chore(release): prepare v1.1.2

nixpkgs update with Bisq update
fix(public-ip): move system.activationScripts under the config attribute
2026-08-21 09:47:43 -05:00 · 2026-08-20 17:04:36 -05:00 · 2026-08-20 17:00:23 -05:00 · 2026-08-20 16:40:00 -05:00 · 2026-08-20 16:12:00 -05:00 · 2026-08-20 14:07:06 -05:00
35 changed files with 1665 additions and 225 deletions
@@ -7,6 +7,36 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

 ---

+## [1.1.2] - 2026-08-21
+
+### Added
+- Nixpkgs update with Bisq update
+- Unify public-IP detection into one privacy-first script
+- Fix Element X discovery and harden federated calling
+- Nixpkgs update
+- Add tor-browser to the package list
+
+### Changed
+- Use pkgs.lndinit, drop local packages/lndinit
+- Update RTL to 0.15.10
+
+### Fixed
+- Move system.activationScripts under the config attribute
+- Reuse Hub external IP for LiveKit, drop egress service calls
+- Derive Restart required from boot default vs running system
+- Reconcile completed updates after polling stalls
+- Migrate i2pd SAM settings for nixpkgs 26.11
+- Order turn-setup after network-online to fix boot-time red dot
+- Restore a Zeus-scannable LND REST connect QR
+
+### Security
+- Prevent LND admin macaroon exposure in curl argv
+
+### Documentation
+- Update desktop and Sovran Hub screenshots
+[1.1.2]: https://git.sovransystems.com/Sovran_Systems/Sovran_SystemsOS/releases/tag/v1.1.2
+
+
 ## [1.1.1] - 2026-08-15

 ### Added
@@ -43,6 +73,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
    manual re-pin.

 ### Fixed
+  - Zeus Connect QR in the Hub was invalid: the LND-only `lndconnect`
+    wrapper used flags Zeus/`lndconnect` do not understand (`--cert`,
+    `--macaroon`), read a non-existent onion path (`free/lnd.onion`),
+    and collided with LND's P2P hidden service. The wrapper now matches
+    the nix-bitcoin contract — dedicated `lnd-rest` onion on port 8080,
+    `--nocert` over Tor, admin macaroon in the URI — so scanning the Hub
+    QR in Zeus works again.
  - The Hub launcher now uses a persistent per-user browser profile
    (`$XDG_STATE_HOME/sovran-hub-browser`) instead of a throwaway `/tmp`
    profile that was deleted on exit. The throwaway profile wiped the
@@ -68,6 +105,23 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
    polling failures instead of hanging indefinitely.
  - Rebuild/update scripts stream `nixos-rebuild` output into the live log
    (it was buffered until completion, making long rebuilds look frozen).
+  - The update modal now reconciles persisted update state after a page/RDP
+    reconnect, bounds every status request with a timeout, prevents overlapping
+    async polls, and reports an explicit "status unavailable" state with a
+    Retry Status action instead of spinning forever. Verbose Nix logs are
+    rendered incrementally and bounded so they cannot stall the browser UI.
+  - Successful staged updates now record the exact NixOS generation. The Hub
+    keeps showing "Restart required" until that generation is active, then
+    clears the marker after reboot (with log-based recovery for older updates).
+  - Pending-reboot state is now derived from NixOS itself — the boot default
+    (`/nix/var/nix/profiles/system`) versus the running `/run/current-system`
+    — instead of from Hub-written marker files. Updates performed from a
+    terminal or support session (which never touch the Hub's status files)
+    previously left a stale `REBOOT_REQUIRED` marker the Hub could never
+    clear, pinning the "Restart required" badge forever even after many
+    reboots. The marker self-heals to `IDLE` on the next status read whenever
+    boot default and running system agree, and recording the informational
+    `.generation` marker can no longer fail an otherwise successful update.

 ---

@@ -21,12 +21,12 @@ Lightning infrastructure, private cloud, and communications platform when you
 are ready.

 [Visit the Website](https://sovransystems.com) ·
-[Download the ISO](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.1.iso) ·
+[Download the ISO](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.2.iso) ·
 [Try it safely in a VM](#try-it-first-in-a-virtual-machine) ·
-[Verify the Download](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.1.iso.sha256) ·
+[Verify the Download](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.2.iso.sha256) ·
 [Build from Source](#build-from-source)

-<img src="assets/desktop-screenshot.png" alt="Sovran_SystemsOS private Bitcoin desktop" width="800" />
+<img src="assets/desktop-screenshot.webp" alt="Sovran_SystemsOS private Bitcoin desktop" width="800" />

 *Bitcoin sovereignty from the first boot.*

@@ -281,6 +281,10 @@ From one place, the Hub helps you:
 - Reach your Bitcoin tools, private cloud, and communications
 - Perform supported system operations without everyday terminal commands

+<img src="assets/sovran-hub-screenshot.webp" alt="The Sovran Hub dashboard" width="800" />
+
+*The Sovran Hub: manage your private infrastructure from one place.*
+
 ### Example home setup

 ```text
@@ -341,8 +345,8 @@ with an imaging application such as [Balena Etcher](https://etcher.balena.io).

 ### 1. Download the ISO and checksum

- [Download Sovran_SystemsOS-1.1.1.iso](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.1.iso)
- [Download Sovran_SystemsOS-1.1.1.iso.sha256](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.1.iso.sha256)
+- [Download Sovran_SystemsOS-1.1.2.iso](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.2.iso)
+- [Download Sovran_SystemsOS-1.1.2.iso.sha256](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.2.iso.sha256)

 The download may take some time. Do not rename or modify the ISO before
 verifying it, and keep both files in the same folder.
@@ -360,16 +364,16 @@ checksum exactly.
 Open a terminal in the download folder and run:

 ```bash
-sha256sum --check Sovran_SystemsOS-1.1.1.iso.sha256
+sha256sum --check Sovran_SystemsOS-1.1.2.iso.sha256
 ```

 A successful comparison reports:

 ```text
-Sovran_SystemsOS-1.1.1.iso: OK
+Sovran_SystemsOS-1.1.2.iso: OK
 ```

-You can also run `sha256sum Sovran_SystemsOS-1.1.1.iso` and compare the output
+You can also run `sha256sum Sovran_SystemsOS-1.1.2.iso` and compare the output
 against the checksum file manually.

 </details>
@@ -380,11 +384,11 @@ against the checksum file manually.
 Open Terminal in the download folder and run:

 ```bash
-shasum -a 256 Sovran_SystemsOS-1.1.1.iso
+shasum -a 256 Sovran_SystemsOS-1.1.2.iso
 ```

 Compare the value shown in Terminal with the value inside
-`Sovran_SystemsOS-1.1.1.iso.sha256`.
+`Sovran_SystemsOS-1.1.2.iso.sha256`.

 </details>

@@ -394,7 +398,7 @@ Compare the value shown in Terminal with the value inside
 Open PowerShell in the download folder and run:

 ```powershell
-Get-FileHash .\Sovran_SystemsOS-1.1.1.iso -Algorithm SHA256
+Get-FileHash .\Sovran_SystemsOS-1.1.2.iso -Algorithm SHA256
 ```

 Compare the value under `Hash` with the published checksum.
@@ -409,7 +413,7 @@ match exactly.

 1. Download and install [Balena Etcher](https://etcher.balena.io), then
   connect the USB drive.
-2. Choose **Flash from file** and select `Sovran_SystemsOS-1.1.1.iso`.
+2. Choose **Flash from file** and select `Sovran_SystemsOS-1.1.2.iso`.
 3. Choose **Select target**, select the USB drive, and review your selection
   carefully.
 4. Choose **Flash** and wait for the writing and verification process to
@@ -852,7 +856,7 @@ primary location for collaboration. Please read our
 ## Privacy. Sovereignty. Bitcoin.

 [Visit Sovran Systems](https://sovransystems.com) ·
-[Download Sovran_SystemsOS](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.1.iso) ·
+[Download Sovran_SystemsOS](https://downloads.sovransystems.com/Sovran_SystemsOS-1.1.2.iso) ·
 [View the License](LICENSE)

 </div>
@@ -1 +1 @@
-1.1.1
+1.1.2
@@ -21,6 +21,7 @@ import subprocess
 import tempfile
 import threading
 import time
+import sys
 import urllib.error
 import urllib.parse
 import urllib.request
@@ -55,6 +56,7 @@ from .security_helpers import (
    load_session_store,
    save_session_store,
 )
+from .update_state import effective_update_status

 logger = logging.getLogger(__name__)

@@ -66,6 +68,7 @@ GITEA_API_BASE = "https://git.sovransystems.com/api/v1/repos/Sovran_Systems/Sovr

 UPDATE_LOG        = "/var/log/sovran-hub-update.log"
 UPDATE_STATUS     = "/var/log/sovran-hub-update.status"
+UPDATE_GENERATION = "/var/log/sovran-hub-update.generation"
 UPDATE_UNIT       = "sovran-hub-update.service"

 REBUILD_LOG    = "/var/log/sovran-hub-rebuild.log"
@@ -132,6 +135,7 @@ _SERVICE_DOMAIN_KEYS = frozenset([
 ])

 INTERNAL_IP_FILE = "/var/lib/secrets/internal-ip"
+EXTERNAL_IP_FILE = "/var/lib/secrets/external-ip"
 ZEUS_CONNECT_FILE = "/var/lib/secrets/zeus-connect-url"

 ONBOARDING_FLAG = "/var/lib/sovran/onboarding-complete"
@@ -487,9 +491,9 @@ SERVICE_DESCRIPTIONS: dict[str, str] = {
        "Sovran_SystemsOS makes running a production-grade payment gateway as simple as flipping a switch."
    ),
    "zeus-connect-setup.service": (
-        "Connect the Zeus mobile wallet to your Lightning node via LND REST. Send and receive "
+        "Connect the Zeus mobile wallet to your Lightning node via LND REST over Tor. Send and receive "
        "Lightning payments from your phone using a direct node connection. "
-        "Scan the QR code to add your node to Zeus — this gives full node admin access."
+        "Scan the QR code to add your node to Zeus, then enable Use Tor — this gives full node admin access."
    ),
    "mempool.service": (
        "Your own blockchain explorer and mempool visualizer. Monitor transactions, "
@@ -950,21 +954,43 @@ def _save_internal_ip(ip: str):
            pass


-def _get_external_ip() -> str:
-    MAX_IP_LENGTH = 46
-    for url in [
-        "https://api.ipify.org",
-        "https://ifconfig.me/ip",
-        "https://icanhazip.com",
-    ]:
+def _save_external_ip(ip: str):
+    """Write the external IP to a file so other services (e.g. LiveKit) can
+    reference it without running their own detection."""
+    if ip and ip != "unavailable":
        try:
-            req = urllib.request.Request(url, method="GET")
-            with urllib.request.urlopen(req, timeout=8) as resp:
-                ip = resp.read().decode().strip()
-                if ip and len(ip) < MAX_IP_LENGTH:
-                    return ip
+            os.makedirs(os.path.dirname(EXTERNAL_IP_FILE), exist_ok=True)
+            with open(EXTERNAL_IP_FILE, "w") as f:
+                f.write(ip)
+        except OSError:
+            pass
+
+
+def _get_external_ip() -> str:
+    """Public IP via the shared detector (/var/lib/sovran/public-ip.py).
+
+    The detector owns discovery (STUN -> DNS -> opt-in HTTPS echo), caches the
+    result in /var/lib/secrets/external-ip, and contacts at most one third
+    party per refresh interval. This function only reads the cache and asks
+    the detector to refresh when it is missing or stale — it performs no
+    per-call external queries of its own.
+    """
+    try:
+        r = subprocess.run(
+            [sys.executable, "/var/lib/sovran/public-ip.py", "check"],
+            capture_output=True, text=True, timeout=20,
+        )
+        if r.returncode == 0 and r.stdout.strip():
+            return r.stdout.strip().splitlines()[0]
    except Exception:
-            continue
+        pass
+    try:
+        with open(EXTERNAL_IP_FILE) as f:
+            ip = f.read().strip()
+        if ip:
+            return ip
+    except OSError:
+        pass
    return "unavailable"


@@ -1532,10 +1558,16 @@ def _evaluate_domain_checklist(

 def _generate_qr_png_bytes(data: str, scale: int = 6, margin: int = 2) -> bytes | None:
    """Generate a QR code PNG and return the raw bytes.
-    Uses qrencode CLI (available on the system via credentials.nix)."""
+    Uses qrencode CLI (available on the system via credentials.nix).
+
+    High error-correction (H) is preferred for short payloads. Long
+    lndconnect URIs can exceed version-40 capacity at H, so fall back to
+    quartile then low ECC — otherwise the Hub shows an empty Zeus QR.
+    """
+    for ecc in ("H", "Q", "L"):
        try:
            result = subprocess.run(
-            ["qrencode", "-o", "-", "-t", "PNG", "-s", str(scale), "-m", str(margin), "-l", "H", data],
+                ["qrencode", "-o", "-", "-t", "PNG", "-s", str(scale), "-m", str(margin), "-l", ecc, data],
                capture_output=True, timeout=10,
            )
            if result.returncode == 0 and result.stdout:
@@ -1548,9 +1580,10 @@ def _generate_qr_png_bytes(data: str, scale: int = 6, margin: int = 2) -> bytes
 def _generate_qr_svg(data: str, scale: int = 10, margin: int = 4) -> str | None:
    """Generate a QR code SVG document (resolution-independent, ideal if the
    user wants to embed the QR in a website or print it at any size)."""
+    for ecc in ("H", "Q", "L"):
        try:
            result = subprocess.run(
-            ["qrencode", "-o", "-", "-t", "SVG", "-s", str(scale), "-m", str(margin), "-l", "H", data],
+                ["qrencode", "-o", "-", "-t", "SVG", "-s", str(scale), "-m", str(margin), "-l", ecc, data],
                capture_output=True, timeout=10,
            )
            if result.returncode == 0 and result.stdout:
@@ -1627,15 +1660,6 @@ def _nwc_lnurl_bech32(alias: str, domain: str) -> str:

 # ── Update helpers (file-based, no systemctl) ────────────────────

-def _read_update_status() -> str:
-    """Read the status file. Returns RUNNING, SUCCESS, REBOOT_REQUIRED, FAILED, or IDLE."""
-    try:
-        with open(UPDATE_STATUS, "r") as f:
-            return f.read().strip()
-    except FileNotFoundError:
-        return "IDLE"
-
-
 def _write_update_status(status: str):
    """Write to the status file."""
    try:
@@ -1645,6 +1669,36 @@ def _write_update_status(status: str):
        pass


+def _read_update_status() -> str:
+    """Read and reconcile the persistent update status.
+
+    ``REBOOT_REQUIRED`` survives Hub/browser restarts before the reboot, but
+    it is a CLAIM about live NixOS state, not the source of truth: the boot
+    default (``/nix/var/nix/profiles/system``) versus the running
+    ``/run/current-system``.  Re-validating on every read keeps the Hub
+    correct when the system was updated from a terminal or support session
+    (which never writes Hub markers), and lets an old marker that predates
+    the reconciliation feature self-heal instead of demanding reboots
+    forever.  The stale marker file is removed once cleared.
+    """
+    try:
+        with open(UPDATE_STATUS, "r") as f:
+            status = f.read().strip()
+    except FileNotFoundError:
+        return "IDLE"
+
+    effective = effective_update_status(status)
+    if effective != status:
+        _write_update_status(effective)
+        try:
+            os.remove(UPDATE_GENERATION)
+        except OSError:
+            pass
+        return effective
+
+    return status
+
+
 def _read_log(offset: int = 0) -> tuple[str, int]:
    """Read the update log file from the given byte offset.
    Returns (new_text, new_offset)."""
@@ -1680,6 +1734,9 @@ def _resolve_credential(cred: dict) -> dict | None:
            qr_data = _generate_qr_base64(result["value"])
            if qr_data:
                result["qrcode"] = qr_data
+            else:
+                # Don't hide the URI if we could not render a scannable QR.
+                qronly = False
        if qronly:
            result["qronly"] = True
        return result
@@ -1714,6 +1771,9 @@ def _resolve_credential(cred: dict) -> dict | None:
        qr_data = _generate_qr_base64(value)
        if qr_data:
            result["qrcode"] = qr_data
+        else:
+            # Don't hide the URI if we could not render a scannable QR.
+            qronly = False

    if qronly:
        result["qronly"] = True
@@ -3682,6 +3742,9 @@ async def api_network():
    # Keep the internal-ip file in sync for credential lookups
    _save_internal_ip(internal)
    _cached_external_ip = external
+    # Persist the external IP so other services (e.g. LiveKit) can reuse the
+    # Hub's detection instead of running their own.
+    _save_external_ip(external)
    return {"internal_ip": internal, "external_ip": external}


@@ -3819,9 +3882,15 @@ async def api_ports_health():
@app.get("/api/updates/check")
 async def api_updates_check():
    loop = asyncio.get_event_loop()
+    status = await loop.run_in_executor(None, _read_update_status)
+    if status in {"RUNNING", "REBOOT_REQUIRED"}:
+        # Avoid a slow remote update check when there is already an operation
+        # the dashboard needs to surface.
+        return {"available": True, "status": status.lower()}
+
    available = await loop.run_in_executor(None, check_for_updates)
    # None means inconclusive (check failed) — report as available so the UI doesn't block
-    return {"available": available is not False}
+    return {"available": available is not False, "status": status.lower()}


@app.get("/api/ping")
@@ -6168,10 +6237,19 @@ async def _startup_migrate_deprecated_features():

 async def _background_domain_reachability_checker():
    """Periodically curl configured domains and cache reachability results."""
+    global _cached_external_ip
    await asyncio.sleep(_DOMAIN_REACHABILITY_STARTUP_DELAY)
    consecutive_failures = 0
    while True:
        try:
+            # Keep the persisted external IP fresh (dynamic WAN IPs), so
+            # services like LiveKit can read /var/lib/secrets/external-ip.
+            loop = asyncio.get_event_loop()
+            external = await loop.run_in_executor(None, _get_external_ip)
+            if external != "unavailable":
+                _cached_external_ip = external
+                _save_external_ip(external)
+
            cfg = load_config()
            services = cfg.get("services", [])

@@ -6181,7 +6259,6 @@ async def _background_domain_reachability_checker():
                if unit is not None
            }

-            loop = asyncio.get_event_loop()
            overrides, *_ = await loop.run_in_executor(None, _read_hub_overrides)

            domains_to_check: list[str] = []
@@ -5,10 +5,16 @@
 const POLL_INTERVAL_SERVICES    = 5000;
 const POLL_INTERVAL_UPDATES     = 1800000;
 const UPDATE_POLL_INTERVAL          = 2000;
-// Max consecutive failed rebuild/update status polls before the page gives up
-// waiting and reloads to re-sync (2s interval → ~2 minutes of failures).
-// A brief Hub restart during activation only causes a handful of failures.
-const STATUS_POLL_MAX_FAILURES  = 60;
+// A pending fetch never rejects by itself. Bound every status request so a
+// wedged browser connection cannot leave the modal spinning forever.
+const STATUS_POLL_FETCH_TIMEOUT      = 15000;
+// Eight timed-out requests plus the poll interval is a little over two minutes.
+// A brief Hub restart or a heavily loaded Nix build remains well inside this.
+const STATUS_POLL_MAX_FAILURES       = 8;
+// Keep verbose Nix output from making textContent updates quadratic and
+// freezing the Hub renderer (especially noticeable over RDP).
+const UPDATE_VISIBLE_LOG_MAX_CHARS   = 250000;
+const UPDATE_VISIBLE_LOG_TRIM_CHARS  = 200000;
 const REBOOT_CHECK_INTERVAL     = 5000;
 const REBOOT_FETCH_TIMEOUT      = 12000;
 const REBOOT_REQUEST_TIMEOUT    = 4000;
@@ -6,6 +6,16 @@
 if ($btnCloseModal) $btnCloseModal.addEventListener("click", closeUpdateModal);
 if ($btnReboot) $btnReboot.addEventListener("click", doReboot);
 if ($btnSave) $btnSave.addEventListener("click", saveErrorReport);
+if ($btnRetryUpdate) $btnRetryUpdate.addEventListener("click", retryUpdateStatus);
+
+// Browser timers and requests may be suspended while an RDP session/tab is in
+// the background. Reconcile immediately when the user returns instead of
+// waiting for the next interval.
+window.addEventListener("focus", resumeUpdateStatusAfterInterruption);
+window.addEventListener("online", resumeUpdateStatusAfterInterruption);
+document.addEventListener("visibilitychange", function() {
+  if (document.visibilityState === "visible") resumeUpdateStatusAfterInterruption();
+});
 if ($credsCloseBtn) $credsCloseBtn.addEventListener("click", closeCredsModal);
 if ($supportCloseBtn) $supportCloseBtn.addEventListener("click", closeSupportModal);

@@ -248,6 +258,11 @@ async function init() {
    setInterval(checkUpdates, POLL_INTERVAL_UPDATES);
    loadAutolaunchToggle();
  }
+
+  // If the page was reloaded or the RDP/browser session resumed during an
+  // update, reopen the modal from the persisted backend state. This also
+  // surfaces a completed update that is waiting for its activation reboot.
+  await restoreUpdateModalIfNeeded();
 }

 document.addEventListener("DOMContentLoaded", init);
@@ -144,3 +144,22 @@ async function apiFetch(path, options) {
  }
  return res.json();
 }
+
+async function apiFetchWithTimeout(path, options, timeoutMs) {
+  var controller = new AbortController();
+  var fetchOptions = Object.assign({}, options || {});
+  fetchOptions.signal = controller.signal;
+  var timer = setTimeout(function() { controller.abort(); }, timeoutMs);
+  try {
+    return await apiFetch(path, fetchOptions);
+  } catch (err) {
+    if (controller.signal.aborted) {
+      var timeoutError = new Error("Request timed out");
+      timeoutError.name = "TimeoutError";
+      throw timeoutError;
+    }
+    throw err;
+  } finally {
+    clearTimeout(timer);
+  }
+}
@@ -8,6 +8,7 @@ function openRebuildModal() {
  _rebuildLogOffset = 0;
  _rebuildServerDown = false;
  _rebuildFinished = false;
+  _rebuildPollInFlight = false;
  _rebuildPollFailures = 0;
  if ($rebuildLog) { $rebuildLog.textContent = ""; $rebuildLog.style.display = "none"; }
  var action = _rebuildIsEnabling ? "Enabling" : "Disabling";
@@ -34,6 +35,7 @@ function appendRebuildLog(text) {
 }

 function startRebuildPoll() {
+  if (_rebuildPollTimer) clearInterval(_rebuildPollTimer);
  pollRebuildStatus();
  _rebuildPollTimer = setInterval(pollRebuildStatus, UPDATE_POLL_INTERVAL);
 }
@@ -43,9 +45,14 @@ function stopRebuildPoll() {
 }

 async function pollRebuildStatus() {
-  if (_rebuildFinished) return;
+  if (_rebuildFinished || _rebuildPollInFlight) return;
+  _rebuildPollInFlight = true;
  try {
-    var data = await apiFetch("/api/rebuild/status?offset=" + _rebuildLogOffset);
+    var data = await apiFetchWithTimeout(
+      "/api/rebuild/status?offset=" + _rebuildLogOffset,
+      { cache: "no-store" },
+      STATUS_POLL_FETCH_TIMEOUT
+    );
    _rebuildPollFailures = 0;
    if (_rebuildServerDown) { _rebuildServerDown = false; }
    if (data.log) appendRebuildLog(data.log);
@@ -61,10 +68,8 @@ async function pollRebuildStatus() {
  } catch (err) {
    _rebuildPollFailures += 1;
    // The Hub restarts itself during activation, which briefly drops this poll.
-    // If polling stays broken long past a normal restart, the page's session
-    // almost certainly no longer matches the server (e.g. the Hub restarted
-    // and sessions were not recovered).  Reload to re-authenticate and show
-    // the real feature state instead of spinning forever.
+    // If polling stays broken long past a normal restart, reload to
+    // re-authenticate and show the resulting feature state.
    if (_rebuildPollFailures >= STATUS_POLL_MAX_FAILURES) {
      _rebuildFinished = true;
      stopRebuildPoll();
@@ -72,6 +77,8 @@ async function pollRebuildStatus() {
      return;
    }
    if (!_rebuildServerDown) { _rebuildServerDown = true; if ($rebuildStatus) $rebuildStatus.textContent = "Applying changes…"; }
+  } finally {
+    _rebuildPollInFlight = false;
  }
 }

@@ -11,9 +11,9 @@ function _getZeusConnectGuideHtml() {
      '<div class="nwc-connect-step"><div class="nwc-step-num">2</div><div>Open Zeus and open the <strong>Wallets</strong> screen.</div></div>' +
      '<div class="nwc-connect-step"><div class="nwc-step-num">3</div><div>Tap the <strong>+ (Add Wallet)</strong> button in the top-right corner.</div></div>' +
      '<div class="nwc-connect-step"><div class="nwc-step-num">4</div><div>On <strong>Wallet Configuration</strong>, tap the <strong>scan icon</strong> in the top-right corner, then scan the QR code above.</div></div>' +
-      '<div class="nwc-connect-step"><div class="nwc-step-num">5</div><div>Zeus detects the LND REST QR and fills in the connection details. Review them, then tap <strong>Save Wallet Config</strong>.</div></div>' +
+      '<div class="nwc-connect-step"><div class="nwc-step-num">5</div><div>Zeus detects the LND REST QR and fills in the connection details. Turn <strong>Use Tor</strong> on (the host is a .onion address), then tap <strong>Save Wallet Config</strong>.</div></div>' +
    '</div>' +
-    '<div class="nwc-connect-note"><strong>💡 Note:</strong> This is <em>not</em> the same as the NWC pairing QR shown in Lightning Wallet Connections — that gives your wallet sandboxed, limited access for everyday spending. LND REST connects Zeus directly to your node for full admin control.</div>' +
+    '<div class="nwc-connect-note"><strong>💡 Note:</strong> This is <em>not</em> the same as the NWC pairing QR shown in Lightning Wallet Connections — that gives your wallet sandboxed, limited access for everyday spending. LND REST connects Zeus directly to your node for full admin control. The QR uses your dedicated LND REST Tor address (no TLS cert) so Zeus can scan and connect over Tor.</div>' +
  '</div>';
 }

@@ -6,9 +6,12 @@ let _servicesCache    = [];
 let _categoryLabels   = {};
 let _updateLog        = "";
 let _updatePollTimer  = null;
+let _updatePollInFlight = false;
 let _updateLogOffset  = 0;
+let _updateVisibleLogChars = 0;
 let _serverWasDown    = false;
 let _updateFinished   = false;
+let _updateStatusUnavailable = false;
 let _updatePollFailures = 0; // consecutive failed update-status polls
 let _supportTimerInt      = null;
 let _supportEnabledAt     = null;
@@ -24,6 +27,7 @@ let _featuresData         = null;
 let _rebuildLog           = "";
 let _rebuildLogOffset     = 0;
 let _rebuildPollTimer     = null;
+let _rebuildPollInFlight  = false;
 let _rebuildFinished      = false;
 let _rebuildServerDown    = false;
 let _rebuildPollFailures  = 0; // consecutive failed rebuild-status polls
@@ -48,6 +52,7 @@ const $modalStatus    = document.getElementById("modal-status");
 const $modalLog       = document.getElementById("modal-log");
 const $btnReboot      = document.getElementById("btn-reboot");
 const $btnSave        = document.getElementById("btn-save-report");
+const $btnRetryUpdate = document.getElementById("btn-retry-update-status");
 const $btnCloseModal  = document.getElementById("btn-close-modal");

 const $rebootOverlay  = document.getElementById("reboot-overlay");
@@ -271,10 +271,19 @@ async function checkUpdates() {
  try {
    var data = await apiFetch("/api/updates/check");
    var hasUpdates = !!data.available;
+    var updateStatus = data.status || "idle";
    var sidebarUpdateBtn = document.getElementById("sidebar-btn-update");
    var sidebarUpdateHint = document.getElementById("sidebar-update-hint");
    if (sidebarUpdateBtn) {
-      if (hasUpdates) {
+      if (updateStatus === "reboot_required") {
+        sidebarUpdateBtn.style.borderColor = "#e5a50a";
+        sidebarUpdateBtn.style.backgroundColor = "rgba(229, 165, 10, 0.10)";
+        if (sidebarUpdateHint) sidebarUpdateHint.textContent = "Restart required";
+      } else if (updateStatus === "running") {
+        sidebarUpdateBtn.style.borderColor = "#3584e4";
+        sidebarUpdateBtn.style.backgroundColor = "rgba(53, 132, 228, 0.10)";
+        if (sidebarUpdateHint) sidebarUpdateHint.textContent = "Update in progress…";
+      } else if (hasUpdates) {
        sidebarUpdateBtn.style.borderColor = "#2ec27e";
        sidebarUpdateBtn.style.backgroundColor = "rgba(46, 194, 126, 0.08)";
        if (sidebarUpdateHint) sidebarUpdateHint.textContent = "Updates available!";
@@ -2,20 +2,45 @@

 // ── Update modal ──────────────────────────────────────────────────

-function openUpdateModal() {
+async function openUpdateModal() {
  if (!$modal) return;
-  apiFetch("/api/updates/check")
+
+  // Reattach before checking for new updates. This makes a browser reload,
+  // RDP reconnect, or suspended tab recover the authoritative systemd-backed
+  // state instead of starting over or claiming the system is merely up to date.
+  try {
+    var current = await apiFetchWithTimeout(
+      "/api/updates/status?offset=0",
+      { cache: "no-store" },
+      STATUS_POLL_FETCH_TIMEOUT
+    );
+    if (current.running || current.result === "reboot_required") {
+      showExistingUpdate(current);
+      return;
+    }
+  } catch (_) {
+    // The normal start path below has its own visible error handling.
+  }
+
+  apiFetchWithTimeout(
+    "/api/updates/check",
+    { cache: "no-store" },
+    STATUS_POLL_FETCH_TIMEOUT
+  )
    .then(function(data) {
      if (!data.available) {
        stopUpdatePoll();
        _updateLog = "";
        _updateLogOffset = 0;
+        _updateVisibleLogChars = 0;
        _updateFinished = true;
+        _updateStatusUnavailable = false;
        if ($modalLog) $modalLog.textContent = "";
        if ($modalStatus) $modalStatus.textContent = "✓ System is already up to date";
        if ($modalSpinner) $modalSpinner.classList.remove("spinning");
        if ($btnReboot) $btnReboot.style.display = "none";
        if ($btnSave) $btnSave.style.display = "none";
+        if ($btnRetryUpdate) $btnRetryUpdate.style.display = "none";
        if ($btnCloseModal) $btnCloseModal.disabled = false;
        $modal.classList.add("open");
        return;
@@ -27,23 +52,69 @@ function openUpdateModal() {
    });
 }

-function _doOpenUpdateModal() {
+function prepareUpdateModal() {
  if (!$modal) return;
+  stopUpdatePoll();
  _updateLog = "";
  _updateLogOffset = 0;
+  _updateVisibleLogChars = 0;
+  _updatePollInFlight = false;
  _serverWasDown = false;
  _updateFinished = false;
+  _updateStatusUnavailable = false;
  _updatePollFailures = 0;
  if ($modalLog) $modalLog.textContent = "";
  if ($modalStatus) $modalStatus.textContent = "Starting update…";
  if ($modalSpinner) $modalSpinner.classList.add("spinning");
  if ($btnReboot) $btnReboot.style.display = "none";
  if ($btnSave) $btnSave.style.display = "none";
+  if ($btnRetryUpdate) $btnRetryUpdate.style.display = "none";
  if ($btnCloseModal) $btnCloseModal.disabled = true;
  $modal.classList.add("open");
+}
+
+function _doOpenUpdateModal() {
+  prepareUpdateModal();
  startUpdate();
 }

+function showExistingUpdate(data) {
+  prepareUpdateModal();
+  if (data.log) appendLog(data.log);
+  _updateLogOffset = Number(data.offset) || 0;
+
+  if (data.running) {
+    if ($modalStatus) $modalStatus.textContent = "Updating…";
+    startUpdatePoll();
+    return;
+  }
+
+  _updateFinished = true;
+  if (data.result === "reboot_required") {
+    onUpdateDone("reboot_required");
+  } else if (data.result === "success") {
+    onUpdateDone(true);
+  } else {
+    onUpdateDone(false);
+  }
+}
+
+async function restoreUpdateModalIfNeeded() {
+  if (!$modal || $modal.classList.contains("open")) return;
+  try {
+    var data = await apiFetchWithTimeout(
+      "/api/updates/status?offset=0",
+      { cache: "no-store" },
+      STATUS_POLL_FETCH_TIMEOUT
+    );
+    if (data.running || data.result === "reboot_required") {
+      showExistingUpdate(data);
+    }
+  } catch (_) {
+    // Dashboard startup must remain usable when status cannot be reached.
+  }
+}
+
 function closeUpdateModal() {
  if (!$modal) return;
  $modal.classList.remove("open");
@@ -53,21 +124,36 @@ function closeUpdateModal() {
 function appendLog(text) {
  if (!text) return;
  _updateLog += text;
-  if ($modalLog) { $modalLog.textContent += text; $modalLog.scrollTop = $modalLog.scrollHeight; }
+  if ($modalLog) {
+    // Appending a text node avoids reparsing/replacing the complete log on
+    // every two-second poll. Trim only occasionally once the visible log is
+    // large; the complete _updateLog remains available for error reports.
+    if (_updateVisibleLogChars + text.length > UPDATE_VISIBLE_LOG_MAX_CHARS) {
+      var tail = _updateLog.slice(-UPDATE_VISIBLE_LOG_TRIM_CHARS);
+      var notice = "[Earlier update output hidden from this view; it remains in the saved report.]\n\n";
+      $modalLog.textContent = notice + tail;
+      _updateVisibleLogChars = notice.length + tail.length;
+    } else {
+      $modalLog.appendChild(document.createTextNode(text));
+      _updateVisibleLogChars += text.length;
+    }
+    $modalLog.scrollTop = $modalLog.scrollHeight;
+  }
 }

 function startUpdate() {
-  fetch("/api/updates/run", { method: "POST" })
-    .then(function(response) {
-      if (!response.ok) return response.text().then(function(t) { throw new Error(t); });
-      return response.json();
-    })
+  apiFetchWithTimeout(
+    "/api/updates/run",
+    { method: "POST" },
+    STATUS_POLL_FETCH_TIMEOUT * 2
+  )
    .then(function(data) {
      if (data.status === "no_updates") {
        if ($modalStatus) $modalStatus.textContent = "✓ System is already up to date";
        if ($modalSpinner) $modalSpinner.classList.remove("spinning");
        if ($btnReboot) $btnReboot.style.display = "none";
        if ($btnSave) $btnSave.style.display = "none";
+        if ($btnRetryUpdate) $btnRetryUpdate.style.display = "none";
        if ($btnCloseModal) $btnCloseModal.disabled = false;
        _updateFinished = true;
        return;
@@ -83,6 +169,7 @@ function startUpdate() {
 }

 function startUpdatePoll() {
+  if (_updatePollTimer) clearInterval(_updatePollTimer);
  pollUpdateStatus();
  _updatePollTimer = setInterval(pollUpdateStatus, UPDATE_POLL_INTERVAL);
 }
@@ -92,33 +179,45 @@ function stopUpdatePoll() {
 }

 async function pollUpdateStatus() {
-  if (_updateFinished) return;
+  // setInterval does not wait for an async callback. The guard prevents a slow
+  // request from creating overlapping, out-of-order status polls.
+  if (_updateFinished || _updatePollInFlight) return;
+  _updatePollInFlight = true;
  try {
-    var data = await apiFetch("/api/updates/status?offset=" + _updateLogOffset);
+    var data = await apiFetchWithTimeout(
+      "/api/updates/status?offset=" + _updateLogOffset,
+      { cache: "no-store" },
+      STATUS_POLL_FETCH_TIMEOUT
+    );
    _updatePollFailures = 0;
    if (_serverWasDown) {
      _serverWasDown = false;
      if (!data.running) {
-        // The update finished while the server was restarting.  Reset to
-        // offset 0 and re-fetch so the complete log is shown from the top.
+        // The update finished while the server or browser connection was away.
+        // Re-fetch from offset 0 so the final result and complete tail agree.
        _updateLog = "";
        _updateLogOffset = 0;
+        _updateVisibleLogChars = 0;
        if ($modalLog) $modalLog.textContent = "";
        try {
-          var fullData = await apiFetch("/api/updates/status?offset=0");
+          var fullData = await apiFetchWithTimeout(
+            "/api/updates/status?offset=0",
+            { cache: "no-store" },
+            STATUS_POLL_FETCH_TIMEOUT
+          );
          if (fullData.log) appendLog(fullData.log);
          _updateLogOffset = fullData.offset;
-        } catch (e) {
-          // If the re-fetch fails, fall through with whatever we have.
+          data = fullData;
+        } catch (_) {
          if (data.log) appendLog(data.log);
          _updateLogOffset = data.offset;
        }
        if (data.result === "reboot_required") {
-          appendLog("[Server restarted — update completed, reboot required.]\n");
+          appendLog("[Reconnected — update completed, reboot required.]\n");
        } else if (data.result === "success") {
-          appendLog("[Server restarted — update completed successfully.]\n");
+          appendLog("[Reconnected — update completed successfully.]\n");
        } else {
-          appendLog("[Server restarted — update encountered an error.]\n");
+          appendLog("[Reconnected — update encountered an error.]\n");
        }
        _updateFinished = true;
        stopUpdatePoll();
@@ -129,7 +228,7 @@ async function pollUpdateStatus() {
        }
        return;
      }
-      appendLog("[Server reconnected]\n");
+      appendLog("[Update status reconnected]\n");
      if ($modalStatus) $modalStatus.textContent = "Updating…";
    }
    if (data.log) appendLog(data.log);
@@ -146,21 +245,57 @@ async function pollUpdateStatus() {
    }
  } catch (err) {
    _updatePollFailures += 1;
-    // Same guard as the rebuild modal: if polling stays broken long past a
-    // normal Hub restart, reload to re-authenticate and show the real state
-    // instead of spinning forever.
    if (_updatePollFailures >= STATUS_POLL_MAX_FAILURES) {
-      _updateFinished = true;
-      stopUpdatePoll();
-      window.location.reload();
+      showUpdateStatusUnavailable();
      return;
    }
-    if (!_serverWasDown) { _serverWasDown = true; appendLog("\n[Server restarting — waiting for it to come back…]\n"); if ($modalStatus) $modalStatus.textContent = "Server restarting…"; }
+    if (!_serverWasDown) {
+      _serverWasDown = true;
+      appendLog("\n[Update status connection interrupted — retrying…]\n");
+      if ($modalStatus) $modalStatus.textContent = "Reconnecting to update…";
+    }
+  } finally {
+    _updatePollInFlight = false;
+  }
+}
+
+function showUpdateStatusUnavailable() {
+  _updateFinished = true;
+  _updateStatusUnavailable = true;
+  stopUpdatePoll();
+  if ($modalSpinner) $modalSpinner.classList.remove("spinning");
+  if ($modalStatus) $modalStatus.textContent = "Update status unavailable — update may still be running";
+  appendLog("\n[The Hub could not confirm update status. The background update was not stopped. Select Retry Status after reconnecting.]\n");
+  if ($btnRetryUpdate) $btnRetryUpdate.style.display = "inline-flex";
+  if ($btnCloseModal) $btnCloseModal.disabled = false;
+}
+
+function retryUpdateStatus() {
+  if (!$modal) return;
+  _updateFinished = false;
+  _updateStatusUnavailable = false;
+  _updatePollFailures = 0;
+  _serverWasDown = true;
+  if ($modalSpinner) $modalSpinner.classList.add("spinning");
+  if ($modalStatus) $modalStatus.textContent = "Reconnecting to update…";
+  if ($btnRetryUpdate) $btnRetryUpdate.style.display = "none";
+  if ($btnCloseModal) $btnCloseModal.disabled = true;
+  startUpdatePoll();
+}
+
+function resumeUpdateStatusAfterInterruption() {
+  if (!$modal || !$modal.classList.contains("open")) return;
+  if (_updateStatusUnavailable) {
+    retryUpdateStatus();
+  } else if (!_updateFinished) {
+    pollUpdateStatus();
  }
 }

 function onUpdateDone(result) {
+  _updateStatusUnavailable = false;
  if ($modalSpinner) $modalSpinner.classList.remove("spinning");
+  if ($btnRetryUpdate) $btnRetryUpdate.style.display = "none";
  if ($btnCloseModal) $btnCloseModal.disabled = false;
  if (result === true) {
    if ($modalStatus) $modalStatus.textContent = "✓ Update complete";
@@ -187,6 +322,7 @@ function saveErrorReport() {
  URL.revokeObjectURL(url);
 }

+
 // ── Reboot ────────────────────────────────────────────────────────

 var _rebootStartTime = 0;
@@ -74,6 +74,7 @@
      <div class="modal-log" id="modal-log" aria-live="polite"></div>
      <div class="modal-footer">
        <button class="btn btn-save" id="btn-save-report" style="display:none">Save Error Report</button>
+        <button class="btn btn-save" id="btn-retry-update-status" style="display:none">Retry Status</button>
        <button class="btn btn-reboot" id="btn-reboot" style="display:none">Restart Entire System</button>
        <button class="btn btn-close-modal" id="btn-close-modal" disabled>Close</button>
      </div>
@@ -0,0 +1,88 @@
+"""Update-state helpers for the Sovran Hub.
+
+The full-system updater stages a NixOS generation with ``nixos-rebuild boot``.
+That generation is not active until the machine reboots — and the same is true
+for updates started from a terminal or an SSH support session, which never go
+near the Hub's status files.
+
+The ONLY reliable indicator that a reboot is pending is NixOS itself: the
+system profile (``/nix/var/nix/profiles/system``), which ``nixos-rebuild``
+points at the newest generation on every ``boot`` AND every ``switch``, versus
+``/run/current-system``, the generation actually running since the last boot.
+When the two differ, a staged generation has not been booted yet.
+
+Earlier revisions reconstructed this from a marker file and log tails written
+by the Hub's own updater.  Any system updated by other means — or whose
+``REBOOT_REQUIRED`` status was written by an updater older than the marker
+feature — left the Hub showing "Restart required" forever: the recorded
+generation could never equal the (since advanced) running one, so the marker
+could never be cleared.
+
+This module has no FastAPI or systemd dependencies so the policy can be tested
+without importing the Hub server.
+"""
+
+from __future__ import annotations
+
+import os
+
+# The NixOS system profile.  ``nixos-rebuild boot`` and ``nixos-rebuild
+# switch`` both add a generation here; ``boot`` additionally makes it the
+# bootloader default.  The path is a symlink chain (``system`` ->
+# ``system-N-link`` -> ``/nix/store/...-nixos-system-...``).
+BOOT_PROFILE_PATH = "/nix/var/nix/profiles/system"
+CURRENT_SYSTEM_PATH = "/run/current-system"
+
+
+def reboot_is_pending(
+    boot_profile_path: str = BOOT_PROFILE_PATH,
+    current_system_path: str = CURRENT_SYSTEM_PATH,
+) -> bool:
+    """Return whether a staged NixOS generation has not been booted yet.
+
+    This is deliberately independent of how the update was started — Hub
+    "Update System", terminal ``nixos-rebuild boot``, or a support session all
+    move the system profile the same way:
+
+    * after ``nixos-rebuild boot``:  profile -> new, current -> old  → pending
+    * after rebooting:               both -> new                     → cleared
+    * after ``nixos-rebuild switch``: both move together             → no reboot
+      ever needed (switch activates immediately)
+    * after a rollback:              both point at the rollback target → cleared
+
+    Unreadable or missing paths are treated as "not pending": the Hub must
+    never demand a reboot it cannot substantiate.
+    """
+    try:
+        boot_default = os.path.realpath(boot_profile_path)
+        current = os.path.realpath(current_system_path)
+    except OSError:
+        return False
+    if not os.path.exists(boot_default) or not os.path.exists(current):
+        return False
+    return boot_default != current
+
+
+def effective_update_status(
+    status: str,
+    boot_profile_path: str = BOOT_PROFILE_PATH,
+    current_system_path: str = CURRENT_SYSTEM_PATH,
+) -> str:
+    """Map a persisted Hub status to the one that reflects live NixOS state.
+
+    Only ``REBOOT_REQUIRED`` is re-validated: it means "the update staged a
+    generation the machine has not booted into", a claim that must stay true
+    no matter which tool performed the last update.  When the boot default IS
+    the running system the claim is stale — the staged generation booted, was
+    superseded by a newer update, or the marker was written by an updater that
+    could never clear it — so the effective status is ``IDLE``.
+
+    All other statuses (``RUNNING``, ``FAILED``, ``SUCCESS``, ``IDLE``) pass
+    through unchanged; RUNNING staleness is handled separately against the
+    systemd unit itself.
+    """
+    if status == "REBOOT_REQUIRED" and not reboot_is_pending(
+        boot_profile_path, current_system_path
+    ):
+        return "IDLE"
+    return status
@@ -5,7 +5,7 @@
  "bitcoind.service": "27.1.0",
  "electrs.service": "0.10.6",
  "lnd.service": "0.18.0",
-  "rtl.service": "0.15.8",
+  "rtl.service": "0.15.10",
  "btcpayserver.service": "2.4.2",
  "albyhub.service": "1.8.0",
  "mempool.service": "3.2.1",
@@ -151,6 +151,7 @@
    dig firefox wp-cli axel
    lk-jwt-service livekit-libwebrtc livekit
    matrix-synapse age onlyoffice-desktopeditors
+    tor-browser
  ];

  # ── Shell ──────────────────────────────────────────────────
@@ -5,11 +5,11 @@
        "nixpkgs": "nixpkgs"
      },
      "locked": {
-        "lastModified": 1785778881,
-        "narHash": "sha256-yHJJTm3o7ZhiVp68G+hh+dz1GKV+ACOT6mQj+eYvJLQ=",
+        "lastModified": 1787246616,
+        "narHash": "sha256-TTbXIBwoaPbzk2o+rGEayqOvrXqKo9JdYJAMDwqjCQ4=",
        "owner": "emmanuelrosa",
        "repo": "btc-clients-nix",
-        "rev": "fc1aca94d839f82e7501fe0d01279d5c1fba6e63",
+        "rev": "8b10c40cb13bac5d100ae1d1fb42eccc0d9c3223",
        "type": "github"
      },
      "original": {
@@ -26,11 +26,11 @@
        ]
      },
      "locked": {
-        "lastModified": 1782949081,
-        "narHash": "sha256-vp6Y/Grm98ESt6ceOkWiHWyZRDV3J1RID4w+6NWK9yA=",
+        "lastModified": 1785627969,
+        "narHash": "sha256-4dtXQk/NMePegK/nWp5NSeuZKLATItOq61lpEvmXqGw=",
        "owner": "hercules-ci",
        "repo": "flake-parts",
-        "rev": "17c9d6cdfc60c64f4ee8d306f9bc0b4ccb51481e",
+        "rev": "427bf4bd9435fdf21321c8cc628c24efc14c0f7a",
        "type": "github"
      },
      "original": {
@@ -56,11 +56,11 @@
    },
    "nixpkgs-stable": {
      "locked": {
-        "lastModified": 1786201459,
-        "narHash": "sha256-CiOTEjmwAmG2AWnaIno9YaCJJmpca2FXPhMAsnrolCg=",
+        "lastModified": 1787101114,
+        "narHash": "sha256-gwrPcFf/rDjHPaVflbDZ040ZDmBTRj/7+s8ZmE2SaIM=",
        "owner": "nixos",
        "repo": "nixpkgs",
-        "rev": "8b8c811c7c2541c30382c5de7ed26be055569c60",
+        "rev": "b18a4b905f8d028dc4476412e6d6891728695379",
        "type": "github"
      },
      "original": {
@@ -72,11 +72,11 @@
    },
    "nixpkgs_2": {
      "locked": {
-        "lastModified": 1786106723,
-        "narHash": "sha256-zDSUbpoeo/9ZmD2+wXnzxoo1+uhL8vxc0b8yuYMKYq0=",
+        "lastModified": 1787135253,
+        "narHash": "sha256-RD2kNWCG+Bjo6h+JVjWVNntZs2GtRoeY2xHjts/FNkA=",
        "owner": "NixOS",
        "repo": "nixpkgs",
-        "rev": "f13ff45afd1bb73e640eaa08a7066dbed07e3238",
+        "rev": "ffb3c9b700e759be2ef13237c9d8f953b32a1e46",
        "type": "github"
      },
      "original": {
@@ -88,11 +88,11 @@
    },
    "nixpkgs_3": {
      "locked": {
-        "lastModified": 1784555310,
-        "narHash": "sha256-/FCliTPgiuV1owejZFNx3Ch9irdvkOfOFl+HHZ+DrtM=",
+        "lastModified": 1787111413,
+        "narHash": "sha256-sFosWtq21eHGJRnTc/hvf4M1obRgLEUMNm/IzllkHMA=",
        "owner": "NixOS",
        "repo": "nixpkgs",
-        "rev": "421eebfd0ec7bccd4abe826ce62d7e6e83129493",
+        "rev": "afe3d8ac4395617bdcdac9f188ac8717a062e014",
        "type": "github"
      },
      "original": {
@@ -109,11 +109,11 @@
        "systems": "systems"
      },
      "locked": {
-        "lastModified": 1785763201,
-        "narHash": "sha256-wA373y/B9orM3HatLu9oS+Ke5lmdZyBl/bdjd2gLMq4=",
+        "lastModified": 1787151631,
+        "narHash": "sha256-EblMdrDFBFNNlUPm5zUQdh0j6gDx+OFQPv7LFE4B5AA=",
        "owner": "nix-community",
        "repo": "nixvim",
-        "rev": "c7be49306b23a952c0151cf4bbeacd944ed82f2a",
+        "rev": "d0d62a2b5027da689b4e8d5ee43f1cf83f2e975d",
        "type": "github"
      },
      "original": {
@@ -131,6 +131,7 @@
      }
    },
    "systems": {
+      "flake": false,
      "locked": {
        "lastModified": 1774449309,
        "narHash": "sha256-brhZ8DmuGtzkCYHJg4HEd602amKm89Y9ytsFZ5uWD1w=",
@@ -284,7 +284,9 @@ let
  nbLib = config.nix-bitcoin.lib;
  secretsDir = config.nix-bitcoin.secretsDir;

-  i2pSAM = config.services.i2pd.proto.sam;
+  # nixpkgs 26.11 moved i2pd's protocol configuration from
+  # `services.i2pd.proto` to the RFC42-style `services.i2pd.settings`.
+  i2pSAM = config.services.i2pd.settings.sam;

  configFile = builtins.toFile "bitcoin.conf" ''
    # We're already logging via journald
@@ -374,7 +376,11 @@ in {

    services.i2pd = mkIf (cfg.i2p != false) {
      enable = true;
-      proto.sam.enable = true;
+      settings.sam = {
+        enabled = true;
+        address = "127.0.0.1";
+        port = 7656;
+      };
    };

    systemd.tmpfiles.rules = [
@@ -150,7 +150,7 @@ let
  nbLib = config.nix-bitcoin.lib;
  secretsDir = config.nix-bitcoin.secretsDir;
  runAsUser = config.nix-bitcoin.runAsUserCmd;
-  lndinit = "${(pkgs.callPackage ../../packages/lndinit {})}/bin/lndinit";
+  lndinit = "${pkgs.lndinit}/bin/lndinit";

  bitcoind = config.services.bitcoind;

@@ -264,13 +264,16 @@ in {
          curl = "${pkgs.curl}/bin/curl -fsS --cacert ${cfg.certPath}";
          restUrl = "https://${nbLib.addressWithPort cfg.restAddress cfg.restPort}/v1";
          # Setting macaroon permissions for other users needs root permissions
+          # The admin macaroon is passed to curl via a fd because argv is
+          # world-readable through /proc/<pid>/cmdline
          script = nbLib.rootScript "lnd-create-macaroons" ''
            umask ug=r,o=
            ${lib.concatMapStrings (macaroon: ''
              echo "Create custom macaroon ${macaroon}"
              macaroonPath="$RUNTIME_DIRECTORY/${macaroon}.macaroon"
+              adminMacaroonHex=$(${pkgs.xxd}/bin/xxd -ps -u -c 99999 '${networkDir}/admin.macaroon')
              ${curl} \
-                -H "Grpc-Metadata-macaroon: $(${pkgs.xxd}/bin/xxd -ps -u -c 99999 '${networkDir}/admin.macaroon')" \
+                -H @<(printf 'Grpc-Metadata-macaroon: %s\n' "$adminMacaroonHex") \
                -X POST \
                -d '{"permissions":[${cfg.macaroons.${macaroon}.permissions}]}' \
                ${restUrl}/macaroon |\
@@ -1,63 +1,115 @@
 { config, lib, pkgs, ... }:

+# LND-only lndconnect wrapper. Restored to the fort-nix/nix-bitcoin contract
+# after the LND-only rewrite shipped a Zeus QR that Zeus cannot use:
+#   - unknown flags (--cert / --macaroon instead of --tlscertpath / --adminmacaroonpath)
+#   - onion hostname read from /var/lib/tor/onion/free/lnd/hostname (does not exist)
+#   - REST hidden service named "lnd", colliding with the LND P2P onion
+#   - TLS cert embedded in the URI (localhost CN + QR too dense to scan)
+#
+# Zeus needs: lndconnect://<lnd-rest-onion>:8080?macaroon=<admin>  (no cert over Tor)
+
 with lib;
 let
  cfg = config.services.lnd;
  operatorName = config.nix-bitcoin.operator.name;
  nbLib = config.nix-bitcoin.lib;
+  runAsUser = config.nix-bitcoin.runAsUserCmd;

-  mkLndconnect = { name, isClightning ? false, enableOnion, onionService, port, certPath, authSecretPath }:
-    let
-      lnd = config.services.lnd;
-      getOnionAddress = "cat ${config.nix-bitcoin.secretsDir}/onion-address-${onionService} 2>/dev/null || echo ${onionService}.onion";
-    in pkgs.writeScriptBin name ''
-      #!${pkgs.bash}/bin/bash
-      set -e
-      certPath="${certPath}"
-      authSecretPath="${authSecretPath}"
-      if [ "${toString enableOnion}" = "1" ]; then
-        host=$(cat /var/lib/tor/onion/${onionService}/hostname 2>/dev/null || echo "${onionService}.onion")
-        port="${toString port}"
+  mkLndconnect = {
+    name,
+    shebang ? "#!${pkgs.stdenv.shell} -e",
+    port,
+    authSecretPath,
+    enableOnion,
+    onionService ? null,
+    certPath ? null
+  }:
+  # lndconnect requires a --configfile argument, although it's unused
+  # https://github.com/LN-Zap/lndconnect/issues/25
+  lib.hiPrio (pkgs.writeScriptBin name ''
+    ${shebang}
+    url=$(
+      ${getExe pkgs.lndconnect} --url \
+        ${optionalString enableOnion "--host=$(cat ${config.nix-bitcoin.onionAddresses.dataDir}/${onionService})"} \
+        --port=${toString port} \
+        ${if enableOnion || certPath == null then "--nocert" else "--tlscertpath='${certPath}'"} \
+        --adminmacaroonpath='${authSecretPath}' \
+        --configfile=/dev/null "$@"
+    )
+
+    # If --url is in args
+    if [[ " $* " =~ " --url " ]]; then
+      echo "$url"
    else
-        host="${nbLib.address lnd.restAddress}"
-        port="${toString lnd.restPort}"
+      # UTF-8 QR is smaller than lndconnect's native output
+      echo -n "$url" | ${getExe pkgs.qrencode} -t UTF8 -o -
    fi
-      # lndconnect is provided by pkgs.lndconnect
-      ${getExe pkgs.lndconnect} --host="$host" --port="$port" --cert="$certPath" --macaroon="$authSecretPath" "$@"
-    '';
+  '');
 in {
  options.services.lnd.lndconnect = {
    enable = mkOption {
      type = types.bool;
      default = false;
-      description = "Enable lndconnect for LND";
+      description = ''
+        Add a `lndconnect` binary to the system environment which prints
+        connection info for lnd clients (Zeus).
+        See: https://github.com/LN-Zap/lndconnect
+
+        Usage:
+        ```bash
+          # Print QR code
+          lndconnect
+
+          # Print URL
+          lndconnect --url
+        ```
+      '';
    };
    onion = mkOption {
      type = types.bool;
      default = false;
-      description = "Expose lndconnect via Tor onion service";
+      description = ''
+        Create an onion service for the lnd REST server,
+        which is used by lndconnect / Zeus.
+      '';
    };
  };

-  config = mkIf cfg.enable (mkMerge [
-    (mkIf cfg.lndconnect.enable {
-      environment.systemPackages = [
-        (mkLndconnect {
+  config = mkIf (cfg.enable && cfg.lndconnect.enable) (mkMerge [
+    {
+      environment.systemPackages = [(
+        mkLndconnect {
          name = "lndconnect";
+          # Run as lnd user because the macaroon and cert are not group-readable
+          shebang = "#!/usr/bin/env -S ${runAsUser} ${cfg.user} ${pkgs.bash}/bin/bash";
          enableOnion = cfg.lndconnect.onion;
-          onionService = "${operatorName}/lnd";
+          onionService = "${cfg.user}/lnd-rest";
          port = cfg.restPort;
          certPath = cfg.certPath;
          authSecretPath = "${cfg.networkDir}/admin.macaroon";
-        })
-      ];
-    })
-    (mkIf (cfg.lndconnect.enable && cfg.lndconnect.onion) {
-      services.tor.relay.onionServices.lnd = nbLib.mkOnionService {
+        }
+      )];
+
+      # LAN / clearnet Zeus needs REST on all interfaces. Tor-only stays on
+      # the existing restAddress (loopback) and is reached via lnd-rest.
+      services.lnd.restAddress = mkIf (!cfg.lndconnect.onion) "0.0.0.0";
+    }
+
+    (mkIf cfg.lndconnect.onion {
+      services.tor = {
+        enable = true;
+        # Dedicated name — must not reuse onionServices.lnd (that's P2P :9735).
+        relay.onionServices.lnd-rest = nbLib.mkOnionService {
+          target.addr = nbLib.address cfg.restAddress;
+          target.port = cfg.restPort;
          port = cfg.restPort;
-        target = { addr = nbLib.address cfg.restAddress; port = cfg.restPort; };
        };
-      nix-bitcoin.onionAddresses.access.${operatorName} = [ "lnd" ];
+      };
+      nix-bitcoin.onionAddresses.access = {
+        ${cfg.user} = [ "lnd-rest" ];
+        ${operatorName} = [ "lnd-rest" ];
+      };
    })
  ]);
 }
@@ -33,7 +33,7 @@
      # /var/lib/njalla/ddns_urls.json.
      NoNewPrivileges    = true;
      ProtectSystem      = "strict";
-      ReadWritePaths     = [ "/var/lib/njalla" ];
+      ReadWritePaths     = [ "/var/lib/njalla" "/var/lib/secrets" ];
      ReadOnlyPaths      = [ "/etc/sovran" ];
      ProtectHome        = true;
      PrivateTmp         = true;
@@ -88,12 +88,15 @@ try:
 except Exception:
    sys.exit(0)  # no URLs configured — nothing to do

-# Resolve current public IP once
+# Resolve current public IP via the shared detector — one script, one cache
+# (STUN -> DNS -> opt-in HTTPS echo; see /var/lib/sovran/public-ip.py).
+# The detector refreshes /var/lib/secrets/external-ip, which the Hub and
+# LiveKit read as well, so the whole system shares a single detected value.
 public_ip = ""
 try:
    r = subprocess.run(
-        ["dig", "@resolver4.opendns.com", "myip.opendns.com", "+short", "-4"],
-        capture_output=True, text=True, timeout=10,
+        [sys.executable, "/var/lib/sovran/public-ip.py", "check"],
+        capture_output=True, text=True, timeout=20,
    )
    raw = r.stdout.strip().splitlines()[0] if r.stdout.strip() else ""
    ipaddress.ip_address(raw)  # validates — raises if not a real IP
@@ -101,6 +104,16 @@ try:
 except Exception:
    pass

+if not public_ip:
+    # Last resort: the shared cache file, if the detector is unavailable.
+    try:
+        with open("/var/lib/secrets/external-ip") as f:
+            raw = f.read().strip()
+        ipaddress.ip_address(raw)
+        public_ip = raw
+    except Exception:
+        pass
+
 if not public_ip:
    sys.exit(0)  # no IP resolved — skip to avoid sending bare ''${IP}

@@ -0,0 +1,323 @@
+# ── Unified public-IP detection (privacy-first) ─────────────────────────────
+#
+# One script, one cache file, every consumer on the system reads the same
+# value. Previously the public IP was detected independently in three places,
+# each phoning home to a different third party:
+#   * the Hub (server.py _get_external_ip) → api.ipify.org / ifconfig.me /
+#     icanhazip.com over HTTPS on every /api/network call and every
+#     background-loop tick
+#   * DDNS (ddns-update.py) → myip.opendns.com via OpenDNS
+#   * LiveKit → STUN (its own embedded detection)
+#
+# This module replaces all of that with a single script
+# (/var/lib/sovran/public-ip.py) that detects the IP once per TTL using the
+# least-exposing mechanism available, and caches it in
+# /var/lib/secrets/external-ip. Consumers (Hub, DDNS, LiveKit) read the cache
+# and only invoke the script when it is missing or stale.
+#
+# Detection chain (first success wins, stops immediately):
+#   1. pin         — sovran_systemsOS.elementCalling.externalIP (baked in)
+#   2. cache       — /var/lib/secrets/external-ip if newer than cacheTTL
+#   3. STUN        — UDP binding request (one packet, no application data,
+#                    no HTTP metadata; the same protocol every WebRTC client
+#                    uses). Server configurable via publicIP.stunServer.
+#   4. DNS         — "myip.opendns.com" A query via publicIP.dnsResolver
+#                    (single DNS query, no HTTP headers)
+#   5. HTTPS echo  — ONLY endpoints listed in publicIP.httpsEcho (empty by
+#                    default → never contacted)
+#
+# Privacy property: while the cache is fresh, zero third parties are
+# contacted. When detection runs, at most ONE party learns the IP per
+# refresh interval (default 5 minutes), and the STUN/DNS mechanisms expose
+# nothing beyond the bare address.
+{
+  config,
+  pkgs,
+  lib,
+  ...
+}:
+
+let
+  stunServer  = config.sovran_systemsOS.publicIP.stunServer;
+  stunPort    = config.sovran_systemsOS.publicIP.stunPort;
+  dnsResolver = config.sovran_systemsOS.publicIP.dnsResolver;
+  httpsEcho   = config.sovran_systemsOS.publicIP.httpsEcho;
+  cacheTTL    = config.sovran_systemsOS.publicIP.cacheTTL;
+
+  # Optional pin shared with element-calling (baked in at build time).
+  pin = if config.sovran_systemsOS.elementCalling.externalIP != null then config.sovran_systemsOS.elementCalling.externalIP else "";
+
+  echoList = lib.concatStringsSep "," (map (u: "'${u}'") httpsEcho);
+in
+{
+  options.sovran_systemsOS.publicIP = {
+    stunServer = lib.mkOption {
+      type = lib.types.str;
+      default = "stun.l.google.com";
+      description = ''
+        STUN server used to discover the public IP over UDP. STUN is the most
+        privacy-preserving detection mechanism: a single stateless packet,
+        no HTTP metadata. Only used when the cache is stale.
+      '';
+    };
+    stunPort = lib.mkOption {
+      type = lib.types.port;
+      default = 19302;
+    };
+    dnsResolver = lib.mkOption {
+      type = lib.types.str;
+      default = "resolver4.opendns.com";
+      description = ''
+        DNS resolver used as fallback (myip.opendns.com trick) when STUN is
+        unavailable (e.g. ISP blocks UDP egress). A single DNS query, no
+        HTTP headers.
+      '';
+    };
+    httpsEcho = lib.mkOption {
+      type = lib.types.listOf lib.types.str;
+      default = [ ];
+      example = [ "https://api.ipify.org" ];
+      description = ''
+        OPT-IN HTTPS endpoints that return the caller's public IP as a bare
+        IPv4 literal. Each listed endpoint observes this server's public IP
+        and HTTP metadata every time detection runs. Empty by default — no
+        HTTPS echo service is ever contacted unless you add one here. This is
+        the last-resort fallback after STUN and DNS.
+      '';
+    };
+    cacheTTL = lib.mkOption {
+      type = lib.types.int;
+      default = 300;
+      description = "Seconds the detected public IP is cached before re-detection.";
+    };
+  };
+
+  # ── Install the unified detector ──────────────────────────────────────────
+  # This module declares `options` above, so ALL configuration must go under
+  # the `config` attribute: NixOS forbids mixing bare top-level settings
+  # (like `system.*`) with the `options`/`config` keyword attributes in the
+  # same module. (Fixes: "Module ... has an unsupported attribute `system'".)
+  config.system.activationScripts.sovranPublicIpInstall = lib.stringAfter [ "users" ] ''
+    install -d -m 0755 /var/lib/sovran
+    cat > /var/lib/sovran/public-ip.py <<'PYEOF'
+#!/usr/bin/env python3
+"""sovran-public-ip — one detector, one cache, every consumer reads the same IP.
+
+Privacy-first detection chain (first success wins):
+  1. pin     — baked in from sovran_systemsOS.elementCalling.externalIP
+  2. cache   — /var/lib/secrets/external-ip if newer than CACHE_TTL seconds
+  3. STUN    — UDP binding request (one packet, no application data)
+  4. DNS     — myip.opendns.com A query via the configured resolver
+  5. HTTPS   — ONLY endpoints baked in from publicIP.httpsEcho (opt-in)
+
+Usage:
+  public-ip.py check     print current public IP (cache first; refresh if stale)
+  public-ip.py refresh   force re-detection, update the cache file, print IP
+
+Exit status: 0 with the IP on stdout on success; 1 if no IP is available
+(cached value, if any, is still printed to stdout with a warning on stderr).
+"""
+import ipaddress
+import os
+import random
+import socket
+import struct
+import sys
+import time
+import urllib.request
+
+CACHE_FILE = "/var/lib/secrets/external-ip"
+PIN        = "${pin}"
+STUN_SERVER = "${stunServer}"
+STUN_PORT   = ${toString stunPort}
+DNS_RESOLVER = "${dnsResolver}"
+DNS_HOST     = "myip.opendns.com"
+ECHO_URLS    = [ ${echoList} ]
+CACHE_TTL    = ${toString cacheTTL}
+TIMEOUT      = 3.0
+
+# ---------------------------------------------------------------------------
+# Detection primitives
+# ---------------------------------------------------------------------------
+
+def is_usable_ip(text: str) -> bool:
+    """True if text is a globally routable IPv4 that LiveKit may advertise."""
+    try:
+        ip = ipaddress.ip_address(text)
+    except ValueError:
+        return False
+    if ip.version != 4:
+        return False
+    if (ip.is_private or ip.is_loopback or ip.is_link_local or ip.is_multicast
+            or ip.is_reserved or ip.is_unspecified or not ip.is_global):
+        return False
+    # RFC 6598 shared (CGNAT) space — not reachable from the internet.
+    if ip in ipaddress.ip_network("100.64.0.0/10"):
+        return False
+    return True
+
+
+def stun_public_ip() -> str | None:
+    """RFC 5389 Binding request over UDP; returns the mapped (public) IPv4."""
+    sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
+    sock.settimeout(TIMEOUT)
+    try:
+        txid = random.randbytes(12)
+        req = struct.pack("!HHI", 0x0001, 0, 0) + txid  # Binding request
+        sock.sendto(req, (STUN_SERVER, STUN_PORT))
+        data, _ = sock.recvfrom(2048)
+    except OSError:
+        return None
+    finally:
+        sock.close()
+
+    if len(data) < 20:
+        return None
+    mtype, _mlen = struct.unpack("!HH", data[:4])
+    if mtype != 0x0101:  # Binding success response
+        return None
+
+    cookie = data[4:8]
+    i = 20
+    while i + 4 <= len(data):
+        atype, alen = struct.unpack("!HH", data[i : i + 4])
+        aval = data[i + 4 : i + 4 + alen]
+        if atype in (0x0001, 0x0020) and len(aval) >= 8:  # MAPPED / XOR-MAPPED
+            family = aval[1]
+            if family == 0x01:  # IPv4
+                raw = aval[4:8]
+                if atype == 0x0020:  # XOR with magic cookie + txid prefix
+                    raw = bytes(b ^ c for b, c in zip(raw, cookie + txid[:4]))
+                return socket.inet_ntop(socket.AF_INET, raw)
+        i += 4 + ((alen + 3) // 4) * 4
+    return None
+
+
+def dns_public_ip() -> str | None:
+    """Minimal DNS A query for myip.opendns.com against the given resolver."""
+    qid = random.randint(0, 0xFFFF)
+    qname = b"".join(bytes([len(p)]) + p.encode() for p in DNS_HOST.split(".")) + b"\x00"
+    query = struct.pack("!HHHHHH", qid, 0x0100, 1, 0, 0, 0) + qname + struct.pack("!HH", 1, 1)
+    sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
+    sock.settimeout(TIMEOUT)
+    try:
+        sock.sendto(query, (DNS_RESOLVER, 53))
+        data, _ = sock.recvfrom(4096)
+    except OSError:
+        return None
+    finally:
+        sock.close()
+
+    try:
+        if len(data) < 12:
+            return None
+        rid, _flags, _qd, an, _ns, _ar = struct.unpack("!HHHHHH", data[:12])
+        if rid != qid or an == 0:
+            return None
+        i = 12
+        for _ in range(_qd):  # skip question
+            while data[i] != 0:
+                i += 1 + data[i]
+            i += 5
+        for _ in range(an):
+            if data[i] & 0xC0 == 0xC0:
+                i += 2
+            else:
+                while data[i] != 0:
+                    i += 1 + data[i]
+                i += 1
+            rtype, _rclass, _ttl, rdlen = struct.unpack("!HHIH", data[i : i + 10])
+            i += 10
+            if rtype == 1 and rdlen == 4:
+                return socket.inet_ntop(socket.AF_INET, data[i : i + 4])
+            i += rdlen
+    except (IndexError, struct.error):
+        return None
+    return None
+
+
+def echo_public_ip() -> str | None:
+    """Opt-in HTTPS echo endpoints (baked in at build time; empty by default)."""
+    for url in ECHO_URLS:
+        try:
+            req = urllib.request.Request(url, headers={"User-Agent": "sovran-public-ip"})
+            with urllib.request.urlopen(req, timeout=TIMEOUT) as resp:
+                text = resp.read().decode().strip()
+            if is_usable_ip(text):
+                return text
+        except Exception:
+            continue
+    return None
+
+
+# ---------------------------------------------------------------------------
+# Cache handling
+# ---------------------------------------------------------------------------
+
+def read_cache() -> str:
+    try:
+        with open(CACHE_FILE) as f:
+            return f.read().strip()
+    except OSError:
+        return ""
+
+
+def write_cache(ip: str) -> None:
+    try:
+        os.makedirs(os.path.dirname(CACHE_FILE), exist_ok=True)
+        tmp = f"{CACHE_FILE}.tmp"
+        with open(tmp, "w") as f:
+            f.write(ip + "\n")
+        os.replace(tmp, CACHE_FILE)
+    except OSError:
+        pass
+
+
+def cache_fresh() -> bool:
+    try:
+        return time.time() - os.path.getmtime(CACHE_FILE) < CACHE_TTL
+    except OSError:
+        return False
+
+
+def detect() -> str:
+    """Run the chain; returns usable IP or an empty string."""
+    if PIN and is_usable_ip(PIN):
+        return PIN
+    for fn in (stun_public_ip, dns_public_ip, echo_public_ip):
+        try:
+            cand = fn()
+        except Exception:
+            continue
+        if cand and is_usable_ip(cand):
+            return cand
+    return ""
+
+
+def main() -> int:
+    force = len(sys.argv) > 1 and sys.argv[1] == "refresh"
+    ip = ""
+    if not force and cache_fresh():
+        ip = read_cache()
+    if not ip:
+        ip = detect()
+        if ip:
+            write_cache(ip)
+        else:
+            stale = read_cache()
+            if stale:
+                print(stale)
+                print("WARNING: detection failed; using last known public IP", file=sys.stderr)
+                return 0
+            print("ERROR: could not determine a public IP (STUN/DNS unreachable)", file=sys.stderr)
+            return 1
+    print(ip)
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())
+PYEOF
+    chmod 0555 /var/lib/sovran/public-ip.py
+  '';
+}
@@ -79,6 +79,41 @@
      };
    };

+    # ── Element Calling (video/audio) tuning ──────────────────
+    elementCalling = {
+      fullAccessHomeservers = lib.mkOption {
+        type = lib.types.listOf lib.types.str;
+        default = [ ];
+        example = [ "matrix.peer.example.com" ];
+        description = ''
+          Additional Matrix server_names (beyond this server itself) that may
+          trigger LiveKit room creation on this server's SFU via lk-jwt-service.
+
+          Not needed for the common federated setup: each participant's client
+          always obtains its token from its own homeserver's JWT service and
+          publishes to its own SFU, and the participant who starts a call
+          creates the room on their own SFU — the remote user merely joins
+          (joining does not require full access).
+
+          Only set this for asymmetric cases: e.g. a peer homeserver that has
+          no focus of its own, or calls whose first participant lands on this
+          server's SFU but belongs to the peer.
+        '';
+      };
+      externalIP = lib.mkOption {
+        type = lib.types.nullOr lib.types.str;
+        default = null;
+        example = "203.0.113.10";
+        description = ''
+          Optional pin: force LiveKit to advertise this public IPv4 in its
+          host/TURN ICE candidates. Not required in normal operation — the
+          module auto-detects the public IP at runtime (HTTPS egress
+          detection, falling back to STUN). Set it only to override a
+          mis-detected address (e.g. multi-WAN/VPN setups).
+        '';
+      };
+    };
+
    # ── Domain setup registry ─────────────────────────────────
    domainRequirements = lib.mkOption {
      type = lib.types.listOf (lib.types.submodule {
@@ -136,9 +136,9 @@ let
    "electrs.service" = if pkgs ? electrs then pkgs.electrs.version else "0.10.6";
    "lnd.service" = if pkgs ? lnd then pkgs.lnd.version else "0.18.0";
    # Keep the fallbacks aligned with the vendored packages used by the
-    # service modules (RTL 0.15.8 and Mempool 3.2.1).  The nixpkgs attrs are
+    # service modules (RTL 0.15.10 and Mempool 3.2.1).  The nixpkgs attrs are
    # optional because these packages are built locally in this repository.
-    "rtl.service" = if pkgs ? clightning-rtl then pkgs.clightning-rtl.version else (if pkgs ? rtl then pkgs.rtl.version else "0.15.8");
+    "rtl.service" = if pkgs ? clightning-rtl then pkgs.clightning-rtl.version else (if pkgs ? rtl then pkgs.rtl.version else "0.15.10");
    # BTCPay Server is intentionally sourced from pkgs.stable by the service
    # module.  Read the configured package here rather than pkgs.btcpayserver
    # (unstable), otherwise the Hub can advertise a version that is not running.
@@ -160,8 +160,10 @@ let

    LOG="/var/log/sovran-hub-update.log"
    STATUS="/var/log/sovran-hub-update.status"
+    GENERATION="/var/log/sovran-hub-update.generation"

    echo "RUNNING" > "$STATUS"
+    rm -f "$GENERATION"
    : > "$LOG"
    exec > >(tee -a "$LOG") 2>&1

@@ -196,6 +198,12 @@ let
      if [ "$BOOT_RC" -ne 0 ]; then
        echo "[ERROR] nixos-rebuild boot failed"
        RC=1
+      elif ! readlink -f /nix/var/nix/profiles/system > "$GENERATION"; then
+        # The marker is informational only.  The Hub derives pending-reboot
+        # state from the NixOS system profile itself, so failing to record
+        # the marker must not fail an otherwise successful update.
+        echo "[WARNING] update succeeded but its staged generation could not be recorded"
+        rm -f "$GENERATION"
      fi
      echo ""
    fi
@@ -33,9 +33,15 @@ lib.mkIf config.sovran_systemsOS.features.element-calling {
    '';
  };

-  ####### ENSURE SERVICES START AFTER KEY EXISTS #######
-  systemd.services.livekit.after = [ "livekit-key-setup.service" "livekit-turn-setup.service" ];
-  systemd.services.livekit.wants = [ "livekit-key-setup.service" "livekit-turn-setup.service" ];
+  ####### ENSURE SERVICES START AFTER KEY & NETWORK EXIST #######
+  # Ordering against network-online.target matters: livekit-turn-setup detects
+  # the primary interface from the IPv4 default route. If it runs before the
+  # network is up (no default route yet) it exits 1 and, being a hard
+  # dependency of livekit.service, takes livekit down with it — the Hub then
+  # shows a "failed" red dot until livekit is restarted manually. See the
+  # livekit-turn-setup block for the matching network-online ordering.
+  systemd.services.livekit.after = [ "network-online.target" "livekit-key-setup.service" "livekit-turn-setup.service" ];
+  systemd.services.livekit.wants = [ "network-online.target" "livekit-key-setup.service" "livekit-turn-setup.service" ];
  systemd.services.lk-jwt-service.after = [ "livekit-key-setup.service" ];
  systemd.services.lk-jwt-service.wants = [ "livekit-key-setup.service" ];

@@ -114,7 +120,12 @@ EOF
  #     substituted) that the overridden ExecStart loads.
  systemd.services.livekit-turn-setup = {
    description = "Stage TURN cert and generate LiveKit runtime config from domain files";
-    after = [ "caddy.service" "livekit-key-setup.service" ];
+    # Wait for a default IPv4 route before detecting the interface, and for
+    # Caddy to have started (cert generation is async, so also see the retry
+    # loop below). Otherwise on a cold boot this unit can fail / produce empty
+    # certs, which breaks livekit.service (requiredBy) and shows a red dot.
+    after = [ "network-online.target" "caddy.service" "livekit-key-setup.service" ];
+    wants = [ "network-online.target" ];
    before = [ "livekit.service" ];
    requiredBy = [ "livekit.service" ];
    wantedBy = [ "multi-user.target" ];
@@ -125,19 +136,39 @@ EOF
    unitConfig = {
      ConditionPathExists = "/var/lib/domains/element-calling";
    };
-    path = [ pkgs.coreutils pkgs.findutils pkgs.iproute2 pkgs.gawk ];
+    path = [ pkgs.coreutils pkgs.findutils pkgs.iproute2 pkgs.gawk pkgs.python3 ];
    script = ''
      MATRIX=$(cat /var/lib/domains/matrix)
+      ELEMENT_CALLING=$(cat /var/lib/domains/element-calling)

      mkdir -p /run/livekit

      # Copy Caddy's already-issued matrix cert/key into LiveKit's state dir.
      # The ACME CA hostname directory can vary, so glob for the domain dir.
+      # Caddy issues ACME certs asynchronously, so on a fresh boot the cert may
+      # not exist yet. Retry (bounded) so we never write an empty turn.crt/key;
+      # otherwise embedded TURN silently breaks until the next livekit restart.
+      CRT=""
+      KEY=""
+      for _ in $(seq 1 30); do
        CRT=$(find /var/lib/caddy -path "*/$MATRIX/$MATRIX.crt" | head -n1)
        KEY=$(find /var/lib/caddy -path "*/$MATRIX/$MATRIX.key" | head -n1)
+        if [ -n "$CRT" ] && [ -n "$KEY" ] \
+           && [ -s "$CRT" ] && [ -s "$KEY" ]; then
+          break
+        fi
+        CRT=""
+        KEY=""
+        echo "Waiting for Caddy to issue the $MATRIX ACME certificate..."
+        sleep 2
+      done
+      if [ -z "$CRT" ] || [ -z "$KEY" ]; then
+        echo "ERROR: Caddy ACME certificate for $MATRIX not available after retries; TURN will not be enabled for LiveKit." >&2
+      else
        cp "$CRT" /var/lib/livekit/turn.crt
        cp "$KEY" /var/lib/livekit/turn.key
        chmod 640 /var/lib/livekit/turn.crt /var/lib/livekit/turn.key
+      fi

      # Detect the primary network interface from the IPv4 default route.
      # Restricting LiveKit to this single interface prevents it from
@@ -158,6 +189,56 @@ EOF
      # rtc.interfaces.includes are only known at runtime, so they are
      # substituted here. The cert/key paths point at the LoadCredential-staged
      # copies under /run/credentials.
+      #
+      # Determine the public IPv4 to advertise in LiveKit ICE candidates.
+      # Remote peers must be able to reach this address, so it must be the
+      # server's public IP — or the router's WAN IP when the server is behind
+      # NAT with port-forwarding. It does not need to be assigned to this box,
+      # and it may be dynamic.
+      #
+      # Reuse the shared detector (/var/lib/sovran/public-ip.py — see
+      # modules/core/public-ip.nix) instead of running our own: one script,
+      # one cache, privacy-first (STUN -> DNS -> opt-in HTTPS echo). Priority:
+      #   1. sovran_systemsOS.elementCalling.externalIP (explicit pin, if set)
+      #   2. /var/lib/secrets/external-ip (the shared cache)
+      #   3. run the detector now (it refreshes the cache)
+      #   4. STUN auto-detection (use_external_ip) as the fallback, with a
+      #      warning — this is where broken installs used to silently end up
+      #      advertising a private IP, causing "call connects but no video".
+      EXTERNAL_IP='${if config.sovran_systemsOS.elementCalling.externalIP != null then config.sovran_systemsOS.elementCalling.externalIP else ""}'
+
+      PUBLIC_IP="$EXTERNAL_IP"
+      if [ -z "$PUBLIC_IP" ] && [ -f /var/lib/secrets/external-ip ]; then
+        PUBLIC_IP=$(tr -d '[:space:]' < /var/lib/secrets/external-ip 2>/dev/null)
+      fi
+      if [ -z "$PUBLIC_IP" ] && [ -x /var/lib/sovran/public-ip.py ]; then
+        PUBLIC_IP=$(python3 /var/lib/sovran/public-ip.py check 2>/dev/null | head -n1)
+      fi
+
+      # Reject non-routable addresses (loopback, private, link-local, CGNAT).
+      # A detected/pinned address like this must never be advertised.
+      if [ -n "$PUBLIC_IP" ] && printf '%s' "$PUBLIC_IP" | grep -qE \
+        '^(0\.|127\.|10\.|100\.64\.|169\.254\.|172\.(1[6-9]|2[0-9]|3[01])\.|192\.168\.)'; then
+        echo "WARNING: external IP '$PUBLIC_IP' is not routable; falling back to STUN auto-detection." >&2
+        PUBLIC_IP=""
+      fi
+
+      if [ -n "$PUBLIC_IP" ]; then
+        cat > /run/livekit/livekit.yaml <<EOF
+port: 7880
+rtc:
+  use_external_ip: false
+  node_ip: $PUBLIC_IP
+  tcp_port: 7881
+  udp_port: 7882
+  port_range_start: 30000
+  port_range_end: 40000
+  interfaces:
+    includes:
+      - $IFACE
+EOF
+        echo "LiveKit will advertise public IP: $PUBLIC_IP"
+      else
        cat > /run/livekit/livekit.yaml <<EOF
 port: 7880
 rtc:
@@ -170,6 +251,20 @@ rtc:
  interfaces:
    includes:
      - $IFACE
+EOF
+        echo "WARNING: could not determine a public IP for LiveKit; using STUN auto-detection. If calls connect without media, check STUN egress or set sovran_systemsOS.elementCalling.externalIP." >&2
+      fi
+
+      # Webhooks → lk-jwt-service. The JWT service validates the HMAC
+      # signature against the same key file it issues tokens with, and uses
+      # the events (participant_left / room_finished) to detect abruptly
+      # disconnected participants instead of waiting for the delayed-event
+      # timeout. The URL hits local Caddy via the /etc/hosts loopback
+      # override and is routed to the JWT service by the element-calling
+      # vhost (/livekit/jwt/sfu_webhook → 8073).
+      LK_KEY=$(cut -d: -f1 < ${livekitKeyFile} | tr -d '[:space:]')
+
+      cat >> /run/livekit/livekit.yaml <<EOF
 room:
  auto_create: false
 turn:
@@ -179,6 +274,10 @@ turn:
  udp_port: 3478
  cert_file: /run/credentials/livekit.service/turn-cert
  key_file: /run/credentials/livekit.service/turn-key
+webhook:
+  api_key: $LK_KEY
+  urls:
+    - https://$ELEMENT_CALLING/livekit/jwt/sfu_webhook
 EOF

      chmod 644 /run/livekit/livekit.yaml
@@ -186,24 +285,17 @@ EOF
  };

  ####### LIVEKIT SERVICE #######
+  # NOTE: the runtime config (rtc ports, TURN, webhook, node_ip) is generated
+  # by livekit-turn-setup and delivered via LoadCredential; the upstream
+  # module's `settings` block is therefore intentionally NOT used (it would
+  # be dead config that silently diverges from what LiveKit actually loads).
+  # The firewall ports are opened explicitly below; openFirewall is left off
+  # so the upstream module does not also open 7880/tcp publicly (Caddy fronts
+  # the SFU on this host).
  services.livekit = {
    enable = true;
-    openFirewall = true;
+    openFirewall = false;
    keyFile = livekitKeyFile;
-    settings = {
-      rtc.use_external_ip = true;
-      rtc.skip_external_ip_validation = true;
-      rtc.tcp_port = 7881;
-      rtc.udp_port = 7882;
-      rtc.port_range_start = 30000;
-      rtc.port_range_end = 40000;
-      room.auto_create = false;
-      turn = {
-        enabled = true;
-        tls_port = 5349;
-        udp_port = 3478;
-      };
-    };
  };

  # Override ExecStart to load the runtime-generated config (which carries the
@@ -247,12 +339,26 @@ EOF
    script = ''
      ELEMENT_CALLING=$(cat /var/lib/domains/element-calling)
      MATRIX=$(cat /var/lib/domains/matrix)
+      FULL_ACCESS_HOMESERVERS="$MATRIX"
+
+      # Federated peers may also be granted LiveKit room-creation (full access)
+      # on this SFU via sovran_systemsOS.elementCalling.fullAccessHomeservers.
+      # Without this, remote users can join existing calls but cannot be the
+      # first to start one on your SFU.
+      EXTRA_HS='${lib.concatStringsSep "," config.sovran_systemsOS.elementCalling.fullAccessHomeservers}'
+      if [ -n "$EXTRA_HS" ]; then
+        FULL_ACCESS_HOMESERVERS="$FULL_ACCESS_HOMESERVERS,$EXTRA_HS"
+      fi

      mkdir -p /run/lk-jwt-service

      cat > /run/lk-jwt-service/env <<EOF
 LIVEKIT_URL=wss://$ELEMENT_CALLING
-LIVEKIT_FULL_ACCESS_HOMESERVERS=$MATRIX
+LIVEKIT_FULL_ACCESS_HOMESERVERS=$FULL_ACCESS_HOMESERVERS
+# Re-check, every 60s, that connected participants are still on the SFU;
+# guards against missed SFU webhooks (e.g. an SFU restart) leaving stale
+# call members in Matrix rooms.
+LIVEKIT_SANITY_CHECK_INTERVAL_SECONDS=60
 EOF

      chmod 640 /run/lk-jwt-service/env
@@ -264,6 +370,9 @@ EOF
    enable = true;
    port = 8073;
    keyFile = livekitKeyFile;
+    # Required by the upstream module's option type, but overridden at runtime
+    # by EnvironmentFile (/run/lk-jwt-service/env, generated above from the
+    # element-calling domain). Kept as a harmless placeholder.
    livekitUrl = "wss://placeholder.local";
  };

@@ -271,6 +380,126 @@ EOF
    "/run/lk-jwt-service/env"
  ];

+  # Restart LiveKit / lk-jwt-service when a rebuild regenerates their runtime
+  # configs (new domains, externalIP, full-access list), mirroring the domain
+  # change flow.
+  # Re-run the config generator and restart LiveKit when a rebuild regenerates
+  # the runtime config, or when the Hub persists a new external IP (dynamic
+  # WAN IPs), so the advertised ICE candidate stays current without a manual
+  # restart. The trigger chain: external-ip change → livekit-turn-setup
+  # re-runs → rewrites livekit.yaml → livekit restarts with the new config.
+  systemd.services.livekit-turn-setup.restartTriggers = [ "/var/lib/secrets/external-ip" ];
+  systemd.services.livekit.restartTriggers = [ "/run/livekit/livekit.yaml" ];
+  systemd.services.lk-jwt-service.restartTriggers = [ "/run/lk-jwt-service/env" ];
+
+  ####### PUBLIC REACHABILITY SELF-CHECK #######
+  # Diagnostic only — never a hard dependency of livekit/caddy. Catches the
+  # classic "call connects but no media" setup errors at boot instead of at
+  # call time:
+  #   * the element-calling domain having no public records (or resolving to
+  #     loopback/link-local/CGNAT for remote peers),
+  #   * the lk-jwt-service being unreachable through Caddy,
+  #   * the MatrixRTC transports endpoint being absent (Element X cannot
+  #     discover calling and shows MISSING_MATRIX_RTC_TRANSPORT).
+  # The check queries only the operator's own DNS provider (the domain's
+  # authoritative nameservers, resolved via the local resolver) plus the
+  # server's own Caddy and public IP — no third-party resolver or service is
+  # contacted. dig queries resolvers directly, so the /etc/hosts loopback
+  # overrides (modules/core/local-domain-loopback.nix) do not influence the
+  # result.
+  systemd.services.element-calling-public-check = {
+    description = "Verify Element Calling domain, JWT service and MatrixRTC transports endpoint are publicly reachable";
+    after = [ "network-online.target" "caddy.service" "livekit.service" "lk-jwt-service.service" ];
+    wants = [ "network-online.target" ];
+    wantedBy = [ "multi-user.target" ];
+    serviceConfig = {
+      Type = "oneshot";
+      RemainAfterExit = true;
+    };
+    unitConfig = {
+      ConditionPathExists = "/var/lib/domains/element-calling";
+    };
+    path = [ pkgs.coreutils pkgs.gawk pkgs.dnsutils pkgs.curl ];
+    script = ''
+      ELEMENT_CALLING=$(cat /var/lib/domains/element-calling)
+      MATRIX=$(cat /var/lib/domains/matrix)
+      FAIL=0
+
+      echo "── Element Calling public reachability self-check ──"
+
+      # 1) Authoritative DNS view (bypassing the /etc/hosts loopback
+      #    overrides). Resolve the domain's own nameservers via the local
+      #    resolver, then query those nameservers directly — the only party
+      #    that sees the query is the DNS provider the operator already uses
+      #    for the domain.
+      NS_LIST=$(dig +short NS "$ELEMENT_CALLING" 2>/dev/null | tr '\n' ' ')
+      if [ -n "$NS_LIST" ]; then
+        IPS=""
+        for NSRV in $NS_LIST; do
+          IPS=$( { dig +short A "$ELEMENT_CALLING" "@$NSRV" 2>/dev/null; dig +short AAAA "$ELEMENT_CALLING" "@$NSRV" 2>/dev/null; } | tr '\n' ' ' )
+          [ -n "$IPS" ] && break
+        done
+        echo "Authoritative nameservers for $ELEMENT_CALLING: $NS_LIST"
+      else
+        echo "WARNING: could not resolve nameservers for $ELEMENT_CALLING via the local resolver; using the local resolver's answer instead." >&2
+        IPS=$( { dig +short A "$ELEMENT_CALLING" 2>/dev/null; dig +short AAAA "$ELEMENT_CALLING" 2>/dev/null; } | tr '\n' ' ' )
+      fi
+
+      if [ -z "$IPS" ]; then
+        echo "ERROR: no A/AAAA records for $ELEMENT_CALLING at its authoritative nameservers. Remote peers cannot reach this LiveKit; calls will connect without media." >&2
+        FAIL=1
+      else
+        echo "Public DNS for $ELEMENT_CALLING: $IPS"
+        for IP in $IPS; do
+          case "$IP" in
+            0.*|127.*|169.254.*|100.64.*|::1|fe80:*|fc*:*|fd*:*)
+              echo "ERROR: $ELEMENT_CALLING resolves to $IP (loopback/link-local/CGNAT). Remote peers cannot reach it." >&2
+              FAIL=1 ;;
+          esac
+        done
+      fi
+
+      # 2) lk-jwt-service healthz through Caddy (validates the proxy chain).
+      if curl -fsS --max-time 10 "https://$ELEMENT_CALLING/livekit/jwt/healthz" >/dev/null 2>&1; then
+        echo "OK: https://$ELEMENT_CALLING/livekit/jwt/healthz responds"
+      else
+        echo "ERROR: https://$ELEMENT_CALLING/livekit/jwt/healthz not reachable through Caddy." >&2
+        FAIL=1
+      fi
+
+      # 3) Same healthz via the first public IP (tests the full NAT path).
+      #    NOTE: if this box is behind the same NAT you are testing through,
+      #    routers without hairpin NAT will fail this step — the warning is
+      #    then expected and harmless; verify from an external device instead.
+      if [ -n "$IPS" ]; then
+        PUBIP=$(echo "$IPS" | awk '{print $1}')
+        if curl -fsS --max-time 15 --resolve "$ELEMENT_CALLING:443:$PUBIP" "https://$ELEMENT_CALLING/livekit/jwt/healthz" >/dev/null 2>&1; then
+          echo "OK: healthz reachable via public IP $PUBIP (NAT path works)"
+        else
+          echo "WARNING: healthz NOT reachable via public IP $PUBIP — check router port-forwarding (443/TCP) and NAT hairpin. Expected if the router lacks hairpin NAT; verify from an external device." >&2
+        fi
+      fi
+
+      # 4) MatrixRTC transports registry (MSC4519) — required by Element X.
+      CODE=$(curl -s -o /dev/null -w "%{http_code}" --max-time 10 "https://$MATRIX/_matrix/client/unstable/org.matrix.msc4143/rtc/transports")
+      case "$CODE" in
+        401|200)
+          echo "OK: MatrixRTC transports endpoint present (HTTP $CODE; auth required is expected)" ;;
+        404)
+          echo "ERROR: /_matrix/client/unstable/org.matrix.msc4143/rtc/transports missing (HTTP 404) — Element X cannot discover calling. Enable msc4143_enabled and matrix_rtc.transports in Synapse." >&2
+          FAIL=1 ;;
+        *)
+          echo "WARNING: transports endpoint returned HTTP $CODE" >&2 ;;
+      esac
+
+      if [ "$FAIL" -eq 1 ]; then
+        echo "── Element Calling self-check FAILED — see errors above ──" >&2
+        exit 1
+      fi
+      echo "── Element Calling self-check passed ──"
+    '';
+  };
+
  ####### SYNAPSE RUNTIME CONFIG (element-calling additions) #######
  systemd.services.element-calling-synapse-config = {
    description = "Generate Synapse runtime config for Element Calling";
@@ -287,6 +516,7 @@ EOF
    path = [ pkgs.coreutils ];
    script = ''
      MATRIX=$(cat /var/lib/domains/matrix)
+      ELEMENT_CALLING=$(cat /var/lib/domains/element-calling)

      mkdir -p /run/matrix-synapse

@@ -296,7 +526,17 @@ public_baseurl: "https://$MATRIX"
 serve_server_wellknown: true
 experimental_features:
  msc3266_enabled: true
+  # MSC4143: enables the MatrixRTC transports registry endpoint
+  # (/_matrix/client/unstable/org.matrix.msc4143/rtc/transports, MSC4519).
+  # Element X requires this endpoint to discover the LiveKit focus; without it
+  # mobile clients fail with MISSING_MATRIX_RTC_TRANSPORT / cannot start calls.
+  msc4143_enabled: true
  msc4222_enabled: true
+# MSC4519: advertise this site's LiveKit focus via the transports registry.
+matrix_rtc:
+  transports:
+    - type: livekit
+      livekit_service_url: "https://$ELEMENT_CALLING/livekit/jwt"
 max_event_delay_duration: "24h"
 rc_message:
  per_second: 0.5
@@ -17,6 +17,7 @@
    ./core/no-sleep.nix
    ./core/cpu-performance.nix
    ./core/local-domain-loopback.nix
+    ./core/public-ip.nix

    # ── Always on (no flag) ───────────────────────────────────
    ./php.nix
@@ -47,29 +47,44 @@ EOF
  systemd.services.zeus-connect-setup = {
    description = "Save Zeus lndconnect URL";
    wantedBy = [ "multi-user.target" ];
-    after = [ "lnd.service" ];
+    after = [ "lnd.service" "onion-addresses.service" ];
+    wants = [ "lnd.service" "onion-addresses.service" ];
    serviceConfig = {
      Type = "oneshot";
      RemainAfterExit = true;
    };
-    path = [ pkgs.coreutils "/run/current-system/sw" ];
+    # sudo is required: the lndconnect wrapper re-execs as the lnd user so it
+    # can read admin.macaroon (not group-readable).
+    path = [ pkgs.coreutils pkgs.gnugrep pkgs.sudo "/run/current-system/sw" ];
    script = ''
      SECRET_FILE="/var/lib/secrets/zeus-connect-url"
      mkdir -p /var/lib/secrets

+      # LND may still be creating the wallet / macaroon, and the dedicated
+      # lnd-rest onion hostname is published by onion-addresses.service.
      URL=""
+      ATTEMPTS=0
+      while [ "$ATTEMPTS" -lt 60 ]; do
        if command -v lndconnect >/dev/null 2>&1; then
-        URL=$(lndconnect --url 2>/dev/null || true)
-      elif command -v lnconnect-clnrest >/dev/null 2>&1; then
-        URL=$(lnconnect-clnrest --url 2>/dev/null || true)
+          URL=$(lndconnect --url 2>/dev/null | tr -d '\r' | tail -n 1 || true)
        fi
+        # Zeus LND REST over Tor: lndconnect://<v3-onion>:8080?macaroon=...
+        if echo "$URL" | grep -q '^lndconnect://' \
+           && echo "$URL" | grep -q '\.onion' \
+           && echo "$URL" | grep -q 'macaroon='; then
+          break
+        fi
+        URL=""
+        ATTEMPTS=$((ATTEMPTS + 1))
+        sleep 2
+      done

      if [ -n "$URL" ]; then
-        echo "$URL" > "$SECRET_FILE"
+        printf '%s\n' "$URL" > "$SECRET_FILE"
        chmod 600 "$SECRET_FILE"
        echo "Zeus connect URL saved."
      else
-        echo "No lndconnect URL available yet."
+        echo "No valid lndconnect URL available yet."
      fi
    '';
  };
@@ -1,19 +0,0 @@
-{ lib, buildGoModule, fetchFromGitHub }:
-
-buildGoModule rec {
-  pname = "lndinit";
-  version = "0.1.3-beta";
-  src = fetchFromGitHub {
-    owner = "lightninglabs";
-    repo = pname;
-    rev = "v${version}";
-    sha256 = "sha256-sO1DpbppCurxr9g9nUl9Vx82FJK1mTcUw3rY1Fm1wEU=";
-  };
-  vendorHash = "sha256-El44BS5Bu0K/klMxkajciU/R6uqiXBMOiLN536QztbE=";
-  subPackages = [ "." ];
-  meta = with lib; {
-    description = "Wallet initializer for lnd (from nix-bitcoin)";
-    homepage = "https://github.com/lightninglabs/lndinit";
-    license = licenses.mit;
-  };
-}
@@ -10,11 +10,11 @@
 }:
 let self = stdenvNoCC.mkDerivation {
  pname = "rtl";
-  version = "0.15.8";
+  version = "0.15.10";

  src = fetchurl {
    url = "https://github.com/Ride-The-Lightning/RTL/archive/refs/tags/v${self.version}.tar.gz";
-    hash = "sha256-8XdGyORxB2dkZRB/Yl7zh+Quqo4L/Y0VmC6Brbr/hqU=";
+    hash = "sha256-r5riYV2FN0OKi0mwj9I1jBeeU1LOv2HVB6CEovPlUuY=";
  };

  passthru = {
@@ -26,7 +26,7 @@ let self = stdenvNoCC.mkDerivation {
      # TODO-EXTERNAL: Remove `npmFlags` when no longer required
      # See: https://github.com/Ride-The-Lightning/RTL/issues/1182
      npmFlags = "--legacy-peer-deps";
-      hash = "sha256-oMqd6nLzS6iQ9w4z2yzpR2unA5qhOq5YdvfoS8IgYLY=";
+      hash = "sha256-NKiWcjqYcHBVIB+vbF3aKXLe2fJRmh/quu8obztP3TA=";
    };
  };

@@ -333,6 +333,33 @@ class TestSshPubkeyValidation(unittest.TestCase):
            _validate_ssh_pubkey("ssh-ed25519")


+# ---------------------------------------------------------------------------
+# LND macaroon command-line safety
+# ---------------------------------------------------------------------------
+
+class TestLndMacaroonCommandLineSafety(unittest.TestCase):
+    """The LND admin macaroon must never be exposed in curl's argv."""
+
+    @classmethod
+    def setUpClass(cls):
+        path = os.path.join(_REPO_ROOT, "modules", "bitcoin", "lnd.nix")
+        with open(path, encoding="utf-8") as f:
+            cls.lnd_module = f.read()
+
+    def test_admin_macaroon_not_interpolated_into_header_argument(self):
+        self.assertNotIn(
+            '-H "Grpc-Metadata-macaroon: $(',
+            self.lnd_module,
+        )
+
+    def test_admin_macaroon_header_is_passed_via_file_descriptor(self):
+        self.assertIn("adminMacaroonHex=$(", self.lnd_module)
+        self.assertIn(
+            """-H @<(printf 'Grpc-Metadata-macaroon: %s\\n' "$adminMacaroonHex")""",
+            self.lnd_module,
+        )
+
+
 # ---------------------------------------------------------------------------
 # Auth-exempt paths
 # ---------------------------------------------------------------------------
@@ -0,0 +1,222 @@
+"""Regression tests for Hub update completion and polling recovery."""
+
+from __future__ import annotations
+
+import os
+import shutil
+import subprocess
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+
+
+_REPO_ROOT = Path(__file__).resolve().parents[1]
+_APP_PARENT = _REPO_ROOT / "app"
+if str(_APP_PARENT) not in sys.path:
+    sys.path.insert(0, str(_APP_PARENT))
+
+from sovran_systemsos_web.update_state import (  # noqa: E402
+    effective_update_status,
+    reboot_is_pending,
+)
+
+
+# Real store paths observed on the incident machine that prompted this rework.
+RUNNING_GENERATION = (
+    "84rsiqi66nc68jbikd26ms50ap831xf8-nixos-system-nixos-26.11.20260817.ec2d622"
+)
+PREVIOUS_GENERATION = (
+    "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-nixos-system-nixos-26.11.20260816.old"
+)
+# What the stale Hub log still claimed was staged: a two-week-old update cycle.
+HUB_LOG_GENERATION = (
+    "yis0saq6p8fhqcaii0h2yzqf0blhdwns-nixos-system-nixos-26.11.20260804.e72e4f2"
+)
+
+
+class TestRebootPendingState(unittest.TestCase):
+    """Reboot-pending state is derived from NixOS, not from Hub marker files.
+
+    The system profile (what boots next) is compared against
+    /run/current-system (what is running).  This stays correct no matter
+    which tool performed the update: Hub "Update System", a terminal
+    ``nixos-rebuild``, or an SSH support session.
+    """
+
+    def setUp(self):
+        self.tmp = tempfile.TemporaryDirectory()
+        self.addCleanup(self.tmp.cleanup)
+        root = Path(self.tmp.name)
+        self.store = root / "store"
+        self.store.mkdir()
+        # /nix/var/nix/profiles/system is a two-hop chain: the diagnostics on
+        # the incident box showed ``system`` -> ``system-148-link`` -> store.
+        self.profile = root / "system"
+        self.profile_entry = root / "system-148-link"
+        # /run/current-system links straight to the running store path.
+        self.current = root / "current-system"
+
+    def _store_dir(self, name: str) -> str:
+        path = self.store / name
+        path.mkdir(exist_ok=True)
+        return str(path)
+
+    def _stage(self, booted: str, running: str) -> None:
+        """Stage ``booted`` as the boot default with ``running`` live."""
+        os.symlink(self._store_dir(booted), self.profile_entry)
+        os.symlink(self.profile_entry, self.profile)
+        os.symlink(self._store_dir(running), self.current)
+
+    def test_hub_update_staged_and_not_yet_rebooted_is_pending(self):
+        self._stage(booted=RUNNING_GENERATION, running=PREVIOUS_GENERATION)
+        self.assertTrue(
+            reboot_is_pending(str(self.profile), str(self.current))
+        )
+
+    def test_staged_generation_booted_is_no_longer_pending(self):
+        self._stage(booted=RUNNING_GENERATION, running=RUNNING_GENERATION)
+        self.assertFalse(
+            reboot_is_pending(str(self.profile), str(self.current))
+        )
+
+    def test_terminal_switch_needs_no_reboot(self):
+        # nixos-rebuild switch moves the profile AND current-system together.
+        self._stage(booted=RUNNING_GENERATION, running=RUNNING_GENERATION)
+        self.assertFalse(
+            reboot_is_pending(str(self.profile), str(self.current))
+        )
+
+    def test_rollback_leaves_nothing_pending(self):
+        # nixos-rebuild switch --rollback points both at the rollback target.
+        self._stage(booted=PREVIOUS_GENERATION, running=PREVIOUS_GENERATION)
+        self.assertFalse(
+            reboot_is_pending(str(self.profile), str(self.current))
+        )
+
+    def test_unverifiable_state_is_never_a_reboot_demand(self):
+        # No profile and no running system readable: the Hub must not nag
+        # about a reboot it cannot substantiate.
+        self.assertFalse(
+            reboot_is_pending(str(self.profile), str(self.current))
+        )
+
+    def test_stale_hub_marker_clears_after_terminal_updates(self):
+        """The exact incident: terminal-updated machine, frozen Hub marker.
+
+        The user's last Hub update (old updater, weeks prior) left
+        REBOOT_REQUIRED behind.  Every update since ran in a terminal and
+        never touched the Hub's files, so the log still records a staged
+        generation from 2026-08-04 while the machine runs a 2026-08-17
+        build.  With the profile and the running system in agreement, the
+        stale claim must reconcile to IDLE regardless of anything the old
+        marker/log files say.
+        """
+        root = Path(self.tmp.name)
+        log = root / "sovran-hub-update.log"
+        log.write_text(
+            "Done. The new configuration is "
+            f"/nix/store/{HUB_LOG_GENERATION}\n"
+            "✓ Update staged successfully\n",
+            encoding="utf-8",
+        )
+        # Deliberately no sovran-hub-update.generation marker: the updater
+        # that produced this state predates the marker feature.
+        self.assertFalse(
+            (root / "sovran-hub-update.generation").exists()
+        )
+
+        self._stage(booted=RUNNING_GENERATION, running=RUNNING_GENERATION)
+        self.assertEqual(
+            effective_update_status(
+                "REBOOT_REQUIRED", str(self.profile), str(self.current)
+            ),
+            "IDLE",
+        )
+
+    def test_genuine_pending_reboot_claim_survives(self):
+        # A staged generation that has NOT been booted yet: the claim is
+        # true and must keep surfacing until the reboot really happens.
+        self._stage(booted=RUNNING_GENERATION, running=PREVIOUS_GENERATION)
+        self.assertEqual(
+            effective_update_status(
+                "REBOOT_REQUIRED", str(self.profile), str(self.current)
+            ),
+            "REBOOT_REQUIRED",
+        )
+
+    def test_other_statuses_pass_through_unchanged(self):
+        self._stage(booted=RUNNING_GENERATION, running=RUNNING_GENERATION)
+        for status in ("RUNNING", "FAILED", "SUCCESS", "IDLE"):
+            self.assertEqual(
+                effective_update_status(
+                    status, str(self.profile), str(self.current)
+                ),
+                status,
+            )
+
+
+class TestUpdatePollingWiring(unittest.TestCase):
+    """Guard the browser failure modes that caused a permanent spinner."""
+
+    @classmethod
+    def setUpClass(cls):
+        js_dir = _REPO_ROOT / "app" / "sovran_systemsos_web" / "static" / "js"
+        cls.update_js = (js_dir / "update.js").read_text(encoding="utf-8")
+        cls.rebuild_js = (js_dir / "rebuild.js").read_text(encoding="utf-8")
+        cls.helpers_js = (js_dir / "helpers.js").read_text(encoding="utf-8")
+        cls.events_js = (js_dir / "events.js").read_text(encoding="utf-8")
+        cls.template = (
+            _REPO_ROOT / "app" / "sovran_systemsos_web" / "templates" / "index.html"
+        ).read_text(encoding="utf-8")
+
+    def test_status_fetches_have_an_abort_timeout(self):
+        self.assertIn("function apiFetchWithTimeout", self.helpers_js)
+        self.assertIn("new AbortController()", self.helpers_js)
+        self.assertIn("controller.abort()", self.helpers_js)
+        self.assertIn("apiFetchWithTimeout(", self.update_js)
+        self.assertIn("STATUS_POLL_FETCH_TIMEOUT", self.update_js)
+
+    def test_async_update_polls_cannot_overlap(self):
+        self.assertIn("_updatePollInFlight", self.update_js)
+        self.assertIn(
+            "if (_updateFinished || _updatePollInFlight) return;", self.update_js
+        )
+        self.assertIn("finally", self.update_js)
+
+    def test_connection_failure_has_explicit_non_running_ui(self):
+        self.assertIn("showUpdateStatusUnavailable", self.update_js)
+        self.assertIn("Update status unavailable", self.update_js)
+        self.assertIn("Retry Status", self.template)
+        self.assertIn("retryUpdateStatus", self.events_js)
+
+    def test_rdp_or_tab_resume_forces_reconciliation(self):
+        self.assertIn("resumeUpdateStatusAfterInterruption", self.events_js)
+        self.assertIn('window.addEventListener("focus"', self.events_js)
+        self.assertIn('document.addEventListener("visibilitychange"', self.events_js)
+
+    def test_verbose_log_rendering_is_bounded(self):
+        self.assertIn("UPDATE_VISIBLE_LOG_MAX_CHARS", self.update_js)
+        self.assertIn("document.createTextNode(text)", self.update_js)
+        self.assertNotIn("$modalLog.textContent += text", self.update_js)
+
+    def test_page_reload_restores_running_or_completed_update(self):
+        self.assertIn("restoreUpdateModalIfNeeded", self.update_js)
+        self.assertIn("await restoreUpdateModalIfNeeded();", self.events_js)
+        self.assertIn('current.result === "reboot_required"', self.update_js)
+
+    def test_all_javascript_remains_syntax_valid(self):
+        node = shutil.which("node")
+        if not node:
+            self.skipTest("node is not available in this test environment")
+        for script in (
+            _REPO_ROOT / "app" / "sovran_systemsos_web" / "static" / "js"
+        ).glob("*.js"):
+            result = subprocess.run(
+                [node, "--check", str(script)], capture_output=True, text=True
+            )
+            self.assertEqual(result.returncode, 0, result.stderr)
+
+
+if __name__ == "__main__":
+    unittest.main()
@@ -0,0 +1,91 @@
+"""Regression tests for the Hub Zeus Connect QR.
+
+The LND-only rewrite of modules/bitcoin/lndconnect.nix shipped a wrapper
+that Zeus cannot use. These tests lock the contract the Hub QR depends on
+without needing lnd / tor / qrencode at test time.
+"""
+
+import os
+import re
+import unittest
+
+_REPO_ROOT = os.path.normpath(os.path.join(os.path.dirname(__file__), ".."))
+
+
+def _read(relpath: str) -> str:
+    with open(os.path.join(_REPO_ROOT, relpath), encoding="utf-8") as fh:
+        return fh.read()
+
+
+class TestLndconnectWrapper(unittest.TestCase):
+    """The system `lndconnect` wrapper must emit a Zeus-scannable URI."""
+
+    @classmethod
+    def setUpClass(cls):
+        cls.src = _read("modules/bitcoin/lndconnect.nix")
+
+    def test_uses_official_lndconnect_flags(self):
+        self.assertIn("--adminmacaroonpath=", self.src)
+        self.assertIn("--configfile=/dev/null", self.src)
+        self.assertIn("--nocert", self.src)
+        self.assertIn("--tlscertpath=", self.src)
+
+    def test_does_not_pass_unknown_short_flags(self):
+        # The broken rewrite called `lndconnect --cert … --macaroon …`.
+        # Those flags do not exist; Zeus then never got a valid URI.
+        self.assertIsNone(re.search(r"--cert=", self.src))
+        self.assertIsNone(re.search(r"--macaroon=", self.src))
+
+    def test_uses_dedicated_lnd_rest_onion(self):
+        self.assertIn("lnd-rest", self.src)
+        self.assertIn('onionServices.lnd-rest', self.src)
+        # Must not collide with the LND P2P onion named `lnd`.
+        self.assertNotIn("onionServices.lnd =", self.src)
+        self.assertNotIn('onionService = "${operatorName}/lnd"', self.src)
+
+    def test_reads_onion_from_onion_addresses_dir(self):
+        self.assertIn("onionAddresses.dataDir", self.src)
+        self.assertNotIn("/var/lib/tor/onion/${onionService}/hostname", self.src)
+
+    def test_omits_tls_cert_over_tor(self):
+        # Onion host + embedded localhost cert = Zeus rejects the QR.
+        self.assertIn('then "--nocert"', self.src)
+
+
+class TestZeusConnectSetup(unittest.TestCase):
+    """zeus-connect-setup must wait for the REST onion and validate the URI."""
+
+    @classmethod
+    def setUpClass(cls):
+        cls.src = _read("modules/wallet-autoconnect.nix")
+
+    def test_waits_for_onion_addresses(self):
+        self.assertIn("onion-addresses.service", self.src)
+
+    def test_rejects_non_lndconnect_output(self):
+        self.assertIn("^lndconnect://", self.src)
+        self.assertIn("\\.onion", self.src)
+        self.assertIn("macaroon=", self.src)
+
+    def test_no_clightning_fallback(self):
+        self.assertNotIn("lnconnect-clnrest", self.src)
+
+
+class TestHubQrFallback(unittest.TestCase):
+    """Hub QR encoding must not die on a payload that is too large for ECC H."""
+
+    def test_qrencode_falls_back_to_lower_ecc(self):
+        src = _read("app/sovran_systemsos_web/server.py")
+        self.assertIn('for ecc in ("H", "Q", "L")', src)
+
+    def test_qronly_does_not_hide_uri_when_qr_fails(self):
+        src = _read("app/sovran_systemsos_web/server.py")
+        self.assertIn("Don't hide the URI if we could not render a scannable QR.", src)
+
+    def test_zeus_guide_mentions_use_tor(self):
+        src = _read("app/sovran_systemsos_web/static/js/service-detail.js")
+        self.assertIn("Use Tor", src)
+
+
+if __name__ == "__main__":
+    unittest.main()
Author	SHA1	Message	Date
naturallaw777	d31887cae6	chore(release): prepare v1.1.2	2026-08-21 09:47:43 -05:00
naturallaw777	669181c75b	nixpkgs update with Bisq update	2026-08-20 17:04:36 -05:00
naturallaw777	47e9bb99b0	fix(public-ip): move system.activationScripts under the config attribute The module mixed the `options` keyword attribute with a bare top-level `system.*` setting. Once a module declares `options` (or `config`), every other top-level attribute must be a reserved module keyword — the nixpkgs unifyModuleSyntax check rejects anything else, so every nixos-rebuild aborted at evaluation time with: error: Module '.../modules/core/public-ip.nix' has an unsupported attribute `system'. ... move all of them (namely: system) into the `config' attribute. Prefix the activation script with `config.` (equivalent to wrapping it in `config = { ... };`) so the module evaluates again. The detector script itself is unchanged. Fixes: `ac6c498615` (-feat(public-ip): unify public-IP detection into one privacy-first script-)	2026-08-20 17:00:23 -05:00
naturallaw777	ac6c498615	feat(public-ip): unify public-IP detection into one privacy-first script The public IP was previously detected independently in three places, each contacting a different third party: the Hub (HTTPS echo via api.ipify.org / ifconfig.me / icanhazip.com on every API call and background tick), DDNS (myip.opendns.com via OpenDNS), and LiveKit (embedded STUN). Consolidate into a single detector with one shared cache so every consumer reads the same value with minimal exposure. - add modules/core/public-ip.nix: installs /var/lib/sovran/public-ip.py (pure Python stdlib, no new deps) writing /var/lib/secrets/external-ip - detection chain (first success wins): explicit pin, fresh cache (default TTL 300s), STUN binding request over UDP (one packet, no metadata), DNS myip.opendns.com query, then OPT-IN HTTPS echo (publicIP.httpsEcho, empty by default — never contacted unless listed) - privacy: while the cache is fresh zero third parties are contacted; at most one party learns the IP per refresh interval, via the least exposing mechanism available - hub (server.py): _get_external_ip() now reads the shared detector / cache instead of calling ipify/ifconfig/icanhazip directly - ddns (njalla.nix): use the shared detector instead of a separate OpenDNS dig; allow the hardened service to write /var/lib/secrets - element-calling: livekit-turn-setup falls back to the shared detector on cold boot; add LiveKit webhooks to lk-jwt-service (sfu_webhook) so abrupt disconnects are cleaned up immediately; set LIVEKIT_SANITY_CHECK_INTERVAL_SECONDS=60 as a missed-webhook guard; drop the dead services.livekit.settings block and set openFirewall=false (Caddy fronts the SFU; no public 7880/tcp) - new options: sovran_systemsOS.publicIP.{stunServer,stunPort, dnsResolver,httpsEcho,cacheTTL}	2026-08-20 16:40:00 -05:00
naturallaw777	224ea99ce4	fix(element-calling): reuse Hub external IP for LiveKit, drop egress service calls	2026-08-20 16:12:00 -05:00
naturallaw777	c54dbfe2a5	feat(element-calling): fix Element X discovery and harden federated calling The element-calling feature only advertised the LiveKit focus via the well-known org.matrix.msc4143.rtc_foci file, and relied on STUN auto-detection for the public IP. Element X queries the MatrixRTC transports registry endpoint and fails with MISSING_MATRIX_RTC_TRANSPORT when it is absent, and blocked STUN egress silently left LiveKit advertising a private IP (call connects but no video across servers). - synapse: enable msc4143_enabled and advertise matrix_rtc.transports (MSC4519) with the site's element-calling URL, so Element X can discover the LiveKit focus instead of erroring out - livekit: determine the public IP to advertise at runtime — explicit pin, then HTTPS egress detection (api.ipify.org / checkip.amazonaws.com / ifconfig.me), then STUN fallback with a warning; reject non-routable results (private/loopback/CGNAT) - lk-jwt-service: append optional extra homeservers to LIVEKIT_FULL_ACCESS_HOMESERVERS via the new sovran_systemsOS.elementCalling.fullAccessHomeservers option - add sovran_systemsOS.elementCalling.externalIP option to pin the advertised public IP for multi-WAN/VPN setups - add element-calling-public-check.service: boot-time diagnostics for public DNS (via 1.1.1.1, bypassing local loopback overrides), JWT healthz through Caddy and via the public IP, and the transports endpoint — turns the silent -no media- failure into a visible error - add restartTriggers so livekit/lk-jwt-service pick up regenerated runtime configs on rebuild	2026-08-20 14:07:06 -05:00
naturallaw777	a1fa40cacf	fix(hub): derive Restart required from boot default vs running system NixOS already knows whether a reboot is pending: /nix/var/nix/profiles/ system vs /run/current-system. Marker files only the Hub's own updater wrote desynced for terminal-updated machines (and markers from older updaters could never clear), pinning the badge on forever. Reconcile REBOOT_REQUIRED against live state on every read; the stale marker self-heals to IDLE. The .generation marker write is now informational.	2026-08-19 11:31:29 -05:00
naturallaw77	48dacbeef3	refactor(lnd): use pkgs.lndinit, drop local packages/lndinit The local packages/lndinit/default.nix is a verbatim copy of the upstream Nixpkgs expression, frozen at v0.1.3-beta (the version that was vendored in from nix-bitcoin before the Aug 10 2026 refactor in commit `1fbeafd`). It carries no Sovran-specific patches, no local overrides, and no behavioral modifications - it is byte-for-byte identical to what Nixpkgs ships, except ~19 minor versions older (Nixpkgs currently ships 0.1.22-beta; the developer upstream lightninglabs/lndinit is at v0.1.36-beta as of June 10 2026). Why this matters ---------------- The Aug 10 2026 refactor (`1fbeafd`, "refactor: move vendor/nix-bitcoin to modules/bitcoin, remove overlays") stated the new convention: No more random vendor/ or pkgs/ dirs - follows Sovran convention: modules/ for NixOS modules, packages/ for packages That refactor successfully removed: * pkgs/sovran-overlay.nix * pkgs/nbxplorer.nix * pkgs/README.md * modules/vendor/ (entire directory) * overlay-sovran from flake.nix It moved the lndinit expression into packages/lndinit/default.nix as an intermediate step, but the file is still a verbatim upstream copy and therefore still incurs the maintenance burden the refactor was meant to eliminate: manual version bumps, manual hash refreshes, and no upstream security or bug-fix flow. Removing it completes the intent of `1fbeafd`. The change ---------- modules/bitcoin/lnd.nix (line 153): - lndinit = "${(pkgs.callPackage ../../packages/lndinit {})}/bin/lndinit"; + lndinit = "${pkgs.lndinit}/bin/lndinit"; The two later uses of `lndinit` in the same file (lines 243 and 247, inside the systemd.services.lnd.preStart block that calls `lndinit gen-seed` and `lndinit init-wallet`) are unchanged because they reference the let-bound `lndinit` value, not the callPackage expression. They continue to work with the new pkgs.lndinit binary path transparently. Removed: * packages/lndinit/default.nix * packages/lndinit/ (now empty directory) No other files in the repository reference packages/lndinit. Verified by: * Git tree search for "packages/lndinit" -> only the file and its parent directory match * Content grep of flake.nix, configuration.nix, modules/bitcoin/default.nix, modules/bitcoin/common.nix, and iso/common.nix -> zero matches Why this is safe ---------------- 1. CLI compatibility. The preStart script only invokes two lndinit subcommands: * `lndinit gen-seed` * `lndinit -v init-wallet --file.seed=... --file.wallet-password=... --init-file.output-wallet-dir=...` Both subcommands and all four flags have been stable since the 0.1.x line. The Nixpkgs 0.1.22-beta binary produces a wallet.db and admin.macaroon in the same on-disk format that 0.1.3-beta did for the same LND version (LND is pinned separately by pkgs.lnd from Nixpkgs and is unaffected by this change). 2. No coupled Go modules or shared vendor tree. The local packages/lndinit/default.nix is a self-contained buildGoModule derivation; it has no shared state with any other Sovran package. 3. Nixpkgs pin is current. flake.nix pins github:NixOS/nixpkgs/nixos-unstable, which has shipped pkgs.lndinit since 2022 and is currently at 0.1.22-beta. There is no "missing attribute" risk. 4. Wallet data is forward-compatible. The wallet.db format is owned by LND, not lndinit. lndinit is only used at first boot to create the seed and initialize the wallet; subsequent LND restarts do not invoke lndinit. So even if a user already initialized a wallet with 0.1.3-beta, the binary being upgraded to 0.1.22-beta is irrelevant - LND owns the wallet from that point on. 5. Single call site. Only modules/bitcoin/lnd.nix references lndinit. No other modules, scripts, or tests need to change. Operational notes ----------------- * After this commit, lndinit updates flow through the normal `nix flake update` workflow (or whatever automated dependency tooling is already in use, e.g. for the recent "chore(deps): update RTL to 0.15.10" commits). No Sovran-side action is needed to pick up future lndinit versions. * If a future LND version requires a specific lndinit version, the pin can be done in flake.nix via a one-line overlay: nixpkgs.overlays = [ (final: prev: { lndinit = prev.lndinit.overrideAttrs (o: { version = "X.Y.Z-beta"; src = prev.fetchFromGitHub { ... }; vendorHash = "..."; }); }) ]; This keeps the upgrade path explicit without bringing the entire expression back into the Sovran tree. * This drops ~20 lines of frozen derivation code, eliminates one source of upstream drift, and reduces the surface area of what Sovran needs to keep current.	2026-08-18 12:30:59 -05:00
naturallaw777	b6e6adbe31	chore(deps): update RTL to 0.15.10 Refresh the RTL source and Node dependency hashes, and align Hub version metadata with the vendored package.	2026-08-18 11:19:23 -05:00
naturallaw777	64624002bb	fix(hub): reconcile completed updates after polling stalls The full-system updater runs as a detached systemd service and can finish successfully even when the browser loses its status connection. In that case the update log and status file correctly report REBOOT_REQUIRED, but the Hub modal can remain on "Updating..." with its controls disabled. There were four independent ways for the frontend to get stuck: * update status fetches had no deadline, so a request that stayed pending never rejected and never advanced the existing failure counter; * setInterval started async polls without waiting for the previous poll, allowing slow requests to overlap and responses to arrive out of order; * each log chunk used textContent +=, replacing the complete and growing Nix build log every two seconds, which could stall browser rendering and was especially visible over RDP; and * page reload, tab resume, and RDP reconnect did not reattach the modal to the update status persisted by the backend. This produced a dangerous UX mismatch: the machine had a fully staged NixOS generation and was ready to reboot, while the Hub continued telling the user that the update was still running. Bound status requests with AbortController, prevent overlapping polls, and replace the endless spinner after sustained failures with an explicit "Update status unavailable" state and Retry Status action. Reconcile state immediately on focus, visibility, online, page startup, and before starting a new update. Use no-store requests and render verbose logs incrementally with a bounded visible tail while retaining the complete report in memory. Apply the same timeout and single-flight protection to rebuild polling. Record the exact generation produced by `nixos-rebuild boot`. The Hub now keeps REBOOT_REQUIRED visible until that generation matches /run/current-system, then clears the marker after reboot. For an update started by an older updater that did not write the marker, recover the staged generation from the final nixos-rebuild log line. The dashboard sidebar also distinguishes update-in-progress and restart-required states. Regression coverage verifies generation marker/log recovery, pre- versus post-reboot detection, request timeout wiring, single-flight polling, connection-loss UX, RDP/tab resume reconciliation, bounded log rendering, page-reload recovery, and JavaScript syntax. Validation: * python3 -m unittest discover -s tests -p 'test_.py' -v (170 passed) node --check app/sovran_systemsos_web/static/js/.js python3 -m py_compile for changed Python modules * git diff --check A Nix evaluation was not available in the development sandbox; the NixOS module should still be evaluated and built in CI or on a test machine before release.	2026-08-18 10:31:28 -05:00
naturallaw777	1ccce429a5	fix(bitcoin): migrate i2pd SAM settings for nixpkgs 26.11 nixpkgs commit c8f9654 refactored the services.i2pd module to use an RFC42-style settings attribute set and removed services.i2pd.proto. After updating the root nixpkgs input from f13ff45 to ec2d622, the vendored bitcoind module failed evaluation on the obsolete services.i2pd.proto.sam.enable definition. The error occurred even with services.bitcoind.i2p at its false default: bitcoind was enabled, so NixOS still validated the obsolete option path inside the conditional i2pd integration. Read the SAM endpoint from services.i2pd.settings.sam and configure its new upstream-style fields explicitly. Keep 127.0.0.1:7656, matching the old typed option defaults that bitcoind uses to generate its i2psam setting. This preserves optional I2P support without activating it by default. i2pd remains disabled until services.bitcoind.i2p is set to true or "only-outgoing". Nixpkgs migration: https://github.com/NixOS/nixpkgs/commit/c8f965411e812060a9377fa4c2d7d0f84e8b10e0	2026-08-18 09:19:15 -05:00
naturallaw777	f832efb0d1	nixpkgs update	2026-08-18 08:59:35 -05:00
Sovran Systems	67ae53ad3f	Add tor-browser to the package list	2026-08-17 19:10:19 -05:00
naturallaw777	db5b9f6b60	fix(livekit): order turn-setup after network-online to fix boot-time red dot livekit-turn-setup.service detects the primary interface from the IPv4 default route, but had no ordering against network-online.target. With NetworkManager+DHCP the default route is applied late at boot, so the oneshot could run before it existed, exit 1, and — being a hard dependency of livekit.service — take livekit down with it. The Hub then showed a 'failed' red dot until livekit was restarted manually. Order both livekit.service and livekit-turn-setup.service after network-online.target. Also add a bounded retry when copying Caddy's ACME cert so we never write an empty turn.crt/turn.key on a fresh boot.	2026-08-17 19:07:54 -05:00
Sovran Systems	2ea1427766	fix: restore a Zeus-scannable LND REST connect QR The LND-only rewrite of lndconnect.nix shipped a wrapper Zeus cannot use: unknown flags (--cert/--macaroon), a non-existent onion path (free/lnd.onion), and a REST hidden service that collided with LND's P2P onion. Restore the nix-bitcoin contract — dedicated lnd-rest onion on port 8080, --nocert over Tor, admin macaroon in the URI — and only persist a valid lndconnect:// URI for the Hub QR.	2026-08-17 18:22:36 -05:00
Sovran Systems	b35b327a07	Merge pull request #435 from naturallaw777/security/lnd-macaroon-argv security: prevent LND admin macaroon exposure in curl argv	2026-08-15 23:03:44 -05:00
naturallaw777	a9ff168fd6	security: prevent LND admin macaroon exposure in curl argv	2026-08-15 23:00:59 -05:00
naturallaw777	b436dbed95	docs: update desktop and Sovran Hub screenshots	2026-08-15 18:53:15 -05:00
@@ -1 +1 @@
 .1.1
 .1.2