If a subscription imports but its nodes won’t connect, check the core, transport, and security settings before switching protocols. VMess and Shadowsocks can encrypt traffic at the protocol layer; common VLESS configurations rely on an outer security layer, while classic Trojan handles proxy authentication inside TLS. This guide is for users switching between desktop and Android or comparing nodes on the same network. First confirm the configurations are compatible, then test initial page load, large-file throughput, and any UDP features you need.
Authentication and encryption: what each protocol protects
A node connects only when the client and server use the same protocol, credentials, and related settings. A password or UUID identifies you; it doesn’t mean the connection is encrypted. And seeing TLS in a configuration doesn’t mean the inner protocol skips authentication.
| Protocol | Authentication credentials | Where encryption happens | What to check for compatibility |
|---|---|---|---|
| VMess | UUID; modern configurations use VMess AEAD request headers | Payload protection depends on the configuration. Options include AES-128-GCM and ChaCha20-Poly1305; an outer TLS layer can also be added. | UUID, system time, payload encryption settings, and transport parameters |
| VLESS | Usually a UUID | Common encryption=none configurations don’t encrypt at the protocol layer; connection protection depends on the outer security settings | Security layer, transport, and any server-required flow; check core versions for extension support |
| Trojan | Password-derived authentication data | Classic Trojan relies on TLS to protect authentication data and proxy traffic | Password, TLS server name, certificate trust, and the server’s actual port |
| Shadowsocks | A pre-shared password or key, depending on the method | Modern AEAD methods provide encryption and integrity at the protocol layer; the base protocol doesn’t include TLS | Encryption method, key format, whether both ends support the same specification, and whether UDP is enabled |
VMess AEAD refers to the request-header mechanism; it doesn’t mean the payload can use only one algorithm. Shadowsocks also has different generations: traditional AEAD methods such as aes-128-gcm and Shadowsocks 2022 methods such as 2022-blake3-aes-128-gcm are not interchangeable. The latter requires a pre-shared key in the corresponding format; simply reusing an old password won’t make the configurations compatible.
Transport combinations: the same port doesn’t mean the same connection
The node editor’s Protocol, Transport, and Security fields are separate. For example, raw TCP and WebSocket can both carry the VMess proxy protocol, but they connect differently. If the server expects requests on a WebSocket path, the client must use the same path. A correct UUID can’t make up for a mismatched path.
- Check the address and port first. In the v2rayN node list, double-click the entry and compare it with the details from your subscription provider. Don’t change every port to
443just because TLS is enabled; 443 is the standard HTTPS port, not a fixed port for every TLS service. - Next, check the transport settings. If both ends use WebSocket, compare the path and the Host required by the server. If both use raw TCP, don’t copy WebSocket fields from another node.
- Finally, check the TLS name and certificate. The connection address, TLS server name, and WebSocket HTTP Host each serve a different purpose. If the certificate name doesn’t match, restore the server name specified by the provider. Don’t leave certificate verification disabled just to suppress the error.
Choosing a protocol: keep a working configuration, then compare alternatives
A single subscription can include all four protocols. A successful import only means the client can read the entries; it doesn’t guarantee that the current core supports each node’s transport and extensions. Before comparing, keep one verified node as a baseline so you can tell what changed when you adjust a configuration.
Choose based on available services and maintenance needs
Already have a VMess configuration? Keep the working node
RecommendedIf you already have a stable VMess AEAD node, use it as a baseline. That makes troubleshooting easier than migrating right away. If authentication fails after switching devices, first check that the system is syncing the date and time automatically, then verify the UUID.
Best for: Existing working subscriptions and troubleshooting connection failures after switching devices
VLESS: Match the core and security layer
Switch only if both ends support the same combination and the security layer is configured as required by the server. With common encryption=none configurations, removing the security layer is not a routine troubleshooting step.
Best for: Setups where you can verify the core, transport parameters, and security settings
Trojan: You have a TLS service you can maintain
The certificate must be valid and its name must match, in addition to using the correct password. If a node that used to work starts failing after a domain or certificate update, check the TLS handshake first instead of changing passwords across the board.
Best for: Setups with clear responsibility for certificate and domain maintenance
Shadowsocks: Confirm the method and UDP requirements
Test only after both ends use compatible AEAD methods and key formats. If an app relies on UDP, also confirm that the client can capture its traffic, the server can relay it, and the network supports UDP.
Best for: Setups using a matching method that aim to avoid an extra transport layer
Performance comparison: separate network overhead from device overhead
Slow initial page loads may be caused by DNS resolution, connection setup, or the handshake. If large downloads slow over time, check sustained throughput, packet loss, and device load. When node locations, server load, or test times differ, compare only the actual experience on those nodes; you can’t use that to rank the four protocols by speed.
Choose a test based on the symptoms
Two performance tests for the same destination
Disconnects or slow first loads on one network
- Use the same device, node, and destination, testing on two networks you’re authorized to use.
- Make five new connection attempts per route. Record the number of successes, the median time for successful attempts, and whether failures occur during connection, the TLS handshake, or the destination request.
- Keep proxy capture and routing settings the same throughout testing. If only one network fails, check that network path first; don’t change the protocol and node at the same time.
Slow sustained downloads on a low-powered device
- On the same device and network, alternate between two compatible configurations. If possible, use servers in the same data center with the same egress.
- Download the same test file continuously. Record steady-state throughput, CPU usage by the core process, and whether the device throttles due to heat. Don’t draw conclusions from cached data or a single peak reading.
- Whether a device has AES hardware acceleration affects comparisons between AES-GCM and ChaCha20-Poly1305. Test only methods supported at both ends; don’t guess which is faster from the algorithm name.
Check connection success rate first, then compare initial load time and sustained throughput. Change just one independently controllable variable per round.
When moving a desktop configuration to Android, note the client and core as well. In v2rayN, check which core the node actually uses. v2rayNG uses Xray, while v2flyNG uses V2Fly. Both Android clients can read the same subscription, but that doesn’t mean they both support Xray-specific extensions in it.
Connection troubleshooting: why a node still won’t work after importing a subscription
Four common issues
A VMess node works on an old device but keeps reporting authentication failures on a new computer. Why?
On Windows, open Settings → Time & language → Date & time, turn on automatic time setting, and sync once. Then compare the UUID, transport, and payload security settings with the old device. Don’t change the protocol type to fix an authentication error.
TCPing returns a value, so why won’t the page load?
TCPing only shows that a TCP connection probe to the remote port got a response; it doesn’t test proxy authentication or the destination request. In the v2rayN node list, right-click and run a real connection latency test. If it succeeds but the browser still fails, check the local listening port under Settings → Parameter settings, then verify the browser’s proxy port and current routing rules.
Trojan won’t connect after a certificate update, even though the password hasn’t changed. Why?
Check the core logs for an expired certificate, name mismatch, or trust failure. Then compare the TLS server name with the one provided by the server, and confirm that the server has loaded the new certificate and any required intermediate certificates. If the handshake fails, fix the certificate first; there’s no need to reset credentials.
Shadowsocks loads web pages, but a voice app keeps reconnecting. Why?
A successful web-page test doesn’t prove that UDP is working. First check whether the app relies on UDP, then verify that the client captures the app’s traffic and the server has UDP relay enabled. With system proxy settings alone, an app may not use the proxy at all. If so, start troubleshooting with routing and traffic capture.