The link falls back because 2.5GbE is not available across the complete physical path.
On a home NAS, the port label proves only one endpoint capability. The client NIC, switch port, adapter, cable pairs, wall termination, driver, and advertised settings must all support a stable 2.5GbE mode. The fastest diagnosis is to confirm the negotiated rate, identify the weakest link partner, and replace one path element at a time.
Confirm That the Problem Is Link Negotiation, Not File Throughput
First read the negotiated link rate on both endpoints and the switch. A 2.5GbE link can transfer slowly because of disks or SMB, while a true 1GbE negotiation is a physical-layer or advertised-capability problem.
A home-network case showed a 2.5GbE path with file copies near 1GbE even though simultaneous transfers could use more capacity, illustrating that throughput and link rate differ. Do not infer negotiation from one file-copy number.
Record the NIC status, switch port speed, duplex, error counters, and test method. If every interface reports 2.5GbE, continue with storage and protocol tests instead of changing cables blindly.
Check What Both Link Partners Actually Advertise
Auto-negotiation selects a mode that both endpoints advertise. A 2.5GbE NAS connected to a 1GbE switch port, dock, USB adapter, or client NIC will correctly settle at 1GbE.
Ethernet negotiation attempts the fastest common link mode. Check the advertised modes on the NAS and client rather than relying on product family names or a port color.
Bypass intermediate docks and converters, then connect the NAS and one known 2.5GbE client through a confirmed multi-gig switch or direct test link. If the rate rises, reinsert each intermediary until the 1GbE boundary returns.
Isolate the Cable, Wall Jack, and Terminations
A cable marked for high speed can still contain a damaged pair, poor termination, excessive untwist, or a wall segment that differs from the visible patch lead. Link negotiation tests the electrical path that actually exists.
A multi-gig switch discussion notes that cable quality affects negotiation even when both ports support the higher mode. The label alone cannot confirm pair integrity or signal margin.
Replace the complete run with one short known-good cable between the same endpoints. If 2.5GbE appears, test the patch leads, couplers, keystone jacks, and in-wall run separately rather than forcing the speed on an unstable path.
Check Driver, Adapter, and Power-State Limits
A capable NIC can advertise only 1GbE when the wrong driver is loaded, the adapter firmware is unstable, a dock exposes a slower controller, or power management changes the link policy.
Multi-gig upgrades can encounter driver and adapter instability. Confirm the exact chipset, active driver, advertised modes, USB or PCIe connection, and any Energy Efficient Ethernet or green-network setting.
Update or roll back one driver at a time, reboot both link partners, and recheck advertised modes after each change. Avoid forcing 2.5GbE until auto-negotiation works on a known-good path, because a forced setting can hide the original compatibility problem.
Use a One-Variable Test Matrix
Keep one known-good endpoint and change only one component per test. This prevents several partially compatible devices from producing an ambiguous success or failure.
Start with NAS โ known-good 2.5GbE client on a short cable, then add the switch, wall run, dock, and original client one at a time. Record negotiated rate and error counters after every insertion.
The matrix below turns the first changed result into the suspect layer.
| Test Result | Likely Layer | Next Action |
|---|---|---|
| Direct link reaches 2.5GbE | Switch, wall run, or intermediary | Add one component at a time |
| Known-good cable remains at 1GbE | Endpoint capability, driver, or port setting | Inspect advertised modes |
| Rate changes when cable moves | Cable or termination | Replace and test the full run |
| Link is 2.5GbE but copy is near 1GbE | Storage, protocol, or client bottleneck | Run network-only and disk tests |
Stop once the result follows one component consistently. Replacing several items after the rate improves destroys the evidence needed to prevent recurrence.
Verify the Stable Link Before Tuning NAS Performance
After 2.5GbE negotiates, leave the link active under sustained load and inspect errors, renegotiations, and disconnects. A link that repeatedly falls back is not ready for large NAS copies.
Do not use SMB Multichannel or jumbo frames to mask a single unstable port. The existing explanation of multiple SMB network paths applies only after each underlying path is valid.
Run a network-only test in both directions, then a large-file NAS copy and a normal small-file workload. The repair is complete when the rate remains 2.5GbE across reboots, error counters stay clean, and application throughput is limited by the next measured layer rather than link fallback.
Support & Tips
More to Read

Can Plex Share a GPU With Another Docker Container?
Plex and another container can often access the same GPU, but you must test driver support, device mapping, video-engine load, memory, and recovery behavior.

How to Tell Whether a Plex Error Comes From the Client or Server
Reproduce the same item on another client, compare the session path, then collect server evidence only after scope tells you where the failure actually...

How to Configure Plex Cache and Transcode Temporary Storage
Protect persistent Plex state while placing transcode temp files on suitable local storage, then verify cleanup, free space, and restart behavior.

