How to Read v2rayN Latency Tests: Ping, Real Connection Latency, and Download Speed

Learn what v2rayN’s three speed tests measure, why ping can work while connections fail, and which results to use when choosing a node.

At a Glance

If the node list shows low latency but webpages still won’t load, first check whether the figure measures probe time to the server or the time to reach a test URL through the node. When choosing a node, use real connection tests to rule out unreliable options, then compare response times or download throughput for your needs. To troubleshoot, use the logs to find where the connection fails.

What Ping, Real Connection Latency, and Download Speed Tests Measure

Ping and TCP Probes

The Windows command-line ping tool measures ICMP round-trip time. In v2rayN’s node list, the menu item “Test server latency (TCP ping)” typically measures the time needed to establish a TCP connection to the node address and specified port. Neither test verifies the proxy protocol itself. For UDP-based transports, a TCP probe also can’t stand in for an actual connection test.

Best for: Checking host responsiveness or TCP port reachability, depending on the probe type.

Real Connection Latency

Recommended

The core uses the node being tested to access the configured test URL and records how long the request takes. The result may include node connection setup, security negotiation, proxy egress, and the test site’s response, depending on the request and whether the connection is reused. It confirms that this particular test request worked, not the latency to every website.

Best for: Finding candidates for everyday browsing and distinguishing an open port from a working proxy.

Download Speed Test

The client requests a test resource through the node and measures the transfer rate over the test window. Results depend on your local network, node load, the route to the destination, and any speed limits imposed by the test server. A fast initial response doesn’t guarantee sustained throughput, and a short test can’t predict the speed of a long download.

Best for: Comparing large-file downloads and sustained transfer performance.

In the v2rayN 7.x desktop app, identify the options in the server-list context menu as “TCP ping,” “Real connection latency,” and “Speed.” Labels may vary by minor version and interface language. Before comparing results, check the real connection test URL and speed test URL under Settings → Parameter settings. If you’ve just changed the test method, retest the candidates before sorting so you don’t mix an earlier TCP result with a new real connection result.

How to Read Milliseconds and Download Speeds

The readings below are examples, not measurements or pass/fail thresholds. Suppose the same node returns these results on the same network. Each figure describes a different process, so they shouldn’t be combined into one ranking.

35 ms
TCP Probe Example
180 ms
Real Connection Latency Example
8 MB/s
Download Speed Example

How to Compare: Track Failures, Then Compare Successful Results

If node A returns 95 ms, then times out twice across three tests, while node B returns 160, 175, and 190 ms, try B for everyday use first. Don’t rank A using only its 95 ms result while ignoring the timeouts, and don’t count timeouts as 0 ms.

You can’t calculate security handshake time by subtracting 35 from 180: the probes have different purposes and timing scopes, and a real connection test also includes the route from the node to the test site. For download speeds, check the units: using decimal units, 8 MB/s is about 64 Mb/s. The first is measured in bytes, the second in bits. If the client uses another unit, convert it before comparing with your broadband plan’s advertised speed.

Ping Works but Connections Fail: Troubleshoot by Failure Stage

Why Does Command-Line Ping Work While the Real Connection Test Times Out?

First, confirm the node’s domain name and actual server port. An ICMP reply only means a probe response was received. If the node uses TCP and its server port is 443, run Test-NetConnection for the domain in PowerShell and specify -Port 443, then check TcpTestSucceeded. If the port isn’t 443, use the actual port. Even a successful test only confirms that a TCP connection can be established.

Why Does the Real Connection Test Fail Even Though the TCP Test Succeeded?

Test just this node, note the time, then check the client logs for errors at the same time. For certificate errors, verify the system clock and server name. For transport negotiation failures, check the transport settings against the provider’s parameters, including the path used by that transport. If the issue started after updating the subscription, compare the old and new configurations to confirm the port, security settings, and user ID were all updated.

Why Won’t Webpages Load in My Browser Even Though the Real Connection Test Shows a Result?

Set the node that passed the test as the active server, then confirm that the system proxy is configured and your browser follows that setting. A successful speed test only confirms that its test request worked; it doesn’t mean your browser is using that node. If only one domain fails, reproduce the issue once and check the relevant logs to see whether the request was routed through the proxy, sent directly, or blocked. Then check how that domain resolves.

Why Did All Nodes Suddenly Stop Returning Test Results? Is the Subscription Invalid?

First, use a previously reliable node to open a site you normally visit, then compare that result with the speed test. If browsing works but the test fails, check whether the test URL is available, or switch to a URL that allows probes and responds reliably, then retest the same node. If regular browsing also fails, look for the earliest DNS, connection, or core startup error in the logs. Don’t reimport the subscription based only on failure indicators in the list.

Choose a Node for Your Needs: Web Response or Sustained Throughput

Two Ways to Compare the Same Candidates

Web Browsing and Lightweight Requests
  • On the same network and test URL, run three real connection tests per node. Record the number of failures and the variation between successful results.
  • For candidates that pass consistently, switch to each as the active node and test with websites you commonly use. Keep the browser and routing mode consistent; don’t compare a page that loads instantly from cache with its first load.
  • If two nodes perform similarly, favor the one that keeps working reliably in the evening rather than chasing a single lowest reading.
Large Files and Sustained Transfers
  • Run download speed tests only on candidates that pass the real connection test. Use the same resource and test window for each node, testing one at a time to avoid competing for local bandwidth.
  • For the top-performing nodes, test sustained speeds with a real downloadable resource. Watch for drops and retries instead of focusing only on the initial peak.
  • If results vary significantly between morning and evening, test again during the hours you actually use the connection. Compare candidates tested at the same time, rather than comparing peak-hour readings with off-peak results.

For browsing, focus on response times and failure rates. For large downloads, focus on sustained throughput. You don’t need one ranking for both.

What to Record When Speed Test Results Fluctuate

Change just one condition at a time when retesting. Pause background downloads, keep your wired or wireless connection the same, and retest the original node. Don’t change the network, test URL, and node all at once, or you won’t know what caused the difference.

  1. Record the test conditions: Note the client and core versions, time of day, connection type, probe type, and test URL. For download results, also record the test window and speed units.
  2. Keep failed results: Record when timeouts occur and save the corresponding logs. If only one of three tests succeeds, list its response time separately instead of summarizing it as “average latency.” Before sharing logs, remove subscription URLs, user IDs, and other credentials.
  3. Include a comparison node: On the same network and test URL, if only one node slows down, check its configuration or route first. If several slow down at once, compare access to the actual sites you use to distinguish a local network change from fluctuations at the test destination.
Download v2rayN