Ethernet Switches
Network Stops After Adding a Switch: Find the Ethernet Loop
A second cable can create a loop instead of extra capacity. Trace the physical path, restore a stable connection and choose the right protection for the next expansion.
If the network fails just after a switch or extra cable is added, check for a second Ethernet path back to the same local network. A loop can circulate traffic until ordinary devices stop responding. Disconnect the newly added duplicate path, keep one known uplink and trace the cabling before reconnecting anything.
You add ports for a printer and two computers. Within moments, WiFi appears connected but pages stall, cameras disappear and the gateway becomes difficult to reach. The new switch may be working exactly as an unmanaged switch normally does: forwarding traffic without understanding that the cabling now goes round in a circle.
This guide separates a network loop from a slow cable, explains what spanning tree does and helps you decide when an inexpensive desktop switch is sufficient.
Recognise a Loop Before Replacing Equipment
The timing is the best first clue. If several unrelated devices fail together immediately after one connection changes, investigate that change first. Rapidly flashing activity lights can accompany a loop, but busy lights alone do not prove it. A large backup can also produce considerable traffic.
| What you observe | What to check first |
|---|---|
| Many devices fail after a second uplink is connected | A duplicate path between the same switches or LAN |
| One computer negotiates only 100 Mb/s | Cable pairs, connectors and port capability |
| Only internet access fails; local files still open | The gateway, ISP and upstream path |
| A managed port is blocked but the rest works | STP or loop-protection events before changing the setting |
A single slow link is a different investigation. Use the 100 Mb/s Ethernet troubleshooting guide when the problem stays with one cable or endpoint. Do not buy a faster switch to solve an untraced topology fault.
Restore One Uplink, Then Trace Both Cable Ends
Start with the most recent addition. Label and photograph it, then remove that added connection if you can identify it confidently. Keep the known working uplink in place. On a shared business network, coordinate with the person responsible for connectivity before unplugging uncertain cables.
- Identify the new switch, its power supply and its upstream connection.
- Look for two leads between the same pair of switches, or a lead connecting two ports on one switch.
- Check whether two wall outlets return to the same patch panel and switch.
- Remove the unintended second path, then check access to the gateway and a local device.
- Reconnect ordinary endpoints one at a time and record which connection changes the behaviour.
Leave the suspected path disconnected while tracing it. Rebooting everything may briefly hide the pattern without removing the loop. If access returns but one managed port remains disabled, inspect its recorded reason before re-enabling it.
See Why Two Cables Can Be Worse Than One
Trace the Path, Not Just the Extra Cable
Two ordinary links are not an automatic bandwidth upgrade.
01 Unplanned duplicate path
Two active paths can loopRemove the unintended path. Check wall outlets and bridged links as well as visible patch leads.
02 Simple stable expansion
One known uplinkKeep ordinary devices on separate endpoint ports. Use supported switching features for deliberate redundancy.
Two normal Ethernet links between switches do not automatically combine their bandwidth. A supported link aggregation arrangement must be configured correctly at both ends. Otherwise, the additional link may create a loop or be blocked by spanning tree.
Unexpected paths can also pass through a bridged device, a second wall outlet or a wireless bridge. Write down every route between the two areas, including wireless links. The diagram describes the path; it does not claim that a particular traffic volume proves a loop.
STP, Loop Protection and Storm Control Do Different Jobs
Spanning Tree Protocol, or STP, prevents a supported switched network from forwarding through every redundant path at once. Rapid STP, or RSTP, is a related faster-converging version. A blocked redundant port can therefore be evidence that the protection is working.
In UniFi, loop protection is a separate safeguard that can disable a port when a loop is detected. Its recovery behaviour differs from STP and may require the port to be re-enabled after the cause is removed. Storm control limits specified traffic rates; it does not correct the wiring.
Keep STP enabled on supported switching infrastructure and make the intended root-switch design deliberate. Do not enable edge-port or BPDU-guard settings blindly on switch-to-switch uplinks. Also check the exact port: integrated gateway switch ports do not necessarily have the same STP support as a dedicated managed switch.
A blocked port is a reason to inspect the topology and event log. Disabling the protection just to make the warning disappear can restore the fault to the whole network.
Choose the Switch for Capacity and Visibility
A small unmanaged switch can be entirely suitable for adding ordinary desktop ports when the topology is simple and has one uplink. A managed switch becomes more useful when you need VLANs, event logs, deliberate redundancy or visibility into individual ports. Do not assume that every unmanaged model provides managed STP controls.
A compact eight-port Gigabit option for a simple, correctly cabled desktop expansion.
A managed multi-gigabit option when its port speeds, PoE and network controls fit the deployment.
These products serve different budgets and workloads. A 10GbE PoE switch is not required merely to cure a cable loop. Count endpoints, uplink demand and power requirements separately. For faster access points and storage, the Ethernet cable and distance guide helps with the physical link.
Make the Fix Last Beyond Today
Keep a simple port map at the cabinet or in the network record. Label both ends of every uplink and distinguish spare patch leads from active connections. If resilience is required, design the redundant paths with supported switching features rather than adding a second lead experimentally.
A useful support record includes the time of the failure, a sketch of the before-and-after cabling, switch models, affected devices and the relevant port event. If switches also disappear from management, the UniFi adoption guide covers discovery and controller reachability after the physical fault is resolved.
The practical choice: restore a single known path first. Keep a sound unmanaged switch for a simple expansion; choose managed switching when the network needs visibility and controlled redundancy. Faster ports and loop protection solve different requirements.
Frequently Asked Questions
Can one cable bring down the whole LAN?
Yes. A cable that joins two points already connected at Layer 2 can create a loop. The impact depends on the topology and available protection; widespread failure immediately after that connection is a useful clue.
Will two cables between switches make the connection twice as fast?
Not automatically. Link aggregation requires compatible equipment and matching configuration at both ends. An unplanned second cable can create a loop or be blocked by spanning tree.
Should I disable STP if an uplink is blocked?
Inspect the reason first. STP may be preventing a redundant path from circulating traffic. Remove an unintended loop or review the intended redundancy design before changing protection.
Does an unmanaged switch always cause loops?
No. A simple unmanaged switch can work well with one uplink and correctly connected endpoints. The physical path and the model’s actual protection features matter.
Why is the port still disabled after I remove the cable?
Some loop-protection mechanisms leave the port disabled until it is manually re-enabled. Confirm the fault is removed, read the event and follow the exact switch’s recovery procedure.
Is a slow cable the same as a loop?
No. A damaged cable may reduce link speed or cause errors on one path. A loop creates an unintended repeated route and can affect many devices together. Use the failure pattern to choose the investigation.
