· Troubleshooting · About 8 min read

V2Ray Slow Speeds: How to Check Node Load, Network Routes, and Local Settings

Slow speeds aren’t always caused by a node. This guide checks node load, network routes, and local settings using real-connection tests and controlled comparisons, then covers transport and routing adjustments.

At a glance

For users who can connect to a node but experience slow downloads, evening slowdowns, or slowness in just one app. Start with a real-connection test to confirm requests pass through the node, then compare download speeds using the same file. Change only one variable at a time—node, network, or local settings—and use the results to choose your next step. Don’t start by reinstalling the client or changing protocols.

First, tell connection delays from slow downloads

In v2rayN, select a node in the server list, right-click it, and choose “Test server real connection latency.” The wording may vary slightly between versions; look for “real connection,” not Ping or TCP latency. This test sends a request through the node to a test address, so it reflects real-world use better than simply checking whether the server port is reachable. It’s still just one brief request: a low reading doesn’t guarantee fast large-file downloads, and a timeout may only mean the test address is temporarily unreachable.

  1. Keep the test conditions consistent. Use the same device, browser, and permitted test file URL. Pause cloud sync, system updates, and other downloads. Make sure the file is actually routed through the node being tested—not sent directly by a routing rule or served from the browser cache.
  2. Record wait time and transfer speed separately. Note roughly how long it takes from sending the request to receiving the first data, then record the speed during the download. If only the start is slow, check connection setup and the destination’s response. If it starts promptly but stays slow, investigate throughput. Don’t combine the two into a single “latency” figure.
  3. Alternate your tests. Don’t finish testing A and wait an hour before testing B. Alternate A, B, A, B, and run each three times. Record the median and number of timeouts. Track timeouts separately; don’t count them as zero milliseconds.
3 runs
Recommended runs per test
100 MiB
Example comparison file size
8 Mb/s
Equivalent rate at 1 MB/s

If mobile data is limited, use a smaller file, but make sure the transfer lasts long enough to measure—don’t rely on a fleeting peak. Browsers typically show MB/s, while broadband plans usually use Mb/s. The units differ by a factor of eight, so convert them before comparing.

Layer 1: Check for node load or a single-node issue

Keep the device, network, test target, and proxy mode fixed; change only the node. Start by comparing two nodes in the same region, from the same provider, using the same transport. Comparing different regions and transports can show a large difference without revealing what caused it. On desktop, right-click the selected node in v2rayN, choose “Set as active server,” and test again. Start a new download after switching; don’t reuse an existing connection.

Example node comparison log (illustrative data, not measured)
Test timeNode ANode BNext step
10:00 AM7.4 MB/s7.1 MB/sKeep as the off-peak baseline
9:00 PM1.2 MB/s6.8 MB/sKeep A; switch networks and test again

Interpretation: If only A slows down in the evening, change only the network in the next test.

If A recovers on a different network, investigate the original access route first. If A is slow on both networks while B is fine, give the provider the time, node name, and comparison results, and ask them to check for throttling, server load, and the outbound route. Client-side speed tests can’t directly reveal server load.

Also distinguish between “everything is slow” and “one site is slow.” If A downloads a second test file at normal speed, note which destination is slow. Investigate that destination’s response or the route from the node to it, rather than assuming A is overloaded across the board.

Layer 2: Switch access networks to check for route fluctuations

On the same Android phone, keep the same v2rayNG node and test over Wi-Fi, then turn Wi-Fi off and test over mobile data. After switching, make sure the phone is using the new network, then stop and restart the proxy and send a fresh request to the same target. This comparison shows whether performance changes with the access network, but it can’t by itself distinguish wireless interference from congestion upstream at the carrier.

  • Speed returns when you move closer to the router: Check Wi-Fi coverage and other devices using the network. If you can connect by Ethernet on a desktop, test through the same router. If Ethernet is stable but Wi-Fi fluctuates, troubleshoot the local wireless connection first.
  • Both Ethernet and Wi-Fi are slow, but mobile data is fine: If several nodes slow down on the same fixed broadband connection, test a destination that’s normally reachable directly over that connection. This helps distinguish general broadband congestion from an issue with the proxy route. Use only destinations that are normally accessible directly; don’t treat a failed direct connection as a speed of zero.
  • Real-connection results vary widely: Record timeouts and when they occur. System Ping can help track round-trip times, but servers may restrict these responses. A missing Ping reply alone doesn’t prove that proxy traffic is being dropped.

context deadline exceeded

Cause: The operation didn’t finish before the timeout. This log entry alone can’t tell whether the node, destination, or access route timed out.

What to do: Compare log entries around the same timestamp. First, use the same node to reach another destination; then retry the original destination on a different network.

connection reset by peer

Cause: The remote peer or a device along the route reset the current connection. This doesn’t necessarily mean there isn’t enough bandwidth.

What to do: Check the connection address in the log to see where the reset occurred. If it started only after changing transport settings, restore the original configuration and test the same destination again.

Layer 3: Check the proxy entry point, routing rules, and transport settings

If only one device is slow, or only one app on the same device is affected, first confirm that the app is actually using the node you just tested. A successful real-connection test doesn’t mean every browser, download manager, or phone app follows the same route. Check how the traffic enters the proxy before considering a transport change.

  1. Windows: Start with a single proxy mode. In v2rayN, use “System Proxy” → “Automatically configure system proxy” to test a browser that follows the system proxy. Turn TUN off for a baseline. If an app has its own proxy setting, open “Settings” → “Parameter Settings” and check the local listening port. For example, enter 127.0.0.1:10808 and select SOCKS only if the client is actually listening on that address and exposing a SOCKS endpoint. Don’t assume this example is the default. To test TUN, switch to it separately and note the settings you had before the change.
  2. Android: Check whether the app is included in the VPN. In v2rayNG settings, find “Per-app proxy” and check whether the app you’re testing is covered by the current mode. The app list works differently in include and exclude modes. After making changes, restart the proxy service, then fully close and reopen the app to avoid reusing an existing connection.
  3. Routing: Check which rule actually matched. In v2rayN, open “Routing Settings” to review the active rule set. If possible, check the core logs to confirm whether the destination actually used the proxy or connected directly. Download URLs may redirect, so the domain serving the file may differ from the one in the address bar. Correct only the rule that matched incorrectly, then test the same file again; there’s no need to clear the entire routing configuration.
  4. Transport: Compare only configurations provided by the server. VMess and VLESS are proxy protocols, not transport methods like WebSocket or gRPC. Don’t change the transport type, path, or TLS settings on a node that already works; mismatched client and server settings can prevent a connection. If the provider offers two working configurations, keep the region, route, and test time as consistent as possible. If the server also changes, record the result as a comparison of the two configurations as a whole.

Common speed test questions

The real-connection test is only 80 ms. Why does video still stutter?

A fast short request doesn’t mean video segments can keep downloading steadily. Use the same node to test the same larger file over an extended period and note whether the speed drops periodically. Then play the same video at the same resolution through another node. If only video stutters, first check whether the video app is routed through the proxy and whether segment domains match different rules.

The speed test page is fast, but downloads are slow. Which should I trust?

For download issues, trust the actual file and comparable downloads. A speed test page may use a different server, multiple parallel connections, or a direct route. Check the logs to confirm the outbound route for both tasks, then test the same download URL with a different node. Don’t use a speed test peak to predict the speed of a single-file download.

It’s slow every evening. Will updating the subscription help?

Updating a subscription fetches configurations published by the provider; it doesn’t directly increase the bandwidth of existing nodes. First, compare two nodes and two access networks on the same evening. Update the relevant subscription group only if the provider confirms a configuration change or adds a testable node, and keep the old results for comparison.

Should I change DNS first to speed things up?

If the main issue is a slow first page load and the logs show DNS timeouts, check which DNS server the current route uses, then change only that setting and test again. If the domain has resolved and the download has started but throughput remains low, compare nodes and network routes first. Changing DNS won’t directly increase throughput on an established connection.

Client Downloads