Ethernet Switch Not Working Fix: Ports, Loops & Power

Updated 2026-10-09 • By Devon Brooks, CISSP (Network Security & VPN Specialist)

When an unmanaged or smart-managed Ethernet switch stops passing traffic across your LAN, the root cause almost always falls into one of four failure modes: a Layer 2 switching loop triggering a broadcast storm, a failing external DC power adapter sagging under load, a PoE budget overload shutdown, or ESD damage to the uplink port. Isolating the fault takes less than five minutes.

Key Technical Takeaways
  • Unison rapid blinking equals a loop: If every active port LED flashes furiously in synchronous lockstep while internet dies, you have a Layer 2 broadcast storm caused by a switching loop.
  • Test the DC power wall-wart under load: Failing 12V or 5V electrolytic capacitors inside external power bricks allow the power LED to light up while switch ASICs brown out whenever multiple ports link up.
  • Check PoE Max LED warnings: Plugging high-draw PTZ cameras or WiFi 7 access points into an 802.3af/at switch can exceed the total watt budget and shut down lower-priority ports.
  • Verify VLAN PVIDs on managed switches: A factory reset or un-saved running-config loss on a smart switch reverts trunk ports back to default VLAN 1 untagged, breaking segmented subnets.

Diagnosing Layer 2 Switching Loops and Broadcast Storms

Standard unmanaged Ethernet switches lack Spanning Tree Protocol (STP / RSTP 802.1w) and forward broadcast frames (FF:FF:FF:FF:FF:FF) out of every port except the ingress port. If someone accidentally plugs both ends of a patch cord into two ports on the same switch, connects two switches together with redundant cables, or bridges wired and mesh WiFi backhaul improperly, ARP requests and DHCP discovery packets circulate infinitely at wire speed.

Within milliseconds, the switch's packet buffer saturates, MAC address tables thrash continuously, and every port LED blinks at maximum frequency in unison. Even your upstream router's LAN interface will freeze when hit with hundreds of thousands of broadcast packets per second. To isolate a loop, unplug all patch cables except the single uplink to the router. Verify the router uplink LED stabilizes, then reconnect downstream drops one by one, waiting 10 seconds per port until you spot the exact cable that triggers synchronous blinking. On managed switches (such as TP-Link Easy Smart or NETGEAR Plus), enable Loop Prevention or RSTP immediately.

Failing DC Power Bricks and Under-Voltage Symptoms

Desktop 5-port and 8-port switches—like the NETGEAR GS308, TP-Link TL-SG108, or D-Link DGS-108—rely on inexpensive external switching power adapters (typically rated at 5V 0.6A, 9V 0.6A, or 12V 1.0A). After three to five years of continuous heat, the filter capacitors inside the wall adapter bulge and lose capacitance. The adapter may still measure 12.0V open-circuit when unplugged, so the switch's green Power LED illuminates normally, fooling you into thinking power is healthy.

However, as soon as three or four Gigabit ports negotiate 1000BASE-T DSP links, current draw spikes and the DC rail sags below the internal buck regulator dropout threshold, causing all ports to reset in an endless 3-second boot loop. Test with a known-good barrel-plug DC supply of identical voltage, matching polarity (almost universally center-positive), and equal or higher amperage rating.

Resolving PoE Overload Shutdowns and MDI/MDI-X Uplink Faults

On Power over Ethernet (PoE) switches, inspect the front-panel PoE Max status LED. A compact 8-port PoE+ switch often has a total power budget of 60W to 65W. If you connect four 802.3at access points or motorized heated outdoor security cameras that simultaneously draw 18W during infrared boot-up, total load exceeds 65W. The PoE controller automatically cuts power to the highest-numbered port (usually Port 8) to protect the power supply unless port priority is configured in the switch GUI.

If non-PoE uplink connectivity fails between an older ISP gateway and your switch, verify Auto-MDI/MDI-X. While Gigabit 1000BASE-T includes automatic crossover detection in the specification, forcing speed/duplex settings disables Auto-MDI/X on many enterprise Cisco Catalyst and MikroTik ports, requiring either auto-negotiation to be re-enabled or a crossover cable.

Isolating Dead Switch ASICs and ESD-Damaged RJ45 Ports

Ethernet runs travelling near outdoor walls, attic AC handlers, or PoE cameras frequently pick up induced electrostatic discharge (ESD) during thunderstorms. Surge current enters through an RJ45 port, arcs across the discrete magnetics transformer module, and destroys the differential transceiver pins on the internal switch ASIC. Sometimes only a single port dies (staying completely dark when plugged into a known-working laptop, or staying permanently lit when empty), while adjacent ports continue forwarding frames.

To test for dead ports, plug a laptop configured with a static IP (e.g., 192.168.1.222/24) into Port 1 while the router sits on Port 2, and run ping -t 192.168.1.1. Move the laptop plug sequentially across every port on the chassis. Mark any dead port with tape so it is never reused, or replace the switch and install grounded inline RJ45 surge arrestors on vulnerable cable runs.

Ethernet Switch LED Diagnostic Matrix

LED Symptom PatternNetwork BehaviorRoot CauseImmediate Fix
All port LEDs flashing rapidly in unisonTotal LAN/WAN freeze; DHCP failsLayer 2 switching loop / broadcast stormUnplug cables one by one; enable STP/Loop Prevention
All port LEDs blink once every 3 sec & dieDevices show 'Cable Unplugged' cycleFailing external DC power adapter under loadReplace DC power supply (match V, polarity, >= Amps)
Single port LED stays lit with no cableThat port passes zero trafficESD surge blew internal PHY transceiverTape over damaged port or RMA switch
PoE Max LED solid/blinking amberCameras/APs on high ports lose powerTotal PoE wattage budget exceededMove 1–2 devices to PoE injector or higher-budget switch
Link LEDs normal, no DHCP/InternetStatic IP works locally, no gatewayUplink cable dead or wrong VLAN PVIDVerify router uplink port & untagged VLAN 1 config

Switch Recovery & Fault Isolation Checklist

  1. Disconnect all downstream patch cords except the router uplink and verify whether internet returns to the primary router.
  2. Power-cycle the switch by pulling the DC barrel plug for 30 seconds to clear hung MAC learning tables and ASIC buffers.
  3. Test the DC power adapter voltage under active port load or swap with a compatible center-positive power supply.
  4. Reconnect Ethernet drops one port at a time while running a continuous ping to the router gateway (192.168.1.1).
  5. Check PoE wattage consumption in the switch web UI or sum the class ratings (802.3af = 15.4W, 802.3at = 30W) of connected devices.
  6. Enable Loop Prevention / RSTP (802.1w) and save the running configuration to NVRAM on smart-managed switches.

Frequently Asked Questions

Does an unmanaged Ethernet switch have an IP address or login page I can reset?

No. Unmanaged switches (such as the NETGEAR GS105/GS308 or TP-Link TL-SG105) operate purely at Layer 2 using hardware MAC address learning tables and do not have a TCP/IP stack, DHCP client, or web interface. Pulling the power cord for 15 seconds completely clears their volatile MAC table.

Does it matter which port on a switch connects to the router?

On virtually all modern Gigabit unmanaged switches, every port supports Auto-MDI/MDI-X and shares a non-blocking backplane, meaning any port (Port 1, Port 5, or Port 8) can serve as the uplink to your router. Only switches with dedicated SFP fiber cages or explicitly labeled 'Uplink / PD' ports require specific placement.

Can a bad device on one switch port take down the entire network?

Yes. Aside from physical cable loops, cheap USB-C Ethernet docks frequently inject pause frames (IEEE 802.3x Flow Control flood) when a laptop enters sleep mode, halting all switch ports until unplugged. Disabling 802.3x Flow Control on managed switches or unplugging the dock when sleeping prevents this outage.

Related Switch Troubleshooting & Engineering Guides

Reviewed by Devon Brooks, CISSP (Network Security & VPN Specialist)

Part of the Packetsaver Network Engineering Team. All configurations and firmware safety instructions follow vendor-verified RFC and IEEE standards. Read our testing methodology →