[transport:manager] tunnel profile missing channelId or…

Error message

[transport:manager] tunnel profile missing channelId or corePubkey

What it means

For kind === 'tunnel' the manager's raceLanAndTunnel() destructures channelId and corePubkey from the profile and refuses to proceed without both — they are the minimum to address the desktop core through the backend relay (channelId names the tunnel; corePubkey seals the first handshake frame). Without them the E2E-encrypted XChaCha20-Poly1305 path cannot even start.

Solutions

Re-pair the device through the devices/pairing flow so channelId and corePubkey are freshly issued and persisted
Validate the tunnel profile fields before createTransportManager and route the user to re-pairing UI when missing
If the core workspace was reset, delete the stale profile on the client first so pairing starts clean

Defensive patterns

Strategy: validation

Validate before calling

function assertTunnelProfileReady(p: ConnectionProfile): void {
  if (p.kind === 'tunnel' && (!p.channelId || !p.corePubkey)) {
    throw new Error('tunnel profile incomplete — re-pair the device to obtain channelId and corePubkey');
  }
}

Type guard

function isPairableTunnelProfile(p: ConnectionProfile): boolean {
  return p.kind !== 'tunnel' || (typeof p.channelId === 'string' && typeof p.corePubkey === 'string' && p.channelId.length > 0 && p.corePubkey.length > 0);
}

Prevention

Persist channelId + corePubkey together at pairing completion, not incrementally
If the core workspace is reset, clear tunnel profiles on the client in the same operation
Route users to the pairing UI when either field is missing instead of attempting connect

When it happens

Trigger: Device pairing completed only partially (channel allocated but core pubkey never recorded); persisted tunnel profile trimmed by a migration; pairing record for this device deleted on the core side (src/openhuman/security/devices/) while the client kept the profile.

Common situations: Pairing flow interrupted (app closed during pair); core workspace reset wiping devices.db while the phone/iOS client kept its tunnel profile; schema change renaming fields.

Related errors

AI-assisted analysis of tinyhumansai/openhuman@7491200858 (2026-08-17). Data as JSON: /api/errors/c5687cd7ed6d8821. Report an issue: GitHub.

Appendix: source

Thrown at app/src/services/transport/TransportManager.ts:114

    if (kind === 'tunnel') {
      return this.raceLanAndTunnel();
    }

    throw new Error(`[transport:manager] unknown profile kind: ${kind}`);
  }

  /**
   * Race LAN (with 2 s timeout) against Tunnel.
   * Whichever responds to `openhuman.ping` first wins.
   * If LAN wins but later fails, caller should call reset() to re-race.
   */
  private async raceLanAndTunnel(): Promise<CoreTransport> {
    const { rpcUrl, channelId, corePubkey, sessionToken, pairingToken, devicePrivkey } =
      this.profile;

    if (!channelId || !corePubkey) {
      throw new Error('[transport:manager] tunnel profile missing channelId or corePubkey');
    }

    const tunnelToken = sessionToken ?? pairingToken;
    if (!tunnelToken) {
      throw new Error('[transport:manager] tunnel profile missing sessionToken or pairingToken');
    }

    const tunnelTransport = new TunnelTransport(
      this.backendSocketUrl,
      channelId,
      corePubkey,
      tunnelToken,
      devicePrivkey,
      sessionToken ? 'session' : 'pairing'
    );

    if (!rpcUrl) {
      // No LAN URL — tunnel only.

View on GitHub (pinned to 7491200858)