Interface and Cable Issues

What it is

When a link misbehaves, the cause is usually the cable, the interface settings, or the device at the other end.

Cisco devices keep counters that tell you what is going wrong, if you know how to read them.

This topic covers how to read interface status and error counters, and how to recognise the most common problems: collisions, bad cables, and a speed or duplex mismatch.

How it works

Start with the interface status, then read the error counters, then check settings at both ends.

Reading interface status

The first check is the status of the interface. Two words are shown: the status of the physical line, then the status of the line protocol (the Layer 2 side).

show ip interface brief
show interfaces GigabitEthernet0/1
show interfaces status

The common combinations are:

A switch port can also be err-disabled, which means the switch shut the port down because of a detected problem, for example a security violation. show interfaces status shows this state.

Reading error counters

show interfaces prints counters. These are the ones to understand.

• Input errors: the total of all receive-side errors.
• CRC: frames that arrived damaged. The checksum at the end of the frame did not match. Causes include a bad or damaged cable, electrical interference, a faulty port, or a duplex mismatch.
• Runts: frames shorter than the 64-byte minimum. They are often a result of collisions or a faulty device.
• Giants: frames larger than the maximum allowed size, normally 1518 bytes for a standard Ethernet frame, or 1522 bytes when an 802.1Q tag is present. A sender configured for a larger MTU can cause these.
• Collisions: happen when two devices on a half-duplex link transmit at once.
• Late collisions: a collision detected after the first 64 bytes of a frame have been sent. This is not normal on a properly working link. The usual causes are a duplex mismatch or a cable that is too long.
• Output errors and drops: problems on the transmit side. A rising count can point to congestion.

You can reset counters with clear counters before you test.

Speed and duplex mismatch

Speed and duplex can be set automatically or by hand. Auto-negotiation lets the two ends agree on the best speed and duplex they both support. It works well when both ends are set to auto.

Trouble starts when one end is set by hand and the other is left on auto. The auto end can usually detect the speed, but it cannot detect the duplex of a manually set partner, and on 10 and 100 Mbps links it falls back to half duplex.

If the other end is fixed at full duplex, the result is a duplex mismatch.

A duplex mismatch has a recognisable pattern. The link comes up and a ping may even work, but real traffic is slow.

The half-duplex side shows collisions and late collisions. The full-duplex side shows CRC errors and runts. The user complaint is "the network is slow", not "the network is down".

A speed mismatch is simpler. If the two ends cannot agree on a common speed, the link usually does not come up at all.

The settings are configured on the interface:

interface GigabitEthernet0/1
 speed 100
 duplex full

The best practice is to leave both ends on auto, or to set both ends to the same fixed values.

Cable-related checks

If counters point to a physical problem, work from the cable outward.

1. Check that the cable is firmly seated, undamaged and the right type.
2. Check that the run does not exceed its length limit.
3. Swap in a known good cable, or move the connection to another port, to see whether the fault follows the cable, the port or the device.

Many Catalyst switches also have a built-in cable test. Run test cable-diagnostics tdr interface INTERFACE, then view the result with show cable-diagnostics tdr interface INTERFACE.