Read a sample: 1. Start With the Link
A user reports that a workstation “keeps losing the network.” The switch port appears to be up. You notice a nonzero error counter and consider replacing the cable.
That may eventually be the right repair. At this moment, however, it is only a possibility. The errors could be old. The workstation could be connected through a patch panel you have not identified. The reported interruption might affect an application while the Ethernet link remains stable. Replacing a cable without establishing those facts could solve the problem, change nothing, or remove evidence you needed.
Ethernet troubleshooting begins with a disciplined distinction: a symptom describes what someone observed; a diagnosis explains why it happened. Between the two lies an investigation.
This chapter gives you a way to organize that investigation. You will build a picture of the link, identify its endpoints and physical path, collect a useful baseline, and make changes only when the evidence and the operating environment justify them.
An Ethernet link is not just a switch port. At minimum, it involves an interface at each end and a path carrying signals between them.
Consider a workstation connected to an access switch:
text
Workstation network interface
|
Workstation patch cable
|
Wall outlet
|
Building cabling
|
Patch panel
|
Switch patch cable
|
Switch interface
The workstation interface and the switch interface are the two endpoints of this Ethernet link. The outlet and patch panel are intermediate connection points in the physical path; they are not additional Ethernet interfaces in the way the workstation and switch ports are.
That distinction matters because a fault anywhere along the path can affect what either endpoint reports. A switch may record receive errors even though the switch itself is functioning properly. The cause could be a damaged cable, a poorly seated connector, a problem at the transmitting endpoint, or another issue affecting the signal. The counter tells you where an observation was recorded, not necessarily which component caused it.
Other links have different paths. A switch-to-switch fiber connection might include transceivers, fiber patch cords, a fiber panel, and building fiber. A short copper connection might run directly between two devices. In every case, start with the same questions:
- What is connected at each end?
- Which interface on each device belongs to this link?
- What physical components lie between those interfaces?
- What information can each endpoint provide?
Do not assume the label on a cable or a port description is current. Labels are valuable starting clues, but they can be wrong or outdated. Confirm the path through available records, neighbor information where applicable, and physical inspection when authorized.
“The network is slow” can describe many conditions. A link may be down, repeatedly dropping, operating at an unexpected speed, or carrying traffic with errors. It may also be healthy while the delay occurs elsewhere.
Your first task is to translate the report into observations that can be tested. Ask what stopped working, when the problem occurred, whether it is continuous or intermittent, and whether other devices are affected. A workstation that cannot reach one remote application presents a different initial picture from a workstation whose Ethernet link repeatedly goes down.
Useful observations include:
- The specific device and interface involved.
- The time and duration of the symptom, as accurately as available.
- Whether the problem affects one device, several devices, or a particular application.
- Whether the link indicator or operating-system connection status changes during the event.
- Any recent changes reported to you, without assuming those changes caused the problem.
These details do not replace interface evidence. They help you decide where to look and provide a time window against which to compare it.
A symptom can also be real without being visible at the moment you inspect the port. An intermittent fault may not be active during your first command. Record that the interface was up when checked, rather than concluding that it has always been up.
A network interface provides several kinds of evidence. The exact fields, labels, and commands depend on the platform, but the categories are broadly useful.
Administrative state reflects whether the interface has been enabled or intentionally disabled in configuration. An administratively disabled interface requires a different investigation from one that is enabled but cannot establish a physical link. Do not enable a disabled port merely because it is down; it may have been disabled deliberately.
Link state indicates whether the interface currently detects an established link according to the device’s reporting. An up state is encouraging, but it does not prove that the path is error-free or that an application is reachable.
Speed and duplex describe how the interface is operating. Speed is the link’s signaling rate, commonly displayed in units such as megabits or gigabits per second. Duplex describes whether transmission and reception can occur simultaneously or must share the medium. An unexpected operational value deserves investigation, but first distinguish what was configured from what the interface actually reports.
Counters record events observed by the interface. Some count frames or bytes; others count conditions such as receive errors, collisions, or discards. A counter is evidence, not a verdict. You need to know what it represents, when it last reset, and whether it is increasing during the problem.
The next chapter examines status combinations in more detail. For now, remember that no single field answers every question. A port can be…