 100Mbps will not link with a Gig interface that is manually configured to a speed of 1000Mbps.

Cisco recommends manually configuring speed and duplex settings on links to devices that are not likely to be changed or moved. For most devices, automatic negotiation of speed and duplex should be allowed to occur.

TROUBLESHOOTING DATA LINK LAYER

The "show interfaces" command is also good for verifying the Data Link layer components. A L2 protocol is required to transmit information from one interface to another. Protocols that operate at the Data Link layer include:

  • Ethernet

  • PPP

  • High-Level Data Link Control

  • Frame Relay

An interface status of up combined with a line protocol of down indicates the presence of a L2 problem. Some examples include:

  • Mismatched encapsulation between linked serial devices

  • Clocking errors

  • Lack of keepalive messages

Verify the L2 encapsulation method by examining the "show interfaces" command. By default, a serial interface is configured to use HDLC encapsulation.

The "show interfaces" command is useful for verifying max transmission unit (MTU). The MTU is the largest frame a device can transmit - sometimes also used to describe the largest packet a router can forward.

The default MTU for an Ethernet frame is 1500 bytes. Because an IP packet has a 20 bytes header, the largest payload that can be carried in an Ethernet frame is 1480 bytes.

If a frame exceeds the MTU of a link, the frame will be fragmented if possible or discarded if the DO-NOT-FRAGMENT bit is set.

TROUBLESHOOTING NETWORK LAYER

Troubleshooting L3 is the most involved task. Troubleshooting requires the verification of correct IPv4 and IPv6 addressing - must understand IPv4 and VLSM and IPv6 addressing.

Network layer troubleshooting might involve the examination of routing tables and routing protocol configurations or default gateway configurations.

FightTheWest
Next

Interfaces and Cabling