Sudden WireGuard speed cap (~100 Mbps) since early September (docker/Gluetun)

[Jommes 1]

I've been using AirVPN via WireGuard for a long time, but I've recently hit a sort of speed cap and I'm trying to figure out if something changed on the backend routing or if I'm hitting a new local bottleneck.

My setup:

• Hardware: Synology DS920+ NAS (Intel Celeron J4125)

• Software: Docker, running Gluetun (V3.38 and the latest 3.41.3)

• Home connection speed: 200 Mbps Down / 200 Mbps Up (Fiber).

My problem: Up until September 2nd, I was consistently hitting 180-200 Mbps through the WireGuard tunnel. Since that date, my download speeds have flatlined at exactly ~100 Mbps (around 11.5 to 13 MB/s).

To locate the issue, I set up two separate Speedtest trackers in Docker over the past weeks:

1. Direct Internet (bypassing VPN): Consistently holds 202 Mbps (see attached graph).

2. Via AirVPN (Gluetun with WireGuard): Flatlines strictly at ~100 Mbps since early September (see attached graph).
    

Troubleshooting steps I've already taken:

• Upgraded Gluetun to the absolute latest build to ensure the newest server lists and TCP optimizations are active

• Tested multiple servers across Germany and The Netherlands

• Manually set wireguard mtu from 1280 (which was perfect) to 1420

• Verified CPU usage on the NAS. Gluetun uses the userspace WireGuard implementation (since Synology DSM lacks kernel support). CPU sits at around 60% during full load. While this is high, the exact same CPU/userspace setup handled 200 Mbps perfectly fine before September 2nd.
   

Does anyone know if there were any routing, load balancing, or WireGuard-specific changes made on the AirVPN servers in early September that could cause this artificial 100 Mbps flatline?
Any suggestions on what else I could tweak on my end are appreciated

[Staff 10645]

Hello!

The sustained change you describe deserves investigation, especially since the same setup previously reached 180–200 Mbps. The plateau alone does not establish an imposed bandwidth cap, and we cannot yet link it to a particular server-side change around September 2.

Since you have already tried different servers, Gluetun versions and MTUs, we suggest two controlled comparisons before changing more settings:

1. On the NAS, run the direct and VPN tests one after the other, using the same Speedtest application/version and a manually selected Speedtest server ID. Pause other downloads/uploads and record both download and upload speeds. Please confirm whether your two trackers already use the same test server, and verify that the VPN test shows an AirVPN exit IP.
2. If another computer is available, preferably connected by Ethernet to the same router, test AirVPN with a native WireGuard client outside Docker. Use the same AirVPN server, endpoint port and Speedtest destination as the NAS test. Stop the Gluetun tunnel for this comparison, or use a separate AirVPN device key so the two clients do not share a key while connected.

If the second computer reaches approximately 200 Mbps while the NAS remains near 100, that would narrow the investigation toward the NAS/container path. If both show the reduction, we should investigate the shared network path and VPN endpoint further.

During the NAS test, please also check CPU usage per core and any CPU limit configured for the container. An overall reading of 60% does not rule out one core reaching its limit, although it does not by itself explain why performance changed.

Please share the exact AirVPN server names and endpoint ports tested, the Speedtest server ID, download/upload results, and test times with your time zone. Those details will let us correlate your measurements with the relevant servers. Take care not to post your WireGuard private key or full configuration file.

Kind regards