PuppyIP Resource Center
Network & Connection Troubleshooting 9 min Published 2026-11-17

VLESS Relay TCP Connects but Resets Before the Response Header: Check Ports and Xray in Layers

A successful TCP handshake proves that the client reached an entry port. If the connection resets before the VLESS response header, also check the Xray inbound and transport settings.

VLESS Xray TCP RST Relay entry Response header Port troubleshooting

Service eligibility and regional restrictions

PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.

Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.

Key Takeaways

  • A TCP handshake does not prove VLESS authentication, TLS/Reality, transport parameters or destination egress succeeded.
  • For an RST before the response header, align client logs, entry listeners, Xray logs and capture direction by timestamp; do not immediately blame an ISP.
  • If changing the entry port restores the connection, that shows the paths behaved differently but does not identify who sent the reset or why.
  • Before changing a port, back up configuration, confirm firewall access and update the client. The new port still needs full protocol verification.

TCP reachability and a working VLESS session are different layers

A successful TCP connection test to a relay address proves only that the TCP handshake reached some listener at that time. A VLESS session also requires matching inbound protocol, client identity, transport and any enabled TLS or Reality options. If the connection resets before a protocol response header appears, the proxy service is not yet proven usable, and a TCP-only test cannot rule out a server-side configuration issue.

Record the client error stage, entry port, protocol/transport mode and exact time first. Then inspect the server listener and matching Xray logs. An RST before the response header can come from several boundaries; without server access, system logs or captures from both ends, the sender cannot be identified. Do not label the symptom as ISP filtering, DDoS protection or invalid credentials without evidence.

Start with read-only TCP and listener checks

On a Windows client, use Test-NetConnection to check the target TCP port. On an authorized Linux relay, use ss to inspect the listening address and process. Test the same hostname and port used by the service configuration, and remember that DNS may return multiple addresses. Server-side commands are only for a host you manage or are authorized to inspect.

If the TCP test fails, check name resolution, routing, firewalls, security groups and the entry process. If it passes, continue at the protocol layer instead of declaring recovery. Hide real addresses, port mappings and server names before sharing results. Never send a client UUID, password or subscription link.

Check TCP reachability from Windows

powershell
Test-NetConnection -ComputerName EXAMPLE.RELAY -Port REPLACE_WITH_CONFIGURED_PORT
In client PowerShell, test the configured relay host and entry port. Success proves only the TCP test passed; it does not prove a VLESS session or UDP.

Inspect the listener on an authorized Linux relay

sh
PORT='REPLACE_WITH_CONFIGURED_PORT'
ss -ltnp "sport = :$PORT"
Run this read-only check only on a relay you are authorized to manage. Replace the placeholder with its actual listener port and redact addresses and process details.

Align client and server protocol logs by timestamp

Enable controlled client diagnostics and inspect only the connection time, destination port and failure stage. On the server, check whether the Xray inbound received the connection, whether it logged an identity/protocol/TLS or transport error, and whether the process restarted at the same time. Unit names differ across distributions; confirm the actual unit in the service manager before reading its time range.

If the client logs show that TCP connects, waits for a protocol response and then receives an RST, while the server has no matching connection log, that alone does not prove an intermediate network drop. Listener log level, the pre-NAT address and log collection scope also affect what is visible. If the server reports an authentication or transport failure, compare the exact fields on both sides. Do not change the UUID, transport, TLS and port together.

Read logs for the confirmed Xray service

sh
journalctl -u REPLACE_WITH_CONFIRMED_XRAY_UNIT --since '10 minutes ago' --no-pager
Use only on a systemd host after confirming the real unit name. Keep the matching time window and redact error lines; do not share full logs or credentials.

Use a port change only as a controlled comparison

Before comparing another entry port, export the Xray configuration and record the original port, bind address, transport parameters, TLS/Reality options and client settings. Confirm that the new port is unused, that the firewall and cloud security group allow the TCP flow, and that an administrative path remains available. Applying a port change can interrupt sessions; do not close the old entry before both configurations are ready.

After changing the port, point the client to the new entry and repeat the same VLESS handshake, response-header and real-request checks. If the new port works, that proves the path result changed; it does not independently prove that an ISP reset the old port. Listener, policy or intermediary state can also differ. Do not present a working port as a universal way around service or network policies.

Verify an actual application request, not just the port

Protocol acceptance requires the actual client configuration to establish a VLESS session and return a valid response from an authorized test destination. Record TCP reachability, whether the client completed its handshake, whether Xray accepted the session and the destination result separately. If UDP is required, test it through an application path that supports UDP. A TCP test, port scan or green dashboard alone cannot replace these checks.

If server access or logs are unavailable, limit the conclusion to client-side observations and list the missing evidence. Do not say that a provider reset the connection or that the issue is fixed when one-sided observation cannot support that statement. Keep software versions and error stages, but remove domains, real IP addresses, user identifiers and full connection links.

Rollback and stop conditions

If the new port is not listening, client parameters do not match or firewall changes affect management access, restore the original listener configuration and client settings from backup, then retest the old port. A port change alters the exposed service surface; open only the entry required by the service, not a broad range.

If timestamps, client/server settings or connection logs cannot be aligned, stop repeated changes and obtain redacted logs or captures from the same time window on both ends. Keep the reset sender marked unknown unless evidence identifies it. Do not generalize one client's result to every network, protocol or relay.

Sources

Frequently Asked Questions

Does a successful Test-NetConnection mean the VLESS node is healthy?

No. It checks the TCP handshake, not VLESS identity, TLS/Reality, transport parameters, destination egress or UDP.

Does an RST before the response header prove ISP blocking?

No. Compare client and server timestamps, listener state and, if needed, captures from both ends. Without server-side evidence, the sender remains unknown.

If a new port works, does that prove the original port was blocked?

No. The paths behaved differently, but listener, policy or intermediary state may also differ. Verify the full protocol session and server records.

Can I send my full VLESS link to support?

Do not send a complete link. It can contain a UUID, password or other sensitive settings. Share only a redacted protocol type, time, port and error stage.