Skip to content

[Bug]: T3 Connect tunnel keeps forwarding to the port recorded at link time after the server's port changes #19247

Closed as not planned

Description

@grahamfarrar

Before submitting

  • I searched existing issues and did not find a duplicate.
  • I included enough detail to reproduce or investigate the problem.

Area

apps/server (T3 Connect managed endpoint)

Steps to reproduce

  1. Link a server to T3 Connect while it listens on one port. Mine recorded 127.0.0.1:3774 in userdata/secrets/cloud-endpoint-confirmed-origin.bin ("origin":{"localHttpHost":"127.0.0.1","localHttpPort":3774}).
  2. Restart the server on a different port. Mine now runs as the desktop background service on 3773.
  3. Open T3 Connect from another device.

Expected behavior

On startup the server notices that the confirmed origin no longer matches the port it's listening on, and updates the tunnel's origin (or re-confirms it).

Actual behavior

The tunnel keeps forwarding to the old port. Every relay request fails, and the server log repeats this from 2026-10-11 through every restart until I fixed it by hand on 2026-10-13:

WARN Relay client reported a transport warning
  tunnelName: 't3coderelay-managedendpoint-prod-…'
  output: 'ERR Request failed error="Unable to reach the origin service. … dial tcp 127.0.0.1:3774: connect: connection refused" … dest=https://prod-….t3coderelay.com/api/t3'

Other devices show the environment under T3 Connect with "Managed endpoint health request failed". The server itself is healthy and reachable directly. Restarting doesn't help, and t3 connect link alone didn't help because the stored origin survives it.

A second server on the same Mac, which kept its port, shows "Relay online", so the relay and account are fine.

Workaround

Run t3 connect link --base-dir <dir>, delete userdata/secrets/cloud-endpoint-confirmed-origin.bin, then restart. The server re-provisioned (new tunnel ID), recorded the correct port, and the relay came online.

Impact

Moderate. T3 Connect silently stops working for that environment after any port change. The only sign is a health error on other devices, and nothing in the UI or CLI points to the stored origin.

Version or commit

Server 0.0.45 (stable), macOS.

Environment

Mac Studio running two t3 serve instances with separate base dirs: the desktop background service on 3773, and a second one on 3775. Both are linked to T3 Connect. Related: #18719, where this failing environment first showed up on iOS.

Activity

  1. juliusmarminge commented on Oct 14, 2026

    @juliusmarminge
    Member

    Note

    Grok responding on behalf of Julius.

    Closing as a duplicate of #7458 — same root cause: the T3 Connect managed tunnel keeps forwarding to a stale local origin port after the server moves (service update, second t3 serve, or an explicit port change), while the live server is healthy elsewhere. Please add any new detail (especially the cloud-endpoint-confirmed-origin.bin persistence / t3 connect link not refreshing origin) on that issue.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    duplicateThis issue or pull request already exists

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions