Network Installer Troubleshooting Centre

Use a layer-by-layer sequence

Randomly replacing equipment is expensive and often hides the real fault. A repeatable sequence starts at the physical path, confirms link state and negotiated parameters, then moves through VLAN, addressing, gateway, DNS and application reachability.

No link light

1. Confirm both devices are powered and the intended interfaces are enabled.
2. Inspect patching and cable IDs. Make sure the physical path is the one shown in the documentation.
3. Substitute a known-good patch lead at one end at a time.
4. Check switch port status for err-disable, PoE fault or administrative shutdown.
5. Test the permanent copper link or inspect/test the fibre path.
6. For fibre, verify optic type, wavelength, connector, fibre type and transmit/receive orientation.

Gigabit endpoint negotiates at 100 Mbps

1000BASE-T uses four pairs. Cisco's Catalyst documentation identifies 1000BASE-T ports as four-pair copper interfaces. A link that falls back to Fast Ethernet can therefore be consistent with a pair fault that still leaves a two-pair 100BASE-TX path working. That is not the only cause, so also check device capability, switch configuration and autonegotiation.

Detailed guide: why a Gigabit link runs at 100 Mbps.

CRC or FCS errors increase

CRC/FCS errors are evidence that received frames are corrupted. Check the physical path, connector condition, cable certification, optics and interface statistics at both ends. Do not clear counters before recording the baseline. If errors increase with traffic or movement of a cable, that observation is useful diagnostic evidence.

PoE endpoint is offline

Separate power from data. Check whether the switch detects a PD, what power class or allocation it reports, whether the port is in a PoE fault state and whether total switch budget remains. Then test cabling. A camera can have a valid IP configuration and still be offline because it is not receiving stable power.

Detailed PoE power fault workflow.

Client has an IP address but no network access

1. Record the client's address, prefix/mask, default gateway and DNS server.
2. Check whether the address came from the expected DHCP scope or is static.
3. Test reachability to the local gateway by IP.
4. Check ARP or neighbour discovery if the gateway is local but unreachable.
5. Test a known remote IP to separate routing from DNS.
6. Only then test DNS names and application ports.

WiFi client is connected but slow

Association alone does not prove adequate RF or airtime. Check RSSI, SNR, channel utilisation, retries, data rate, band/channel width and whether the problem follows one client or all clients. Then check the AP's Ethernet uplink, switch errors, PoE state and upstream congestion.

Fibre link is down

Inspect and clean both ends, verify polarity and patch schedule, confirm the optics match the fibre and application, then compare live transmit/receive levels against the transceiver's manufacturer specification. If loss is high or the route is uncertain, OLTS and OTDR evidence can isolate total loss and event location.

Intermittent faults

Intermittent problems require time correlation. Record exact timestamps and compare them with switch logs, interface counters, PoE events, UPS events, WiFi logs and application monitoring. A statement such as "it drops sometimes" is difficult to diagnose; "port Gi1/0/14 logged link down at 10:42:18 while CRC errors rose from 0 to 245" is evidence.

Related resources

Testing & Certification Centre | Copper Handbook | Fibre Handbook | PoE Field Guide | WiFi Handbook