Merge pull request #330 from naturallaw777/copilot/implement-manual-backup-workflow

Implement reliable exFAT Manual Backup with tar artifacts, DB exports, and lifecycle hardening

app/sovran_systemsos_web/scripts/sovran-hub-backup.sh

@@ -17,13 +17,24 @@ set -euo pipefail
 BACKUP_LOG="/var/log/sovran-hub-backup.log"
 BACKUP_STATUS="/var/log/sovran-hub-backup.status"
 MEDIA_ROOT="/run/media"
-MIN_FREE_GB=10
 HUB_CONFIG_JSON="/var/lib/sovran-hub/config.json"
 ROLE_STATE_NIX="/etc/nixos/role-state.nix"
+SECOND_DRIVE_MOUNT="/run/media/Second_Drive"
+SAFETY_MARGIN_BYTES=$((1024 * 1024 * 1024))
 
 # ── Internal drive labels/paths to NEVER use as backup targets ───
 INTERNAL_LABELS=("BTCEcoandBackup" "sovran_systemsos")
-INTERNAL_MOUNTS=("/run/media/Second_Drive" "/boot/efi" "/")
+INTERNAL_MOUNTS=("$SECOND_DRIVE_MOUNT" "/boot/efi" "/")
+
+FAILED_ALREADY=0
+BACKUP_COMPLETE=0
+LND_STOPPED=0
+LND_UNITS_TO_RESTART=()
+
+ARCHIVE_FILES=()
+DB_DUMP_FILES=()
+MANIFEST_EXCLUDES=()
+LND_BACKUP_NOTES=()
 
 # ── Logging helpers ──────────────────────────────────────────────
 
@@ -37,16 +48,58 @@ set_status() {
 }
 
 fail() {
+  FAILED_ALREADY=1
   log "ERROR: $*"
   set_status "FAILED"
   exit 1
 }
 
+cleanup() {
+  local rc=$?
+  local restart_failed=0
+
+  if [[ "$LND_STOPPED" -eq 1 ]]; then
+    log "Restarting previously active LND-related services…"
+    for (( idx=${#LND_UNITS_TO_RESTART[@]}-1 ; idx>=0 ; idx-- )); do
+      local unit="${LND_UNITS_TO_RESTART[$idx]}"
+      if systemctl start "$unit"; then
+        log "Started $unit"
+      else
+        log "ERROR: Failed to start $unit"
+        restart_failed=1
+      fi
+    done
+    LND_STOPPED=0
+  fi
+
+  if [[ "$restart_failed" -eq 1 ]]; then
+    rc=1
+    FAILED_ALREADY=1
+    set_status "FAILED"
+  fi
+
+  if [[ "$BACKUP_COMPLETE" -eq 1 && "$rc" -eq 0 ]]; then
+    return
+  fi
+
+  if [[ "$FAILED_ALREADY" -eq 0 ]]; then
+    log "ERROR: Backup terminated unexpectedly (exit code $rc)."
+    set_status "FAILED"
+  fi
+}
+
+trap cleanup EXIT
+trap 'exit 1' INT TERM
+
+require_cmd() {
+  local cmd="$1"
+  command -v "$cmd" >/dev/null 2>&1 || fail "Required command not found: $cmd"
+}
+
 # ── Check whether a mount point is an internal drive ────────────
 
 is_internal() {
   local mnt="$1"
-  # Reject known internal mount points and their subdirectories
   for internal in "${INTERNAL_MOUNTS[@]}"; do
     if [[ "$mnt" == "$internal" || "$mnt" == "${internal}/"* ]]; then
       return 0
@@ -59,39 +112,37 @@ is_internal() {
 
 find_external_drive() {
   local target=""
-  # lsblk JSON output: NAME,LABEL,MOUNTPOINT,HOTPLUG,RM,TYPE
-  if command -v lsblk &>/dev/null; then
-    while IFS=$'\t' read -r dev_type hotplug removable label mountpoint; do
-      # Must be a partition or disk, and be removable/hotplug
-      [[ "$dev_type" == "part" || "$dev_type" == "disk" ]] || continue
-      [[ "$hotplug" == "1" || "$removable" == "1" ]] || continue
-      [[ -n "$mountpoint" ]] || continue
 
-      # Filter out internal labels
-      local skip=0
-      for lbl in "${INTERNAL_LABELS[@]}"; do
-        [[ "$label" == "$lbl" ]] && skip=1 && break
-      done
-      [[ "$skip" -eq 1 ]] && continue
+  while IFS=$'\t' read -r dev_type hotplug removable label mountpoint; do
+    [[ "$dev_type" == "part" || "$dev_type" == "disk" ]] || continue
+    [[ "$hotplug" == "1" || "$removable" == "1" ]] || continue
+    [[ -n "$mountpoint" ]] || continue
 
-      # Filter out internal mount points
-      is_internal "$mountpoint" && continue
+    local skip=0
+    for lbl in "${INTERNAL_LABELS[@]}"; do
+      [[ "$label" == "$lbl" ]] && skip=1 && break
+    done
+    [[ "$skip" -eq 1 ]] && continue
 
-      if mountpoint -q "$mountpoint" 2>/dev/null; then
-        target="$mountpoint"
-        break
-      fi
-    done < <(lsblk -J -o NAME,LABEL,MOUNTPOINT,HOTPLUG,RM,TYPE 2>/dev/null | \
-      python3 -c "
+    is_internal "$mountpoint" && continue
+
+    if mountpoint -q "$mountpoint" 2>/dev/null; then
+      target="$mountpoint"
+      break
+    fi
+  done < <(lsblk -J -o NAME,LABEL,MOUNTPOINT,HOTPLUG,RM,TYPE 2>/dev/null | \
+    python3 -c "
 import sys, json
-data = json.load(sys.stdin)
+
 def flatten(devs):
     for d in devs:
         yield d
         for c in d.get('children', []):
             yield from flatten([c])
+
+data = json.load(sys.stdin)
 for d in flatten(data.get('blockdevices', [])):
-    print('\t'.join([
+    print('\\t'.join([
         d.get('type') or '',
         str(d.get('hotplug') or '0'),
         str(d.get('rm') or '0'),
@@ -99,24 +150,10 @@ for d in flatten(data.get('blockdevices', [])):
         d.get('mountpoint') or '',
     ]))
 " 2>/dev/null || true)
-  fi
 
-  # Fallback: walk /run/media/ if lsblk produced nothing
   if [[ -z "$target" && -d "$MEDIA_ROOT" ]]; then
     while IFS= read -r -d '' mnt; do
       is_internal "$mnt" && continue
-      # Check label via lsblk on the device backing this mount
-      local dev
-      dev=$(findmnt -n -o SOURCE "$mnt" 2>/dev/null || true)
-      if [[ -n "$dev" ]]; then
-        local lbl
-        lbl=$(lsblk -n -o LABEL "$dev" 2>/dev/null || true)
-        local skip=0
-        for internal_lbl in "${INTERNAL_LABELS[@]}"; do
-          [[ "$lbl" == "$internal_lbl" ]] && skip=1 && break
-        done
-        [[ "$skip" -eq 1 ]] && continue
-      fi
       if mountpoint -q "$mnt" 2>/dev/null; then
         target="$mnt"
         break
@@ -128,16 +165,10 @@ for d in flatten(data.get('blockdevices', [])):
 }
 
 # ── Detect the configured system role ───────────────────────────
-#
-# Priority:
-#   1. Hub config JSON  (/var/lib/sovran-hub/config.json)  — "role" key
-#   2. role-state.nix   (/etc/nixos/role-state.nix)        — grep for true flag
-#   3. Default: server_plus_desktop
 
 detect_role() {
   local role="server_plus_desktop"
 
-  # 1. Try the Hub config JSON
   if [[ -f "$HUB_CONFIG_JSON" ]] && command -v python3 &>/dev/null; then
     local r
     r=$(python3 -c \
@@ -149,7 +180,6 @@ detect_role() {
     fi
   fi
 
-  # 2. Fall back to parsing role-state.nix
   if [[ -f "$ROLE_STATE_NIX" ]]; then
     if grep -q 'roles\.desktop = lib\.mkDefault true' "$ROLE_STATE_NIX" 2>/dev/null; then
       role="desktop"
@@ -161,6 +191,64 @@ detect_role() {
   echo "$role"
 }
 
+validate_target_mount() {
+  local target="$1"
+  [[ "$target" == "${MEDIA_ROOT}/"* ]] || fail "Target '$target' must be mounted under $MEDIA_ROOT."
+  [[ -d "$target" ]] || fail "Target path '$target' does not exist."
+  mountpoint -q "$target" || fail "Target path '$target' is not a mount point."
+
+  local fstype=""
+  fstype=$(findmnt -n -o FSTYPE -T "$target" 2>/dev/null || true)
+  [[ -n "$fstype" ]] || fail "Could not determine filesystem type for '$target'."
+
+  if [[ "$fstype" != "exfat" && "$fstype" != "fuseblk" ]]; then
+    fail "Target '$target' must be exFAT (detected filesystem: $fstype)."
+  fi
+
+  if [[ "$fstype" == "fuseblk" ]]; then
+    local src_dev blk_type
+    src_dev=$(findmnt -n -o SOURCE -T "$target" 2>/dev/null || true)
+    blk_type=""
+    if [[ -n "$src_dev" ]]; then
+      blk_type=$(lsblk -no FSTYPE "$src_dev" 2>/dev/null || true)
+      [[ -z "$blk_type" ]] && blk_type=$(blkid -o value -s TYPE "$src_dev" 2>/dev/null || true)
+    fi
+    if [[ "$blk_type" != "exfat" && "$blk_type" != "fuseblk" ]]; then
+      fail "Target '$target' is fuseblk but not identified as exFAT-compatible."
+    fi
+  fi
+
+  local write_test
+  write_test="$target/.sovran-write-test-$$"
+  if ! ( : > "$write_test" && echo "ok" >> "$write_test" && rm -f "$write_test" ); then
+    fail "Target '$target' is not writable."
+  fi
+
+  log "Verified backup target filesystem: $fstype"
+}
+
+has_unit() {
+  systemctl cat "$1" >/dev/null 2>&1
+}
+
+is_unit_active() {
+  systemctl is-active --quiet "$1"
+}
+
+estimate_path_bytes() {
+  local path="$1"
+  shift || true
+  [[ -e "$path" ]] || {
+    echo 0
+    return
+  }
+
+  local size
+  size=$(du -s -B1 -x "$@" "$path" 2>/dev/null | awk '{print $1}' || true)
+  [[ -n "$size" ]] || size=0
+  echo "$size"
+}
+
 # ── Initialise log file ──────────────────────────────────────────
 
 : > "$BACKUP_LOG"
@@ -169,14 +257,30 @@ set_status "RUNNING"
 log "=== Sovran_SystemsOS External Hub Backup ==="
 log "Starting backup process…"
 
+require_cmd tar
+require_cmd sha256sum
+require_cmd findmnt
+require_cmd lsblk
+require_cmd mountpoint
+require_cmd df
+require_cmd du
+require_cmd awk
+require_cmd sort
+require_cmd find
+require_cmd systemctl
+require_cmd hostname
+require_cmd date
+require_cmd python3
+require_cmd runuser
+
 # ── Detect system role ───────────────────────────────────────────
 
 ROLE="$(detect_role)"
 case "$ROLE" in
-  desktop)            ROLE_LABEL="Desktop Only" ;;
-  node)               ROLE_LABEL="Node (Bitcoin-only)" ;;
-  server_plus_desktop) ROLE_LABEL="Server + Desktop" ;;
-  *)                  ROLE_LABEL="$ROLE" ;;
+  desktop)              ROLE_LABEL="Desktop Only" ;;
+  node)                 ROLE_LABEL="Node (Bitcoin-only)" ;;
+  server_plus_desktop)  ROLE_LABEL="Server + Desktop" ;;
+  *)                    ROLE_LABEL="$ROLE" ;;
 esac
 log "Detected role: $ROLE_LABEL"
 
@@ -184,7 +288,6 @@ log "Detected role: $ROLE_LABEL"
 
 if [[ -n "${BACKUP_TARGET:-}" ]]; then
   TARGET="$BACKUP_TARGET"
-  # Safety: never allow internal drives even if explicitly passed
   if is_internal "$TARGET"; then
     fail "Target '$TARGET' is an internal system drive and cannot be used for external backup."
   fi
@@ -193,106 +296,362 @@ else
   log "Auto-detecting external USB drives…"
   TARGET="$(find_external_drive)"
   if [[ -z "$TARGET" ]]; then
-    fail "No external USB drive detected. " \
-         "Please plug in an exFAT-formatted USB drive (≥500 GB) and try again."
+    fail "No external USB drive detected. Please plug in an exFAT-formatted USB drive and try again."
   fi
   log "Detected external drive: $TARGET"
 fi
 
-# ── Verify mount point ───────────────────────────────────────────
+validate_target_mount "$TARGET"
 
-[[ -d "$TARGET" ]] || fail "Target path '$TARGET' does not exist."
-mountpoint -q "$TARGET" || fail "Target path '$TARGET' is not a mount point."
+# ── Plan role-aware source scope and exclusions ─────────────────
 
-# ── Check free disk space (require ≥ 10 GB) ──────────────────────
+LND_AVAILABLE=0
+if [[ "$ROLE" != "desktop" ]] && [[ -d /var/lib/lnd ]] && has_unit "lnd.service"; then
+  LND_AVAILABLE=1
+fi
 
-FREE_KB=$(df -k --output=avail "$TARGET" | tail -1)
-FREE_GB=$(( FREE_KB / 1024 / 1024 ))
+if [[ "$ROLE" == "desktop" ]]; then
+  MANIFEST_EXCLUDES+=("/etc/nix-bitcoin-secrets (not applicable for Desktop Only role)")
+else
+  MANIFEST_EXCLUDES+=("/etc/nix-bitcoin-secrets skipped when path absent")
+fi
+
+MANIFEST_EXCLUDES+=(
+  "/run/media/Second_Drive (never traversed)"
+  "/run/media/Second_Drive/BTCEcoandBackup/Bitcoin_Node (excluded; internal second-drive data)"
+  "/run/media/Second_Drive/BTCEcoandBackup/Electrs_Data (excluded; internal second-drive data)"
+  "/var/lib/bitcoind (excluded from manual backup)"
+  "/var/lib/electrs (excluded from manual backup)"
+  "/var/lib/*/log and /var/lib/*/logs"
+  "/var/lib/*/cache and /var/lib/*/tmp"
+  "/home/*/.cache and /home/*/.local/share/Trash"
+)
+
+if [[ "$ROLE" == "desktop" || "$LND_AVAILABLE" -eq 1 ]]; then
+  MANIFEST_EXCLUDES+=("/var/lib/lnd from general /var/lib archive")
+fi
+
+# ── Estimate required free space ─────────────────────────────────
+
+ETC_NIXOS_BYTES=$(estimate_path_bytes /etc/nixos)
+HOME_BYTES=$(estimate_path_bytes /home --exclude='*/.cache' --exclude='*/.local/share/Trash' --exclude='*/Trash')
+SECRETS_BYTES=0
+if [[ "$ROLE" != "desktop" ]]; then
+  SECRETS_BYTES=$(estimate_path_bytes /etc/nix-bitcoin-secrets)
+fi
+
+VAR_LIB_BYTES=$(estimate_path_bytes /var/lib \
+  --exclude='bitcoind' \
+  --exclude='electrs' \
+  --exclude='lnd' \
+  --exclude='*/log' \
+  --exclude='*/logs' \
+  --exclude='*/cache' \
+  --exclude='*/tmp')
+
+LND_BYTES=0
+if [[ "$LND_AVAILABLE" -eq 1 ]]; then
+  LND_BYTES=$(estimate_path_bytes /var/lib/lnd)
+fi
+
+ESTIMATED_BYTES=$(( ETC_NIXOS_BYTES + HOME_BYTES + SECRETS_BYTES + VAR_LIB_BYTES + LND_BYTES ))
+# Require 20% growth headroom plus an additional fixed 1 GiB safety margin.
+REQUIRED_BYTES=$(( ESTIMATED_BYTES + (ESTIMATED_BYTES / 5) + SAFETY_MARGIN_BYTES ))
+
+FREE_BYTES=$(df -B1 --output=avail "$TARGET" | tail -1 | tr -d ' ')
+FREE_GB=$(( FREE_BYTES / 1024 / 1024 / 1024 ))
+REQUIRED_GB=$(( REQUIRED_BYTES / 1024 / 1024 / 1024 ))
+
+log "Estimated backup size: $(( ESTIMATED_BYTES / 1024 / 1024 / 1024 )) GB"
+log "Required free space (with safety margin): ${REQUIRED_GB} GB"
 log "Free space on drive: ${FREE_GB} GB"
-(( FREE_GB >= MIN_FREE_GB )) || \
-  fail "Not enough free space on drive (${FREE_GB} GB available, ${MIN_FREE_GB} GB required)."
+
+(( FREE_BYTES >= REQUIRED_BYTES )) || \
+  fail "Not enough free space on drive (${FREE_GB} GB available, ${REQUIRED_GB} GB required)."
 
 # ── Create timestamped backup directory ─────────────────────────
 
 TIMESTAMP="$(date '+%Y%m%d_%H%M%S')"
 BACKUP_DIR="${TARGET}/Sovran_SystemsOS_Backup/${TIMESTAMP}"
-mkdir -p "$BACKUP_DIR"
+DB_DUMP_DIR="$BACKUP_DIR/database-dumps"
+mkdir -p "$BACKUP_DIR" "$DB_DUMP_DIR"
 log "Backup destination: $BACKUP_DIR"
 
-# ── Stage 1/4: NixOS configuration ──────────────────────────────
+create_tar_archive() {
+  local archive_name="$1"
+  shift
+  local archive_path="$BACKUP_DIR/$archive_name"
+
+  log "Creating $archive_name …"
+  tar \
+    --create \
+    --file "$archive_path" \
+    --numeric-owner \
+    --acls \
+    --xattrs \
+    --sparse \
+    --one-file-system \
+    "$@"
+
+  ARCHIVE_FILES+=("$archive_name")
+  log "Created archive: $archive_name"
+}
+
+export_postgresql_dumps() {
+  if ! command -v pg_dump >/dev/null 2>&1 || ! has_unit "postgresql.service"; then
+    log "PostgreSQL tools/service not available — skipping PostgreSQL exports."
+    return
+  fi
+
+  if ! is_unit_active "postgresql.service"; then
+    log "PostgreSQL service is not active — skipping PostgreSQL exports."
+    return
+  fi
+
+  log "Exporting PostgreSQL globals and databases…"
+  local globals_file="$DB_DUMP_DIR/postgresql_globals.sql"
+  runuser -u postgres -- pg_dumpall --globals-only > "$globals_file" || \
+    fail "Failed to export PostgreSQL globals."
+  DB_DUMP_FILES+=("database-dumps/postgresql_globals.sql")
+
+  local dbs
+  dbs=$(runuser -u postgres -- psql -Atqc "SELECT datname FROM pg_database WHERE datistemplate = false AND datallowconn AND datname <> 'postgres';" 2>/dev/null || true)
+
+  if [[ -z "$dbs" ]]; then
+    log "No non-template PostgreSQL application databases found."
+    return
+  fi
+
+  while IFS= read -r db; do
+    [[ -n "$db" ]] || continue
+    local safe_db
+    safe_db="$(echo "$db" | tr -c '[:alnum:]_.-' '_')"
+    local out_file="$DB_DUMP_DIR/postgresql_${safe_db}.dump"
+    runuser -u postgres -- pg_dump --format=custom --file "$out_file" "$db" || \
+      fail "Failed to export PostgreSQL database '$db'."
+    DB_DUMP_FILES+=("database-dumps/postgresql_${safe_db}.dump")
+  done <<< "$dbs"
+}
+
+export_mariadb_dumps() {
+  local dump_cmd=""
+  local query_cmd=""
+  local mariadb_unit=""
+
+  if command -v mariadb-dump >/dev/null 2>&1; then
+    dump_cmd="mariadb-dump"
+  elif command -v mysqldump >/dev/null 2>&1; then
+    dump_cmd="mysqldump"
+  fi
+
+  if command -v mariadb >/dev/null 2>&1; then
+    query_cmd="mariadb"
+  elif command -v mysql >/dev/null 2>&1; then
+    query_cmd="mysql"
+  fi
+
+  if [[ -z "$dump_cmd" || -z "$query_cmd" ]]; then
+    log "MariaDB dump/query tools not available — skipping MariaDB exports."
+    return
+  fi
+
+  if has_unit "mariadb.service" && is_unit_active "mariadb.service"; then
+    mariadb_unit="mariadb.service"
+  elif has_unit "mysql.service" && is_unit_active "mysql.service"; then
+    mariadb_unit="mysql.service"
+  else
+    log "MariaDB service is not active — skipping MariaDB exports."
+    return
+  fi
+
+  log "Exporting MariaDB databases from ${mariadb_unit}…"
+
+  local dbs
+  dbs=$($query_cmd -N -e "SHOW DATABASES" 2>/dev/null || true)
+  if [[ -z "$dbs" ]]; then
+    log "No MariaDB databases found."
+    return
+  fi
+
+  while IFS= read -r db; do
+    [[ -n "$db" ]] || continue
+    case "$db" in
+      information_schema|performance_schema|mysql|sys) continue ;;
+    esac
+
+    local safe_db out_file
+    safe_db="$(echo "$db" | tr -c '[:alnum:]_.-' '_')"
+    out_file="$DB_DUMP_DIR/mariadb_${safe_db}.sql"
+
+    $dump_cmd --single-transaction --quick --routines --events --triggers "$db" > "$out_file" || \
+      fail "Failed to export MariaDB database '$db'."
+
+    DB_DUMP_FILES+=("database-dumps/mariadb_${safe_db}.sql")
+  done <<< "$dbs"
+}
+
+export_lnd_scb_if_possible() {
+  [[ "$LND_AVAILABLE" -eq 1 ]] || return
+
+  local scb_file="$BACKUP_DIR/lnd-static-channel-backup.scb"
+  local attempts=(
+    "lncli exportchanbackup --all --output_file $scb_file"
+    "lncli -n mainnet exportchanbackup --all --output_file $scb_file"
+    "runuser -u lnd -- lncli exportchanbackup --all --output_file $scb_file"
+    "runuser -u lnd -- lncli -n mainnet exportchanbackup --all --output_file $scb_file"
+  )
+
+  if ! command -v lncli >/dev/null 2>&1; then
+    log "lncli not available — skipping Static Channel Backup export."
+    LND_BACKUP_NOTES+=("Static Channel Backup skipped (lncli unavailable)")
+    return
+  fi
+
+  if ! is_unit_active "lnd.service"; then
+    log "LND service is not active — skipping Static Channel Backup export."
+    LND_BACKUP_NOTES+=("Static Channel Backup skipped (lnd.service inactive)")
+    return
+  fi
+
+  log "Exporting LND Static Channel Backup…"
+  local attempt
+  for attempt in "${attempts[@]}"; do
+    if eval "$attempt" >/dev/null 2>&1; then
+      DB_DUMP_FILES+=("lnd-static-channel-backup.scb")
+      LND_BACKUP_NOTES+=("Static Channel Backup exported via lncli")
+      log "LND Static Channel Backup exported."
+      return
+    fi
+  done
+
+  log "WARNING: Unable to export LND Static Channel Backup with available lncli invocations."
+  LND_BACKUP_NOTES+=("Static Channel Backup export failed (no compatible lncli invocation succeeded)")
+}
+
+capture_active_lnd_dependents() {
+  [[ "$LND_AVAILABLE" -eq 1 ]] || return
+
+  LND_UNITS_TO_RESTART=()
+  local raw_units=""
+  raw_units=$(systemctl show lnd.service -p RequiredBy -p WantedBy --value 2>/dev/null | tr ' ' '\n' | grep '\.service$' | sort -u || true)
+
+  while IFS= read -r unit; do
+    [[ -n "$unit" ]] || continue
+    if is_unit_active "$unit"; then
+      LND_UNITS_TO_RESTART+=("$unit")
+    fi
+  done <<< "$raw_units"
+
+  if is_unit_active "lnd.service"; then
+    LND_UNITS_TO_RESTART+=("lnd.service")
+  fi
+}
+
+stop_lnd_stack_if_needed() {
+  [[ "$LND_AVAILABLE" -eq 1 ]] || return
+
+  capture_active_lnd_dependents
+
+  if [[ "${#LND_UNITS_TO_RESTART[@]}" -eq 0 ]]; then
+    log "No active LND-related services needed stopping."
+    return
+  fi
+
+  log "Stopping active services that depend on LND for clean /var/lib/lnd archive…"
+
+  local unit
+  for unit in "${LND_UNITS_TO_RESTART[@]}"; do
+    if [[ "$unit" == "lnd.service" ]]; then
+      continue
+    fi
+    systemctl stop "$unit" || fail "Failed to stop dependent service: $unit"
+    log "Stopped $unit"
+  done
+
+  if printf '%s\n' "${LND_UNITS_TO_RESTART[@]}" | grep -qx 'lnd.service'; then
+    systemctl stop lnd.service || fail "Failed to stop lnd.service"
+    log "Stopped lnd.service"
+  fi
+
+  LND_STOPPED=1
+}
+
+# ── Stage 1/5: NixOS configuration ──────────────────────────────
 
 log ""
-log "── Stage 1/4: NixOS configuration (/etc/nixos) ──────────────"
+log "── Stage 1/5: NixOS configuration (/etc/nixos) ──────────────"
 if [[ -d /etc/nixos ]]; then
-  rsync -a --info=progress2 /etc/nixos/ "$BACKUP_DIR/nixos/" 2>&1 | tee -a "$BACKUP_LOG" || \
-    fail "Stage 1 failed while copying /etc/nixos"
+  create_tar_archive "etc-nixos.tar" -C / etc/nixos
   log "Stage 1 complete."
 else
   log "WARNING: /etc/nixos not found — skipping."
 fi
 
-# ── Stage 2/4: Secrets ──────────────────────────────────────────
+# ── Stage 2/5: Secrets ──────────────────────────────────────────
 
 log ""
-log "── Stage 2/4: Secrets ───────────────────────────────────────"
-mkdir -p "$BACKUP_DIR/secrets"
-
+log "── Stage 2/5: Secrets (/etc/nix-bitcoin-secrets) ───────────"
 if [[ "$ROLE" == "desktop" ]]; then
   log "Skipping /etc/nix-bitcoin-secrets — not applicable for Desktop Only role."
+elif [[ -e /etc/nix-bitcoin-secrets ]]; then
+  create_tar_archive "etc-nix-bitcoin-secrets.tar" -C / etc/nix-bitcoin-secrets
 else
-  if [[ -e /etc/nix-bitcoin-secrets ]]; then
-    rsync -a --info=progress2 /etc/nix-bitcoin-secrets "$BACKUP_DIR/secrets/" 2>&1 | tee -a "$BACKUP_LOG" || \
-      log "WARNING: Could not copy /etc/nix-bitcoin-secrets — continuing."
-  else
-    log "  (not found: /etc/nix-bitcoin-secrets — skipping)"
-  fi
+  log "(not found: /etc/nix-bitcoin-secrets — skipping)"
 fi
-
 log "Stage 2 complete."
 
-# ── Stage 3/4: Home directory ────────────────────────────────────
+# ── Stage 3/5: Home directory ───────────────────────────────────
 
 log ""
-log "── Stage 3/4: Home directory (/home) ───────────────────────"
+log "── Stage 3/5: Home directory (/home) ───────────────────────"
 if [[ -d /home ]]; then
-  rsync -a --info=progress2 \
-    --exclude='.cache/' \
-    --exclude='.local/share/Trash/' \
-    --exclude='*/Trash/' \
-    /home/ "$BACKUP_DIR/home/" 2>&1 | tee -a "$BACKUP_LOG" || \
-    fail "Stage 3 failed while copying /home"
+  create_tar_archive "home.tar" \
+    -C / \
+    --exclude='home/*/.cache' \
+    --exclude='home/*/.local/share/Trash' \
+    --exclude='home/*/Trash' \
+    home
   log "Stage 3 complete."
 else
   log "WARNING: /home not found — skipping."
 fi
 
-# ── Stage 4/4: System data ───────────────────────────────────────
+# ── Stage 4/5: Database exports + LND artifacts ────────────────
 
 log ""
-log "── Stage 4/4: System data (/var/lib) ────────────────────────"
-if [[ "$ROLE" == "desktop" ]]; then
-  if [[ -d /var/lib ]]; then
-    rsync -a --info=progress2 \
-      --filter='- /lnd/***' \
-      --exclude='logs/' \
-      --exclude='log/' \
-      --exclude='*/logs/' \
-      --exclude='*/log/' \
-      /var/lib/ "$BACKUP_DIR/var-lib/" 2>&1 | tee -a "$BACKUP_LOG" || \
-      fail "Stage 4 failed while copying /var/lib for Desktop Only role"
-    log "Stage 4 complete (Desktop Only role excludes /var/lib/lnd)."
-  else
-    log "WARNING: /var/lib not found — skipping."
+log "── Stage 4/5: Database and LND consistency exports ─────────"
+export_postgresql_dumps
+export_mariadb_dumps
+export_lnd_scb_if_possible
+
+if [[ "$LND_AVAILABLE" -eq 1 ]]; then
+  stop_lnd_stack_if_needed
+  create_tar_archive "var-lib-lnd-clean.tar" -C / var/lib/lnd
+  LND_BACKUP_NOTES+=("Created clean raw /var/lib/lnd archive after controlled service stop")
+fi
+
+log "Stage 4 complete."
+
+# ── Stage 5/5: System data ──────────────────────────────────────
+
+log ""
+log "── Stage 5/5: System data (/var/lib) ───────────────────────"
+if [[ -d /var/lib ]]; then
+  VAR_LIB_EXCLUDES=(
+    --exclude='var/lib/bitcoind'
+    --exclude='var/lib/electrs'
+    --exclude='var/lib/*/log'
+    --exclude='var/lib/*/logs'
+    --exclude='var/lib/*/cache'
+    --exclude='var/lib/*/tmp'
+  )
+
+  if [[ "$ROLE" == "desktop" || "$LND_AVAILABLE" -eq 1 ]]; then
+    VAR_LIB_EXCLUDES+=(--exclude='var/lib/lnd')
   fi
-elif [[ -d /var/lib ]]; then
-  rsync -a --info=progress2 \
-    --exclude='logs/' \
-    --exclude='log/' \
-    --exclude='*/logs/' \
-    --exclude='*/log/' \
-    /var/lib/ "$BACKUP_DIR/var-lib/" 2>&1 | tee -a "$BACKUP_LOG" || \
-    fail "Stage 4 failed while copying /var/lib"
-  log "Stage 4 complete."
+
+  create_tar_archive "var-lib.tar" -C / "${VAR_LIB_EXCLUDES[@]}" var/lib
+  log "Stage 5 complete."
 else
   log "WARNING: /var/lib not found — skipping."
 fi
@@ -301,21 +660,81 @@ fi
 
 log ""
 log "Generating BACKUP_MANIFEST.txt …"
+MANIFEST_FILE="$BACKUP_DIR/BACKUP_MANIFEST.txt"
+CHECKSUM_FILE="$BACKUP_DIR/SHA256SUMS.txt"
+
 {
   echo "Sovran_SystemsOS Backup Manifest"
-  echo "Generated: $(date)"
+  echo "Generated: $(date -u '+%Y-%m-%dT%H:%M:%SZ')"
+  echo "Timestamp: $TIMESTAMP"
   echo "Hostname:  $(hostname)"
   echo "Role:      $ROLE_LABEL"
   echo "Target:    $TARGET"
   echo ""
-  echo "Contents:"
-  find "$BACKUP_DIR" -mindepth 1 -maxdepth 2 | sort
-} > "$BACKUP_DIR/BACKUP_MANIFEST.txt"
-log "Manifest written to $BACKUP_DIR/BACKUP_MANIFEST.txt"
+  echo "Source paths included:"
+  echo "- /etc/nixos"
+  echo "- /home"
+  if [[ "$ROLE" != "desktop" ]]; then
+    echo "- /etc/nix-bitcoin-secrets (when present)"
+  fi
+  echo "- /var/lib"
+  echo ""
+  echo "Exclusions:"
+  for ex in "${MANIFEST_EXCLUDES[@]}"; do
+    echo "- $ex"
+  done
+  echo ""
+  echo "Archives:"
+  for archive in "${ARCHIVE_FILES[@]}"; do
+    echo "- $archive"
+  done
+  echo ""
+  echo "Database and LND exports:"
+  if [[ "${#DB_DUMP_FILES[@]}" -eq 0 && "${#LND_BACKUP_NOTES[@]}" -eq 0 ]]; then
+    echo "- none"
+  else
+    for dump in "${DB_DUMP_FILES[@]}"; do
+      echo "- $dump"
+    done
+    for note in "${LND_BACKUP_NOTES[@]}"; do
+      echo "- $note"
+    done
+  fi
+  echo ""
+  echo "Restore guidance:"
+  echo "- Verify artifacts: cd <backup_dir> && sha256sum -c SHA256SUMS.txt"
+  echo "- Extract a tar archive: sudo tar --acls --xattrs --numeric-owner -xpf <archive>.tar -C /"
+  echo "- PostgreSQL globals: sudo -u postgres psql -f database-dumps/postgresql_globals.sql"
+  echo "- PostgreSQL DB dump: sudo -u postgres pg_restore --create --clean --if-exists -d postgres database-dumps/postgresql_<db>.dump"
+  echo "- MariaDB DB dump: mariadb <db_name> < database-dumps/mariadb_<db>.sql"
+  echo "- LND SCB: keep lnd-static-channel-backup.scb with wallet seed for channel recovery procedures"
+  echo ""
+  echo "Important note: Bitcoin blockchain and Electrs index data are intentionally excluded"
+  echo "from manual external backup because they already live on the internal second drive"
+  echo "(/run/media/Second_Drive) and are reconstructable/internal-backup data."
+  echo ""
+  echo "Artifact listing:"
+  find "$BACKUP_DIR" -mindepth 1 -maxdepth 2 -type f | sort
+} > "$MANIFEST_FILE"
+
+# ── Generate checksums for all backup artifacts ─────────────────
+
+log "Generating SHA-256 checksums …"
+(
+  cd "$BACKUP_DIR"
+  while IFS= read -r -d '' file; do
+    sha256sum "$file"
+  done < <(find . -mindepth 1 -maxdepth 2 -type f ! -name 'SHA256SUMS.txt' -print0 | sort -z)
+) > "$CHECKSUM_FILE"
+
+log "Manifest written to $MANIFEST_FILE"
+log "Checksums written to $CHECKSUM_FILE"
 
 # ── Done ─────────────────────────────────────────────────────────
 
 log ""
 log "All Finished! Your data is now backed up to a third location."
 log "Please eject the drive safely before removing it from your Sovran Pro."
+
+BACKUP_COMPLETE=1
 set_status "SUCCESS"

app/sovran_systemsos_web/server.py

@@ -8,6 +8,7 @@ import contextlib
 import glob
 import hashlib
 import hmac
+import ipaddress
 import json
 import logging
 import os
@@ -80,6 +81,15 @@ DOMAINS_DIR = "/var/lib/domains"
 NOSTR_NPUB_FILE   = "/var/lib/secrets/nostr_npub"
 NJALLA_SCRIPT     = "/var/lib/njalla/njalla.sh"
 
+# Systemd service that rewrites the Sovran-managed /etc/hosts loopback block
+SOVRAN_HOSTS_SERVICE = "sovran-hosts-update.service"
+
+# Domain keys that produce a public HTTPS virtual host via Caddy
+_SERVICE_DOMAIN_KEYS = frozenset([
+    "matrix", "wordpress", "nextcloud", "btcpayserver",
+    "vaultwarden", "haven", "element-calling",
+])
+
 INTERNAL_IP_FILE = "/var/lib/secrets/internal-ip"
 ZEUS_CONNECT_FILE = "/var/lib/secrets/zeus-connect-url"
 
@@ -964,18 +974,94 @@ def _check_port_status(
     return "closed"
 
 
+
+# Regex for validating domain values written into /etc/hosts.  Rejects anything
+# containing whitespace, newlines, or characters that could escape a hosts entry.
+# NOTE: The equivalent pattern in modules/core/local-domain-loopback.nix (shell
+# grep -E) must be kept in sync with this Python regex.
+_SAFE_DOMAIN_RE = re.compile(
+    r'^(?:[a-zA-Z0-9](?:[a-zA-Z0-9\-]{0,61}[a-zA-Z0-9])?\.)+[a-zA-Z]{2,}$'
+)
+
+
+def _validate_domain_value(domain: str) -> bool:
+    """Return True if *domain* is a valid hostname safe to write into /etc/hosts.
+
+    Rejects values containing whitespace, newlines, or other characters that
+    could inject additional entries or corrupt the hosts file.
+    """
+    if not domain or len(domain) > 253:
+        return False
+    # Guard against newline / whitespace injection before regex check.
+    if any(c in domain for c in ('\n', '\r', ' ', '\t', '#')):
+        return False
+    return bool(_SAFE_DOMAIN_RE.match(domain))
+
+
+def _is_loopback_address(ip: str) -> bool:
+    """Return True if *ip* is a loopback address (127.0.0.0/8 or ::1)."""
+    try:
+        return ipaddress.ip_address(ip).is_loopback
+    except ValueError:
+        return False
+
+
+def _resolve_all_addresses(domain: str) -> list[str]:
+    """Return all unique IP addresses that *domain* resolves to, or an empty list.
+
+    The first element is the address that the system resolver would normally
+    use for a connection.  All elements are checked when determining whether
+    any address matches the expected public IP or is a loopback address.
+    """
+    try:
+        results = socket.getaddrinfo(domain, None)
+        unique_addresses: list[str] = []
+        for r in results:
+            addr = r[4][0]
+            if addr not in unique_addresses:
+                unique_addresses.append(addr)
+        return unique_addresses
+    except Exception:
+        return []
+
+
+def _trigger_hosts_update() -> None:
+    """Start the sovran-hosts-update systemd service (best-effort, no-op if unavailable)."""
+    try:
+        subprocess.run(
+            ["systemctl", "start", SOVRAN_HOSTS_SERVICE],
+            timeout=30,
+            check=False,
+            capture_output=True,
+        )
+    except Exception:
+        pass
+
+
 def _check_domain_reachable(domain: str) -> dict:
-    """Curl the domain to verify end-to-end HTTPS reachability."""
+    """Check HTTPS reachability for *domain* via local Caddy (loopback).
+
+    Using ``--resolve`` ensures the request reaches Caddy on this computer
+    without depending on router NAT loopback or the public DNS result.
+    A successful local check is sufficient to confirm that Caddy and the
+    virtual-host configuration are working correctly.
+    """
     try:
         result = subprocess.run(
-            ["curl", "-sS", "-o", "/dev/null", "-w", "%{http_code}", "--max-time", "10", f"https://{domain}"],
+            [
+                "curl", "-sS", "-o", "/dev/null", "-w", "%{http_code}",
+                "--max-time", "10",
+                "--resolve", f"{domain}:443:127.0.0.1",
+                "--resolve", f"{domain}:80:127.0.0.1",
+                f"https://{domain}",
+            ],
             capture_output=True,
             text=True,
             timeout=15,
         )
         status_code = result.stdout.strip()
         if status_code and status_code.isdigit() and int(status_code) > 0:
-            return {"reachable": True, "status_code": int(status_code)}
+            return {"reachable": True, "status_code": int(status_code), "via_loopback": True}
         return {"reachable": False, "error": result.stderr.strip() or "No response"}
     except subprocess.TimeoutExpired:
         return {"reachable": False, "error": "timeout"}
@@ -984,25 +1070,28 @@ def _check_domain_reachable(domain: str) -> dict:
 
 
 def _check_domain_health_fast(domain: str | None, external_ip: str) -> bool:
-    """Fast domain issue check for tile health (no curl/subprocess calls)."""
+    """Fast domain issue check for tile health (no curl/subprocess calls).
+
+    Returns ``True`` when a domain issue is detected that warrants
+    ``needs_attention``, ``False`` otherwise.
+    Loopback resolution is treated as an intentional server-local override,
+    not a DNS mismatch.
+    """
     if not domain:
         return True
 
-    resolved_ip: str | None = None
-    try:
-        results = socket.getaddrinfo(domain, None)
-        if results:
-            resolved_ip = results[0][4][0]
-    except socket.gaierror:
-        resolved_ip = None
-    except Exception:
-        resolved_ip = None
-
-    if not resolved_ip:
+    addrs = _resolve_all_addresses(domain)
+    if not addrs:
         return True
+
+    # If every resolved address is loopback the intentional /etc/hosts
+    # override is in place — this is healthy, not a mismatch.
+    if all(_is_loopback_address(a) for a in addrs):
+        return False
+
     if external_ip == "unavailable":
         return False
-    return resolved_ip != external_ip
+    return not any(a == external_ip for a in addrs)
 
 
 def _is_domain_reachable_cached(domain: str) -> bool | None:
@@ -1070,15 +1159,8 @@ def _evaluate_domain_checklist(
         "detail": domain,
     })
 
-    resolved_ip: str | None = None
-    try:
-        results = socket.getaddrinfo(domain, None)
-        if results:
-            resolved_ip = results[0][4][0]
-    except socket.gaierror:
-        resolved_ip = None
-    except Exception:
-        resolved_ip = None
+    addrs = _resolve_all_addresses(domain)
+    resolved_ip: str | None = addrs[0] if addrs else None
 
     if not resolved_ip:
         domain_status = {
@@ -1105,7 +1187,31 @@ def _evaluate_domain_checklist(
             "has_issues": True,
         }
 
-    if external_ip == "unavailable":
+    # Detect intentional server-local loopback override from /etc/hosts.
+    # When all addresses are loopback the public DNS is not checked via the
+    # system resolver (which would always return the override).  We proceed
+    # to the reachability check so Caddy health can still be verified.
+    loopback_override = all(_is_loopback_address(a) for a in addrs)
+
+    if loopback_override:
+        domain_status = {
+            "status": "local_override",
+            "resolved_ip": resolved_ip,
+            "expected_ip": external_ip,
+        }
+        steps.append({
+            "step": 2,
+            "label": "DNS / Local Override",
+            "status": "ok",
+            "detail": (
+                "Server-local loopback override is active — this computer routes the domain "
+                "directly to Caddy without going through the router. "
+                "Public DNS cannot be verified from this computer while the override is in place. "
+                "To check your public DNS from outside, use a tool such as "
+                "https://dnschecker.org or run: dig @1.1.1.1 " + domain
+            ),
+        })
+    elif external_ip == "unavailable":
         domain_status = {
             "status": "error",
             "resolved_ip": resolved_ip,
@@ -1117,7 +1223,7 @@ def _evaluate_domain_checklist(
             "status": "warning",
             "detail": f"Resolves to {resolved_ip} (external IP unavailable for comparison)",
         })
-    elif resolved_ip != external_ip:
+    elif not any(a == external_ip for a in addrs):
         domain_status = {
             "status": "dns_mismatch",
             "resolved_ip": resolved_ip,
@@ -1345,6 +1451,12 @@ def _read_backup_status() -> str:
         return "IDLE"
 
 
+def _write_backup_status(value: str) -> None:
+    """Write backup status file."""
+    with open(BACKUP_STATUS, "w") as f:
+        f.write(value)
+
+
 def _read_backup_log(offset: int = 0) -> tuple[str, int]:
     """Read the backup log file from the given byte offset.
     Returns (new_text, new_offset)."""
@@ -1361,6 +1473,12 @@ def _read_backup_log(offset: int = 0) -> tuple[str, int]:
         return "", 0
 
 
+def _append_backup_log(line: str) -> None:
+    """Append one line to backup log."""
+    with open(BACKUP_LOG, "a") as f:
+        f.write(line.rstrip("\n") + "\n")
+
+
 _INTERNAL_LABELS  = {"BTCEcoandBackup", "sovran_systemsos"}
 _INTERNAL_MOUNTS  = {"/", "/boot/efi"}
 _INTERNAL_MOUNT_PREFIX = "/run/media/Second_Drive"
@@ -1375,6 +1493,41 @@ def _is_internal_mount(mnt: str) -> bool:
     return False
 
 
+def _is_supported_backup_fstype(path: str, fstype: str) -> bool:
+    """Return whether the target filesystem type is supported for manual backup."""
+    fstype = (fstype or "").lower()
+    if fstype == "exfat":
+        return True
+    if fstype != "fuseblk":
+        return False
+
+    src_dev = ""
+    try:
+        result = subprocess.run(
+            ["findmnt", "-n", "-o", "SOURCE", "-T", path],
+            capture_output=True, text=True, timeout=5,
+        )
+        if result.returncode == 0:
+            src_dev = result.stdout.strip()
+    except Exception:
+        src_dev = ""
+
+    if not src_dev:
+        return False
+
+    for cmd in (
+        ["lsblk", "-no", "FSTYPE", src_dev],
+        ["blkid", "-o", "value", "-s", "TYPE", src_dev],
+    ):
+        try:
+            result = subprocess.run(cmd, capture_output=True, text=True, timeout=5)
+            if result.returncode == 0 and result.stdout.strip().lower() in {"exfat", "fuseblk"}:
+                return True
+        except Exception:
+            continue
+    return False
+
+
 def _detect_external_drives() -> list[dict]:
     """Scan for mounted external USB drives.
 
@@ -1384,7 +1537,8 @@ def _detect_external_drives() -> list[dict]:
     /run/media/ directly if lsblk is unavailable, applying the same
     label/path filters.
 
-    Returns a list of dicts with name, path, free_gb, total_gb.
+    Returns:
+        list[dict]: Each dict contains name, path, free_gb, total_gb, fstype.
     """
     import json as _json
     import subprocess as _subprocess
@@ -1395,7 +1549,7 @@ def _detect_external_drives() -> list[dict]:
     # ── Primary path: lsblk JSON ────────────────────────────────
     try:
         result = _subprocess.run(
-            ["lsblk", "-J", "-o", "NAME,LABEL,MOUNTPOINT,HOTPLUG,RM,TYPE"],
+            ["lsblk", "-J", "-o", "NAME,LABEL,FSTYPE,MOUNTPOINT,HOTPLUG,RM,TYPE"],
             capture_output=True, text=True, timeout=10
         )
         if result.returncode == 0:
@@ -1413,6 +1567,7 @@ def _detect_external_drives() -> list[dict]:
                 hotplug    = str(dev.get("hotplug", "0"))
                 rm         = str(dev.get("rm", "0"))
                 label      = dev.get("label") or ""
+                fstype     = (dev.get("fstype") or "").lower()
                 mountpoint = dev.get("mountpoint") or ""
 
                 if dev_type not in ("part", "disk"):
@@ -1438,6 +1593,7 @@ def _detect_external_drives() -> list[dict]:
                         "path":     mountpoint,
                         "free_gb":  free_gb,
                         "total_gb": total_gb,
+                        "fstype":   fstype,
                     })
                     seen_paths.add(mountpoint)
                 except OSError:
@@ -1471,11 +1627,20 @@ def _detect_external_drives() -> list[dict]:
                     st = os.statvfs(drive_path)
                     total_gb = round((st.f_blocks * st.f_frsize) / (1024 ** 3), 1)
                     free_gb  = round((st.f_bavail * st.f_frsize) / (1024 ** 3), 1)
+                    fstype = ""
+                    try:
+                        fstype = _subprocess.run(
+                            ["findmnt", "-n", "-o", "FSTYPE", "-T", drive_path],
+                            capture_output=True, text=True, timeout=5
+                        ).stdout.strip().lower()
+                    except Exception:
+                        fstype = ""
                     drives.append({
                         "name":     drive_name,
                         "path":     drive_path,
                         "free_gb":  free_gb,
                         "total_gb": total_gb,
+                        "fstype":   fstype,
                     })
                     seen_paths.add(drive_path)
                 except OSError:
@@ -2749,19 +2914,17 @@ async def api_services():
                     break
         has_domain_issues = False
         if needs_domain and domain and enabled:
+            addrs = _resolve_all_addresses(domain)
             dns_ok = True
-            try:
-                results = socket.getaddrinfo(domain, None)
-                if results:
-                    resolved_ip = results[0][4][0]
-                    if (
-                        _cached_external_ip != "unavailable"
-                        and resolved_ip != _cached_external_ip
-                    ):
-                        dns_ok = False
-                else:
-                    dns_ok = False
-            except (socket.gaierror, Exception):
+            if not addrs:
+                dns_ok = False
+            elif all(_is_loopback_address(a) for a in addrs):
+                # Intentional server-local /etc/hosts override — not a mismatch.
+                dns_ok = True
+            elif (
+                _cached_external_ip != "unavailable"
+                and not any(a == _cached_external_ip for a in addrs)
+            ):
                 dns_ok = False
 
             if not dns_ok:
@@ -3578,6 +3741,22 @@ async def api_backup_drives():
     return {"drives": drives}
 
 
+async def _monitor_backup_subprocess(proc: asyncio.subprocess.Process) -> None:
+    """Mark status FAILED if backup subprocess exits unexpectedly."""
+    rc = await proc.wait()
+    if rc == 0:
+        return
+
+    loop = asyncio.get_event_loop()
+    status = await loop.run_in_executor(None, _read_backup_status)
+    if status in {"SUCCESS", "FAILED"}:
+        return
+
+    msg = f"[{time.strftime('%Y-%m-%d %H:%M:%S')}] ERROR: Backup subprocess exited unexpectedly (code {rc})."
+    await loop.run_in_executor(None, _append_backup_log, msg)
+    await loop.run_in_executor(None, _write_backup_status, "FAILED")
+
+
 @app.post("/api/backup/run")
 async def api_backup_run(target: str = ""):
     """Start the backup script as a background subprocess.
@@ -3588,6 +3767,26 @@ async def api_backup_run(target: str = ""):
     if status == "RUNNING":
         return {"ok": True, "status": "already_running"}
 
+    drives = await loop.run_in_executor(None, _detect_external_drives)
+    if not drives:
+        raise HTTPException(status_code=400, detail="No external backup drive detected.")
+
+    drive_map = {d.get("path", ""): d for d in drives if d.get("path")}
+    if target:
+        if target not in drive_map:
+            raise HTTPException(status_code=400, detail="Selected backup target is not an available external drive.")
+        selected = drive_map[target]
+    else:
+        selected = drives[0]
+
+    selected_target = selected.get("path", "")
+    selected_fstype = (selected.get("fstype") or "").lower()
+    if selected_fstype and not _is_supported_backup_fstype(selected_target, selected_fstype):
+        raise HTTPException(
+            status_code=400,
+            detail=f"Selected drive filesystem '{selected_fstype}' is not supported for manual backup.",
+        )
+
     # Clear stale log before starting
     try:
         with open(BACKUP_LOG, "w") as f:
@@ -3595,21 +3794,34 @@ async def api_backup_run(target: str = ""):
     except OSError:
         pass
 
-    env = dict(os.environ)
-    if target:
-        env["BACKUP_TARGET"] = target
+    try:
+        await loop.run_in_executor(None, _write_backup_status, "RUNNING")
+    except OSError as exc:
+        raise HTTPException(status_code=500, detail=f"Could not set backup status: {exc}")
 
-    # Fire-and-forget: the script writes its own status/log files.
-    # Progress is read by the client via /api/backup/status (same pattern
-    # as /api/updates/run and the rebuild feature).
-    await asyncio.create_subprocess_exec(
-        "/usr/bin/env", "bash", BACKUP_SCRIPT,
-        stdout=asyncio.subprocess.DEVNULL,
-        stderr=asyncio.subprocess.DEVNULL,
-        env=env,
+    await loop.run_in_executor(
+        None,
+        _append_backup_log,
+        f"[{time.strftime('%Y-%m-%d %H:%M:%S')}] Starting backup process…",
     )
 
-    return {"ok": True, "status": "started"}
+    env = dict(os.environ)
+    env["BACKUP_TARGET"] = selected_target
+
+    try:
+        proc = await asyncio.create_subprocess_exec(
+            "/usr/bin/env", "bash", BACKUP_SCRIPT,
+            stdout=asyncio.subprocess.DEVNULL,
+            stderr=asyncio.subprocess.DEVNULL,
+            env=env,
+        )
+    except Exception as exc:
+        await loop.run_in_executor(None, _append_backup_log, f"[{time.strftime('%Y-%m-%d %H:%M:%S')}] ERROR: Failed to launch backup script: {exc}")
+        await loop.run_in_executor(None, _write_backup_status, "FAILED")
+        raise HTTPException(status_code=500, detail="Failed to launch backup process.")
+
+    asyncio.create_task(_monitor_backup_subprocess(proc))
+    return {"ok": True, "status": "started", "target": selected_target}
 
 
 # ── Feature Manager endpoints ─────────────────────────────────────
@@ -3886,6 +4098,12 @@ async def api_domains_set(req: DomainSetRequest):
         except Exception:
             pass
 
+    # Regenerate the server-local /etc/hosts loopback entries so the newly
+    # saved domain is immediately reachable on this computer without NAT
+    # loopback support on the router.
+    if req.domain_name in _SERVICE_DOMAIN_KEYS:
+        _trigger_hosts_update()
+
     return {"ok": True}
 
 
@@ -3933,38 +4151,35 @@ async def api_domains_check(req: DomainCheckRequest):
     external_ip = _cached_external_ip
 
     def check_domain(domain: str) -> dict:
-        try:
-            results = socket.getaddrinfo(domain, None)
-            if not results:
-                return {
-                    "domain": domain, "status": "unresolvable",
-                    "resolved_ip": None, "expected_ip": external_ip,
-                }
-            resolved_ip = results[0][4][0]
-            if external_ip == "unavailable":
-                return {
-                    "domain": domain, "status": "error",
-                    "resolved_ip": resolved_ip, "expected_ip": external_ip,
-                }
-            if resolved_ip == external_ip:
-                return {
-                    "domain": domain, "status": "connected",
-                    "resolved_ip": resolved_ip, "expected_ip": external_ip,
-                }
-            return {
-                "domain": domain, "status": "dns_mismatch",
-                "resolved_ip": resolved_ip, "expected_ip": external_ip,
-            }
-        except socket.gaierror:
+        addrs = _resolve_all_addresses(domain)
+        if not addrs:
             return {
                 "domain": domain, "status": "unresolvable",
                 "resolved_ip": None, "expected_ip": external_ip,
             }
-        except Exception:
+        resolved_ip = addrs[0]
+        # Server-local /etc/hosts loopback override — report as such rather
+        # than as a DNS mismatch.  Public DNS cannot be verified from this
+        # computer when the override is active.
+        if all(_is_loopback_address(a) for a in addrs):
+            return {
+                "domain": domain, "status": "local_override",
+                "resolved_ip": resolved_ip, "expected_ip": external_ip,
+            }
+        if external_ip == "unavailable":
             return {
                 "domain": domain, "status": "error",
-                "resolved_ip": None, "expected_ip": external_ip,
+                "resolved_ip": resolved_ip, "expected_ip": external_ip,
             }
+        if any(a == external_ip for a in addrs):
+            return {
+                "domain": domain, "status": "connected",
+                "resolved_ip": resolved_ip, "expected_ip": external_ip,
+            }
+        return {
+            "domain": domain, "status": "dns_mismatch",
+            "resolved_ip": resolved_ip, "expected_ip": external_ip,
+        }
 
     check_results = await asyncio.gather(*[
         loop.run_in_executor(None, check_domain, d) for d in req.domains

app/sovran_systemsos_web/static/css/header.css

@@ -91,3 +91,30 @@
   border-color: var(--accent-color);
   color: var(--accent-color);
 }
+
+/* ── Header reboot button ───────────────────────────────────────── */
+
+.btn-header-reboot {
+  background: transparent;
+  border: 1px solid rgba(184, 125, 0, 0.35);
+  color: #c98d08;
+  font-size: 0.78rem;
+  font-weight: 600;
+  padding: 4px 12px;
+  border-radius: var(--radius-btn);
+  cursor: pointer;
+  transition: border-color 0.15s, color 0.15s, background-color 0.15s;
+}
+
+.btn-header-reboot:hover {
+  border-color: #b87d00;
+  color: #e0a010;
+  background-color: rgba(184, 125, 0, 0.1);
+}
+
+@media (max-width: 480px) {
+  .btn-header-reboot {
+    padding: 4px 8px;
+    font-size: 0.72rem;
+  }
+}

app/sovran_systemsos_web/static/css/modals.css

@@ -102,6 +102,48 @@ button.btn-reboot:hover:not(:disabled) {
   background-color: #529E7E;
 }
 
+/* Restart = AMBER (manual restart action) */
+.btn-restart-amber {
+  background-color: #b87d00;
+  color: #fff;
+}
+
+.btn-restart-amber:hover:not(:disabled) {
+  background-color: #9a6800;
+}
+
+/* Restart conflict warning box */
+.restart-conflict-box {
+  background-color: rgba(180, 100, 0, 0.12);
+  border-left: 3px solid #c97a00;
+  border-radius: 6px;
+  padding: 12px 14px;
+  margin-bottom: 14px;
+}
+
+.restart-conflict-title {
+  font-size: 0.88rem;
+  font-weight: 700;
+  color: #e69000;
+  margin: 0 0 6px 0;
+}
+
+.restart-conflict-desc {
+  font-size: 0.83rem;
+  color: var(--text-secondary);
+  line-height: 1.5;
+  margin: 0;
+}
+
+/* Reboot error card actions row */
+.reboot-error-actions {
+  display: flex;
+  gap: 12px;
+  justify-content: center;
+  flex-wrap: wrap;
+  margin-top: 20px;
+}
+
 .btn-save {
   background-color: var(--yellow);
   color: #0A1A10;

app/sovran_systemsos_web/static/js/events.js

@@ -44,6 +44,28 @@ if ($upgradeCloseBtn) $upgradeCloseBtn.addEventListener("click", closeUpgradeMod
 if ($upgradeCancelBtn) $upgradeCancelBtn.addEventListener("click", closeUpgradeModal);
 if ($upgradeModal) $upgradeModal.addEventListener("click", function(e) { if (e.target === $upgradeModal) closeUpgradeModal(); });
 
+// Restart confirm dialog
+if ($restartConfirmCancel) $restartConfirmCancel.addEventListener("click", closeRestartConfirmDialog);
+if ($restartConfirmModal) $restartConfirmModal.addEventListener("click", function(e) { if (e.target === $restartConfirmModal) closeRestartConfirmDialog(); });
+if ($restartConfirmModal) $restartConfirmModal.addEventListener("keydown", function(e) { if (e.key === "Escape") closeRestartConfirmDialog(); });
+
+// Header Reboot button
+if ($headerRebootBtn) $headerRebootBtn.addEventListener("click", function() { openRestartConfirmDialog(); });
+if ($restartConfirmOk) $restartConfirmOk.addEventListener("click", function() {
+  if ($restartConfirmOk.disabled) return;
+  $restartConfirmOk.disabled = true;
+  closeRestartConfirmDialog();
+  doReboot();
+});
+
+// Reboot error card buttons
+var $rebootErrorCloseBtn = document.getElementById("reboot-error-close-btn");
+var $rebootErrorRetryBtn = document.getElementById("reboot-error-retry-btn");
+if ($rebootErrorCloseBtn) $rebootErrorCloseBtn.addEventListener("click", function() {
+  if ($rebootOverlay) $rebootOverlay.classList.remove("visible");
+});
+if ($rebootErrorRetryBtn) $rebootErrorRetryBtn.addEventListener("click", doReboot);
+
 // ── Upgrade modal functions ───────────────────────────────────────
 
 function openUpgradeModal() {
@@ -54,6 +76,37 @@ function closeUpgradeModal() {
   if ($upgradeModal) $upgradeModal.classList.remove("open");
 }
 
+// ── Restart confirm dialog functions ─────────────────────────────
+
+var _restartDialogOpener = null;
+
+function openRestartConfirmDialog() {
+  if (!$restartConfirmModal) return;
+  _restartDialogOpener = document.activeElement;
+
+  // Detect conflicting operations
+  var isOperationInProgress = !!_updatePollTimer || !!_rebuildPollTimer;
+  if ($restartConflictBox) $restartConflictBox.style.display = isOperationInProgress ? "" : "none";
+  if ($restartConfirmOk) $restartConfirmOk.disabled = isOperationInProgress;
+
+  $restartConfirmModal.classList.add("open");
+
+  // Focus Cancel initially for safety
+  var cancelBtn = document.getElementById("restart-confirm-cancel-btn");
+  if (cancelBtn) setTimeout(function() { cancelBtn.focus(); }, 50);
+}
+
+function closeRestartConfirmDialog() {
+  if ($restartConfirmModal) $restartConfirmModal.classList.remove("open");
+  // Re-enable confirm button for next open
+  if ($restartConfirmOk) $restartConfirmOk.disabled = false;
+  // Return focus to the element that opened the dialog
+  if (_restartDialogOpener && _restartDialogOpener.focus) {
+    try { _restartDialogOpener.focus(); } catch (_) {}
+    _restartDialogOpener = null;
+  }
+}
+
 async function doUpgradeToServer() {
   var confirmBtn = $upgradeConfirmBtn;
   if (confirmBtn) { confirmBtn.disabled = true; confirmBtn.textContent = "Upgrading…"; }

app/sovran_systemsos_web/static/js/rebuild.js

@@ -69,7 +69,7 @@ function onRebuildDone(result) {
     // Auto-reload the page after a short delay so tiles and toggles reflect the new state
     setTimeout(function() { window.location.reload(); }, 1200);
   } else if (result === "reboot_required") {
-    if ($rebuildStatus) $rebuildStatus.textContent = "✓ Done — reboot required";
+    if ($rebuildStatus) $rebuildStatus.textContent = "✓ Done — restart required";
     if ($rebuildReboot) $rebuildReboot.style.display = "inline-flex";
   } else {
     if ($rebuildStatus) $rebuildStatus.textContent = "✗ Something went wrong";

app/sovran_systemsos_web/static/js/security.js

@@ -145,28 +145,25 @@ function openSecurityModal() {
           if (rebootBtn) {
             // Keep button disabled for 5 seconds to prevent accidental clicks
             var countdown = 5;
-            rebootBtn.textContent = "I have written down my new password \u2014 Reboot now (" + countdown + ")";
+            rebootBtn.textContent = "I have written down my new password \u2014 Restart Entire System (" + countdown + ")";
             var timer = setInterval(function() {
               countdown--;
               if (countdown <= 0) {
                 clearInterval(timer);
                 rebootBtn.disabled = false;
-                rebootBtn.textContent = "I have written down my new password \u2014 Reboot now";
+                rebootBtn.textContent = "I have written down my new password \u2014 Restart Entire System";
               } else {
-                rebootBtn.textContent = "I have written down my new password \u2014 Reboot now (" + countdown + ")";
+                rebootBtn.textContent = "I have written down my new password \u2014 Restart Entire System (" + countdown + ")";
               }
             }, 1000);
 
             rebootBtn.addEventListener("click", function() {
               rebootBtn.disabled = true;
-              rebootBtn.textContent = "Rebooting\u2026";
-              if ($rebootOverlay) $rebootOverlay.classList.add("visible");
-              _rebootStartTime = Date.now();
-              _serverWentDown = false;
-              setTimeout(waitForServerReboot, REBOOT_INITIAL_DELAY);
-              var rebootCtrl = new AbortController();
-              setTimeout(function() { rebootCtrl.abort(); }, REBOOT_REQUEST_TIMEOUT);
-              fetch("/api/reboot", { method: "POST", signal: rebootCtrl.signal }).catch(function() {});
+              rebootBtn.textContent = "Restarting\u2026";
+              // Hide the security reset overlay so the shared reboot overlay is visible
+              var $secResetOverlay2 = document.getElementById("security-reset-overlay");
+              if ($secResetOverlay2) $secResetOverlay2.classList.remove("visible");
+              doReboot();
             }, { once: true });
           }
         } catch (err) {

app/sovran_systemsos_web/static/js/state.js

@@ -49,6 +49,9 @@ const $btnSave        = document.getElementById("btn-save-report");
 const $btnCloseModal  = document.getElementById("btn-close-modal");
 
 const $rebootOverlay  = document.getElementById("reboot-overlay");
+const $rebootMainCard = document.getElementById("reboot-main-card");
+const $rebootErrorCard = document.getElementById("reboot-error-card");
+const $rebootSubmessage = document.getElementById("reboot-submessage");
 
 const $credsModal     = document.getElementById("creds-modal");
 const $credsTitle     = document.getElementById("creds-modal-title");
@@ -101,5 +104,14 @@ const $upgradeConfirmBtn  = document.getElementById("upgrade-confirm-btn");
 const $upgradeCancelBtn   = document.getElementById("upgrade-cancel-btn");
 const $upgradeCloseBtn    = document.getElementById("upgrade-close-btn");
 
+// Restart confirm dialog
+const $restartConfirmModal  = document.getElementById("restart-confirm-modal");
+const $restartConfirmOk     = document.getElementById("restart-confirm-ok-btn");
+const $restartConfirmCancel = document.getElementById("restart-confirm-cancel-btn");
+const $restartConflictBox   = document.getElementById("restart-conflict-box");
+
+// Header reboot button
+const $headerRebootBtn = document.getElementById("btn-header-reboot");
+
 // System status banner
 // (removed — health is now shown per-tile via the composite health field)

app/sovran_systemsos_web/static/js/support.js

@@ -491,7 +491,7 @@ function renderBackupReady(drives) {
       '<div class="support-steps-title">Requirements</div>',
       '<ol class="support-backup-steps">',
         '<li>USB hard drive plugged into one of the open USB ports on your Sovran Pro</li>',
-        '<li>At least 500 GB of free space on the drive</li>',
+        '<li>Enough free space for your selected backup data (the backup checks this before starting)</li>',
         '<li>Drive must be formatted as <strong>exFAT</strong></li>',
       '</ol>',
     '</div>',
@@ -584,9 +584,10 @@ async function pollBackupStatus() {
       logDiv.scrollTop = logDiv.scrollHeight;
     }
     _backupLogOffset = data.offset;
-    if (!data.running) {
+    const result = (data.result || "").toLowerCase();
+    if (result === "success" || result === "failed") {
       stopBackupPoll();
-      renderBackupDone(data.result === "success");
+      renderBackupDone(result === "success");
     }
   } catch (_) {}
 }

app/sovran_systemsos_web/static/js/update.js

@@ -154,7 +154,7 @@ function onUpdateDone(result) {
     if ($modalStatus) $modalStatus.textContent = "✓ Update complete";
     if ($btnReboot) $btnReboot.style.display = "inline-flex";
   } else if (result === "reboot_required") {
-    if ($modalStatus) $modalStatus.textContent = "✓ Update complete — reboot required";
+    if ($modalStatus) $modalStatus.textContent = "✓ Update complete — restart required";
     if ($btnReboot) $btnReboot.style.display = "inline-flex";
   } else {
     if ($modalStatus) $modalStatus.textContent = "✗ Update failed";
@@ -179,23 +179,50 @@ function saveErrorReport() {
 
 var _rebootStartTime = 0;
 var _serverWentDown = false;
+var _rebootFailed = false;
+
+function _setRebootStatus(msg) {
+  if ($rebootSubmessage) $rebootSubmessage.textContent = msg;
+}
 
 function doReboot() {
   if ($modal) $modal.classList.remove("open");
   if ($rebuildModal) $rebuildModal.classList.remove("open");
   stopUpdatePoll();
   stopRebuildPoll();
+  // Reset overlay to main card
+  if ($rebootMainCard) $rebootMainCard.style.display = "";
+  if ($rebootErrorCard) $rebootErrorCard.style.display = "none";
+  _setRebootStatus("Sending restart request\u2026");
   if ($rebootOverlay) $rebootOverlay.classList.add("visible");
   _rebootStartTime = Date.now();
   _serverWentDown = false;
+  _rebootFailed = false;
   var rebootCtrl = new AbortController();
   setTimeout(function() { rebootCtrl.abort(); }, REBOOT_REQUEST_TIMEOUT);
-  fetch("/api/reboot", { method: "POST", signal: rebootCtrl.signal }).catch(function() {});
+  fetch("/api/reboot", { method: "POST", signal: rebootCtrl.signal })
+    .then(function(res) {
+      if (!res.ok) {
+        // Definitive HTTP error — server rejected the request before going down
+        _rebootFailed = true;
+        if ($rebootMainCard) $rebootMainCard.style.display = "none";
+        if ($rebootErrorCard) $rebootErrorCard.style.display = "";
+        // Leave overlay visible so the error card is shown
+      }
+      // HTTP 2xx: request accepted, proceed with polling
+    })
+    .catch(function() {
+      // Connection dropped or request aborted — the server is likely already going
+      // down as part of the restart. Treat as success and continue polling.
+    });
   // Wait before the first check — NixOS shutdown after an update can take 20-40s
   setTimeout(waitForServerReboot, REBOOT_INITIAL_DELAY);
 }
 
 function waitForServerReboot() {
+  if (_rebootFailed) return;
+  // Update status on first check (server hasn't gone down yet)
+  if (!_serverWentDown) _setRebootStatus("Waiting for the computer to shut down\u2026");
   var controller = new AbortController();
   var timeoutId = setTimeout(function() { controller.abort(); }, REBOOT_FETCH_TIMEOUT);
 
@@ -205,18 +232,23 @@ function waitForServerReboot() {
       if (_serverWentDown) {
         // Server is responding after having been down — reboot is complete.
         // Any response (even 401/500) means the server process is back.
+        _setRebootStatus("System is back online. Reconnecting\u2026");
         window.location.reload();
       } else if ((Date.now() - _rebootStartTime) < 90000) {
         // Server still responding but hasn't gone down yet — keep waiting
         setTimeout(waitForServerReboot, REBOOT_CHECK_INTERVAL);
       } else {
         // Been over 90 seconds and server is responding — just reload
+        _setRebootStatus("System is back online. Reconnecting\u2026");
         window.location.reload();
       }
     })
     .catch(function() {
       clearTimeout(timeoutId);
-      _serverWentDown = true;
+      if (!_serverWentDown) {
+        _serverWentDown = true;
+        _setRebootStatus("The computer is restarting\u2026");
+      }
       setTimeout(waitForServerReboot, REBOOT_CHECK_INTERVAL);
     });
 }

app/sovran_systemsos_web/templates/index.html

@@ -24,6 +24,7 @@
     <span class="title">Sovran_SystemsOS Hub</span>
     <div class="header-buttons">
       <span class="role-badge" id="role-badge">Loading…</span>
+      <button class="btn btn-header-reboot" id="btn-header-reboot" title="Restart the entire computer">Reboot</button>
       <button class="btn btn-logout" id="btn-logout" title="Sign out">Sign Out</button>
     </div>
   </header>
@@ -61,7 +62,7 @@
       <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-reboot" id="btn-reboot" style="display:none">Reboot</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>
     </div>
@@ -164,7 +165,7 @@
       <div class="modal-log" id="rebuild-log" aria-live="polite"></div>
       <div class="modal-footer">
         <button class="btn btn-save" id="rebuild-save-report" style="display:none">Save Error Report</button>
-        <button class="btn btn-reboot" id="rebuild-reboot-btn" style="display:none">Reboot</button>
+        <button class="btn btn-reboot" id="rebuild-reboot-btn" style="display:none">Restart Entire System</button>
         <button class="btn btn-close-modal" id="rebuild-close-btn" disabled>Close</button>
       </div>
     </div>
@@ -240,26 +241,61 @@
         You will need it to log in to your computer<br />and the Sovran Hub at <em>sovransystemsos.local</em>.
       </p>
       <button class="security-reset-reboot-btn" id="security-reset-reboot-btn" disabled>
-        I have written down my new password — Reboot now
+        I have written down my new password — Restart Entire System
       </button>
     </div>
   </div>
 
   <!-- Reboot overlay -->
   <div class="reboot-overlay" id="reboot-overlay">
-    <div class="reboot-card">
-      <div class="reboot-icon">↻</div>
-      <h2 class="reboot-title">System Rebooting</h2>
+    <!-- Normal restarting card -->
+    <div class="reboot-card" id="reboot-main-card">
+      <div class="reboot-icon" aria-hidden="true">↻</div>
+      <h2 class="reboot-title">Restarting Entire System</h2>
       <p class="reboot-message">
-        Sovran_SystemsOS is now restarting.<br />
-        This page will automatically reconnect once the system is back online.
+        The entire computer is restarting, including the desktop and all hosted services.<br />
+        This page will reconnect automatically when Sovran_SystemsOS is back online.
       </p>
-      <div class="reboot-dots">
+      <div class="reboot-dots" aria-hidden="true">
         <span class="reboot-dot"></span>
         <span class="reboot-dot"></span>
         <span class="reboot-dot"></span>
       </div>
-      <p class="reboot-submessage">Stay tuned…</p>
+      <p class="reboot-submessage" id="reboot-submessage" aria-live="polite">Sending restart request…</p>
+    </div>
+    <!-- Error card (shown if restart request fails definitively) -->
+    <div class="reboot-card" id="reboot-error-card" style="display:none">
+      <div class="reboot-icon" aria-hidden="true">⚠</div>
+      <h2 class="reboot-title">Restart could not be started</h2>
+      <p class="reboot-message">
+        The computer did not begin restarting. No services were intentionally stopped. Please try again.
+      </p>
+      <div class="reboot-error-actions">
+        <button class="btn btn-close-modal" id="reboot-error-close-btn">Close</button>
+        <button class="btn btn-restart-amber" id="reboot-error-retry-btn">Try Again</button>
+      </div>
+    </div>
+  </div>
+
+  <!-- Restart Confirm Dialog -->
+  <div class="modal-overlay" id="restart-confirm-modal" role="dialog" aria-modal="true" aria-labelledby="restart-confirm-title">
+    <div class="creds-dialog domain-narrow-dialog">
+      <div class="creds-header">
+        <span class="creds-title" id="restart-confirm-title">Restart the entire computer?</span>
+      </div>
+      <div class="creds-body">
+        <div id="restart-conflict-box" class="restart-conflict-box" style="display:none">
+          <p class="restart-conflict-title">The system cannot restart right now.</p>
+          <p class="restart-conflict-desc">A system update, rebuild, backup, restore, or security operation is currently running. Wait for it to finish, then try again.</p>
+        </div>
+        <p class="support-desc"><strong>This will reboot the physical machine running Sovran_SystemsOS — not just the Hub.</strong></p>
+        <p class="support-desc">The desktop and all hosted services will stop temporarily and restart with the computer. Anyone currently using these services will be disconnected.</p>
+        <p class="support-desc">The system usually returns within 1–3 minutes. This page will reconnect automatically.</p>
+        <div class="domain-field-actions">
+          <button class="btn btn-close-modal" id="restart-confirm-cancel-btn">Cancel</button>
+          <button class="btn btn-restart-amber" id="restart-confirm-ok-btn">Restart Entire System</button>
+        </div>
+      </div>
     </div>
   </div>
 

app/tests/test_local_domain_loopback_nix.py

@@ -0,0 +1,193 @@
+"""Structural regression tests for modules/core/local-domain-loopback.nix.
+
+Verifies that the sovran-hosts-update helper:
+  - is built as a pkgs.writeShellApplication with explicit runtimeInputs
+    (gawk, gnugrep, coreutils);
+  - uses ``lib.getExe hostsUpdateScript`` for both the systemd ExecStart and
+    the activation script so that both contexts share the same Nix-store
+    executable;
+  - does NOT point ExecStart at the raw /etc path;
+  - includes element-calling in the supported domain list;
+  - uses ``awk -v`` for safe marker-variable passing rather than interpolating
+    marker text directly into the awk program;
+  - retains the domain validation regex (idempotency / injection prevention);
+  - exposes /etc/sovran-hosts-update.sh as a symlink via ``source =`` (not a
+    second raw ``text =`` body);
+  - does NOT rely on environment.systemPackages for the helper's dependencies.
+"""
+
+import re
+import shutil
+import subprocess
+import unittest
+from pathlib import Path
+
+NIX_STRING_INDENT = 6
+REPO_ROOT = Path(__file__).resolve().parents[2]
+FLAKE_SOURCE = (REPO_ROOT / "flake.nix").read_text()
+PRIMARY_NIXOS_CONFIGURATION_MATCH = re.search(
+    r"nixosConfigurations\.([A-Za-z0-9_-]+)\s*=",
+    FLAKE_SOURCE,
+)
+if PRIMARY_NIXOS_CONFIGURATION_MATCH is None:
+    raise RuntimeError("Could not determine the primary nixosConfigurations entry from flake.nix")
+PRIMARY_NIXOS_CONFIGURATION = PRIMARY_NIXOS_CONFIGURATION_MATCH.group(1)
+HELPER_BUILD_ATTR = (
+    f'.#nixosConfigurations.{PRIMARY_NIXOS_CONFIGURATION}.config.environment.etc.'
+    '"sovran-hosts-update.sh".source'
+)
+NIX_FILE = (
+    REPO_ROOT
+    / "modules"
+    / "core"
+    / "local-domain-loopback.nix"
+)
+
+
+class LocalDomainLoopbackNixStructureTests(unittest.TestCase):
+    def setUp(self):
+        self.source = NIX_FILE.read_text()
+
+    def _helper_script(self) -> str:
+        start = self.source.index("text = ''") + len("text = ''")
+        end = self.source.index("    '';", start)
+        return "\n".join(
+            line[NIX_STRING_INDENT:]
+            if line.startswith(" " * NIX_STRING_INDENT)
+            else line
+            for line in self.source[start:end].splitlines()
+        ).lstrip("\n")
+
+    # ── writeShellApplication and explicit runtimeInputs ────────────────────
+
+    def test_uses_write_shell_application(self):
+        self.assertIn("pkgs.writeShellApplication", self.source)
+
+    def test_runtime_inputs_includes_coreutils(self):
+        self.assertIn("pkgs.coreutils", self.source)
+
+    def test_runtime_inputs_includes_gawk(self):
+        self.assertIn("pkgs.gawk", self.source)
+
+    def test_runtime_inputs_includes_gnugrep(self):
+        self.assertIn("pkgs.gnugrep", self.source)
+
+    def test_runtime_inputs_block_present(self):
+        self.assertIn("runtimeInputs", self.source)
+
+    # ── Both execution paths use lib.getExe ─────────────────────────────────
+
+    def test_exec_start_uses_lib_get_exe(self):
+        """systemd ExecStart must reference the Nix-store executable."""
+        self.assertIn("ExecStart = lib.getExe hostsUpdateScript", self.source)
+
+    def test_activation_script_uses_lib_get_exe(self):
+        """Activation text must call the same Nix-store executable."""
+        self.assertIn("${lib.getExe hostsUpdateScript}", self.source)
+
+    def test_exec_start_does_not_point_to_etc_path(self):
+        """ExecStart must NOT use the raw /etc path (which lacks a deterministic PATH)."""
+        self.assertNotIn('ExecStart = "/etc/sovran-hosts-update.sh"', self.source)
+
+    # ── /etc symlink uses source =, not a second text = body ────────────────
+
+    def test_etc_entry_uses_source_not_text(self):
+        """The /etc/sovran-hosts-update.sh entry must be a symlink (source =),
+        not a second raw script body (text =)."""
+        self.assertIn(
+            'environment.etc."sovran-hosts-update.sh".source', self.source
+        )
+
+    def test_etc_source_points_to_get_exe(self):
+        self.assertIn(
+            'environment.etc."sovran-hosts-update.sh".source = lib.getExe hostsUpdateScript',
+            self.source,
+        )
+
+    # ── element-calling domain is supported ─────────────────────────────────
+
+    def test_element_calling_domain_key_present(self):
+        self.assertIn("element-calling", self.source)
+
+    # ── Robust awk -v variable passing ──────────────────────────────────────
+
+    def test_awk_uses_dash_v_for_begin_marker(self):
+        """awk must receive the begin marker via -v, not by shell interpolation."""
+        self.assertIn('awk -v begin=', self.source)
+
+    def test_awk_uses_dash_v_for_end_marker(self):
+        self.assertIn('-v end=', self.source)
+
+    def test_awk_does_not_interpolate_marker_into_program(self):
+        """The old pattern interpolated $BEGIN_MARKER directly into the awk source."""
+        self.assertNotIn('/$BEGIN_MARKER', self.source)
+        self.assertNotIn('/$END_MARKER', self.source)
+
+    # ── Domain validation ────────────────────────────────────────────────────
+
+    def test_domain_validation_regex_present(self):
+        """The hostname validation regex must still be present for injection prevention."""
+        self.assertIn("grep -qE", self.source)
+        self.assertIn("[a-zA-Z0-9]", self.source)
+
+    def test_invalid_domain_warning_present(self):
+        self.assertIn("skipping invalid domain value", self.source)
+
+    # ── No environment.systemPackages reliance ───────────────────────────────
+
+    def test_no_environment_system_packages_for_helper(self):
+        """The helper's tools are declared via runtimeInputs; the module must
+        not add them to environment.systemPackages."""
+        self.assertNotIn("environment.systemPackages", self.source)
+
+    # ── Idempotency: existing Sovran block is removed before rewriting ───────
+
+    def test_existing_block_removal_logic_present(self):
+        """awk strip of the managed block must be present for idempotency."""
+        self.assertIn("skip=1", self.source)
+        self.assertIn("skip=0", self.source)
+
+    def test_managed_block_uses_grouped_append_redirect(self):
+        self.assertIn('} >> "$TMP"', self.source)
+        self.assertEqual(self.source.count('>> "$TMP"'), 1)
+
+    def test_helper_script_passes_shellcheck(self):
+        shellcheck = shutil.which("shellcheck")
+        if shellcheck is None:
+            self.skipTest("shellcheck is not installed")
+        proc = subprocess.run(
+            [shellcheck, "-s", "bash", "-"],
+            input=self._helper_script(),
+            text=True,
+            capture_output=True,
+            check=False,
+        )
+        output = proc.stdout + proc.stderr
+        self.assertEqual(proc.returncode, 0, output)
+
+    def test_helper_derivation_builds_when_nix_available(self):
+        nix = shutil.which("nix")
+        if nix is None:
+            self.skipTest("nix is not installed")
+        proc = subprocess.run(
+            [
+                nix,
+                "build",
+                HELPER_BUILD_ATTR,
+                "--no-link",
+            ],
+            cwd=REPO_ROOT,
+            text=True,
+            capture_output=True,
+            check=False,
+        )
+        self.assertEqual(proc.returncode, 0, proc.stdout + proc.stderr)
+
+    # ── Activation script warns on failure rather than silently swallowing ───
+
+    def test_activation_script_emits_warning_on_failure(self):
+        self.assertIn("warning: sovran-hosts-update", self.source)
+
+
+if __name__ == "__main__":
+    unittest.main()

app/tests/test_loopback_diagnostics.py

@@ -0,0 +1,399 @@
+"""Tests for server-local loopback diagnostics and domain validation.
+
+Covers:
+- Domain value validation and injection prevention.
+- Loopback address detection (IPv4 and IPv6).
+- _resolve_all_addresses returning multiple addresses.
+- _check_domain_health_fast with loopback resolution.
+- _evaluate_domain_checklist with loopback override — no false dns_mismatch.
+- _evaluate_domain_checklist with genuine DNS mismatch — still reports error.
+- api_services health stays "healthy" when domain resolves to loopback.
+- api_services health stays "needs_attention" when DNS is genuinely wrong.
+- api_domains_check returns "local_override" for loopback-resolved domains.
+"""
+
+import unittest
+from pathlib import Path
+from unittest.mock import MagicMock, mock_open, patch
+import sys
+import types
+
+sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
+
+
+# ---------------------------------------------------------------------------
+# Minimal stubs so server.py can be imported without the full FastAPI stack.
+# ---------------------------------------------------------------------------
+
+def _install_web_stubs():
+    if "fastapi" in sys.modules:
+        return
+
+    class _HTTPException(Exception):
+        def __init__(self, status_code=None, detail=None):
+            super().__init__(detail)
+            self.status_code = status_code
+            self.detail = detail
+
+    class _FastAPI:
+        def __init__(self, *args, **kwargs):
+            pass
+
+        def mount(self, *args, **kwargs):
+            return None
+
+        def add_middleware(self, *args, **kwargs):
+            return None
+
+        def __getattr__(self, _name):
+            def _decorator_factory(*args, **kwargs):
+                def _decorator(func):
+                    return func
+                return _decorator
+            return _decorator_factory
+
+    class _BaseModel:
+        pass
+
+    class _StaticFiles:
+        def __init__(self, *args, **kwargs):
+            pass
+
+    class _Jinja2Templates:
+        def __init__(self, *args, **kwargs):
+            pass
+
+    class _BaseHTTPMiddleware:
+        pass
+
+    fastapi_module = types.ModuleType("fastapi")
+    fastapi_module.FastAPI = _FastAPI
+    fastapi_module.HTTPException = _HTTPException
+    sys.modules["fastapi"] = fastapi_module
+
+    responses_module = types.ModuleType("fastapi.responses")
+    responses_module.HTMLResponse = object
+    responses_module.JSONResponse = object
+    responses_module.RedirectResponse = object
+    sys.modules["fastapi.responses"] = responses_module
+
+    staticfiles_module = types.ModuleType("fastapi.staticfiles")
+    staticfiles_module.StaticFiles = _StaticFiles
+    sys.modules["fastapi.staticfiles"] = staticfiles_module
+
+    templating_module = types.ModuleType("fastapi.templating")
+    templating_module.Jinja2Templates = _Jinja2Templates
+    sys.modules["fastapi.templating"] = templating_module
+
+    requests_module = types.ModuleType("fastapi.requests")
+    requests_module.Request = object
+    sys.modules["fastapi.requests"] = requests_module
+
+    pydantic_module = types.ModuleType("pydantic")
+    pydantic_module.BaseModel = _BaseModel
+    sys.modules["pydantic"] = pydantic_module
+
+    starlette_base_module = types.ModuleType("starlette.middleware.base")
+    starlette_base_module.BaseHTTPMiddleware = _BaseHTTPMiddleware
+    sys.modules["starlette.middleware.base"] = starlette_base_module
+
+    starlette_middleware_module = types.ModuleType("starlette.middleware")
+    starlette_middleware_module.base = starlette_base_module
+    sys.modules["starlette.middleware"] = starlette_middleware_module
+
+    starlette_module = types.ModuleType("starlette")
+    starlette_module.middleware = starlette_middleware_module
+    sys.modules["starlette"] = starlette_module
+
+
+_install_web_stubs()
+from sovran_systemsos_web import server  # noqa: E402
+
+
+# ===========================================================================
+# Domain value validation
+# ===========================================================================
+
+class TestValidateDomainValue(unittest.TestCase):
+    """_validate_domain_value must reject anything that could corrupt /etc/hosts."""
+
+    def _v(self, value: str) -> bool:
+        return server._validate_domain_value(value)
+
+    # -- Valid values --------------------------------------------------------
+
+    def test_simple_domain_valid(self):
+        self.assertTrue(self._v("cloud.example.com"))
+
+    def test_subdomain_valid(self):
+        self.assertTrue(self._v("matrix.home.example.org"))
+
+    def test_single_label_with_tld_valid(self):
+        self.assertTrue(self._v("example.com"))
+
+    def test_hyphen_in_domain_valid(self):
+        self.assertTrue(self._v("my-nextcloud.example.com"))
+
+    # -- Injection / malformed values ----------------------------------------
+
+    def test_empty_string_invalid(self):
+        self.assertFalse(self._v(""))
+
+    def test_newline_injection_invalid(self):
+        self.assertFalse(self._v("evil.com\n127.0.0.1 other.host"))
+
+    def test_carriage_return_injection_invalid(self):
+        self.assertFalse(self._v("evil.com\r127.0.0.1 other.host"))
+
+    def test_space_injection_invalid(self):
+        self.assertFalse(self._v("evil.com 127.0.0.1"))
+
+    def test_hash_comment_injection_invalid(self):
+        self.assertFalse(self._v("evil.com# comment"))
+
+    def test_bare_hostname_no_dot_invalid(self):
+        self.assertFalse(self._v("localhost"))
+
+    def test_bare_ip_invalid(self):
+        self.assertFalse(self._v("192.168.1.1"))
+
+    def test_too_long_invalid(self):
+        self.assertFalse(self._v("a" * 254 + ".com"))
+
+    def test_leading_dot_invalid(self):
+        self.assertFalse(self._v(".example.com"))
+
+    def test_trailing_dot_invalid(self):
+        self.assertFalse(self._v("example.com."))
+
+
+# ===========================================================================
+# Loopback address detection
+# ===========================================================================
+
+class TestIsLoopbackAddress(unittest.TestCase):
+
+    def test_ipv4_loopback(self):
+        self.assertTrue(server._is_loopback_address("127.0.0.1"))
+
+    def test_ipv4_loopback_other(self):
+        self.assertTrue(server._is_loopback_address("127.0.0.2"))
+
+    def test_ipv4_loopback_high(self):
+        self.assertTrue(server._is_loopback_address("127.255.255.255"))
+
+    def test_ipv6_loopback(self):
+        self.assertTrue(server._is_loopback_address("::1"))
+
+    def test_public_ipv4_not_loopback(self):
+        self.assertFalse(server._is_loopback_address("203.0.113.10"))
+
+    def test_private_ipv4_not_loopback(self):
+        self.assertFalse(server._is_loopback_address("192.168.1.50"))
+
+    def test_ipv6_public_not_loopback(self):
+        self.assertFalse(server._is_loopback_address("2001:db8::1"))
+
+    def test_invalid_string_not_loopback(self):
+        self.assertFalse(server._is_loopback_address("not-an-ip"))
+
+
+# ===========================================================================
+# _check_domain_health_fast
+# ===========================================================================
+
+class TestCheckDomainHealthFast(unittest.TestCase):
+    """_check_domain_health_fast returns True when there is an issue,
+    False when everything looks fine."""
+
+    def _fast(self, domain, external_ip, resolved_addrs):
+        with patch.object(server, "_resolve_all_addresses", return_value=resolved_addrs):
+            return server._check_domain_health_fast(domain, external_ip)
+
+    def test_no_domain_no_issue(self):
+        # None/empty domain: the fast check reports True (handled by checklist).
+        result = server._check_domain_health_fast(None, "203.0.113.10")
+        self.assertTrue(result)
+
+    def test_empty_domain_no_issue(self):
+        result = server._check_domain_health_fast("", "203.0.113.10")
+        self.assertTrue(result)
+
+    def test_loopback_ipv4_no_issue(self):
+        """Loopback override must not be flagged as a DNS mismatch."""
+        result = self._fast("cloud.example.com", "203.0.113.10", ["127.0.0.1"])
+        self.assertFalse(result)
+
+    def test_loopback_ipv6_no_issue(self):
+        result = self._fast("cloud.example.com", "203.0.113.10", ["::1"])
+        self.assertFalse(result)
+
+    def test_matches_external_ip_no_issue(self):
+        result = self._fast("cloud.example.com", "203.0.113.10", ["203.0.113.10"])
+        self.assertFalse(result)
+
+    def test_mismatch_is_an_issue(self):
+        result = self._fast("cloud.example.com", "203.0.113.10", ["198.51.100.1"])
+        self.assertTrue(result)
+
+    def test_unavailable_external_ip_no_issue(self):
+        result = self._fast("cloud.example.com", "unavailable", ["198.51.100.1"])
+        self.assertFalse(result)
+
+    def test_multiple_addresses_one_matches_no_issue(self):
+        """If any resolved address matches external_ip the check should pass."""
+        result = self._fast(
+            "cloud.example.com", "203.0.113.10",
+            ["198.51.100.1", "203.0.113.10"],
+        )
+        self.assertFalse(result)
+
+
+# ===========================================================================
+# _evaluate_domain_checklist — loopback override path
+# ===========================================================================
+
+class TestEvaluateDomainChecklistLoopback(unittest.TestCase):
+
+    def _eval(self, domain, external_ip, resolved_addrs, reachable_result=None):
+        with (
+            patch.object(server, "_resolve_all_addresses", return_value=resolved_addrs),
+            patch.object(server, "_check_domain_reachable",
+                         return_value=reachable_result or {"reachable": True, "status_code": 200}),
+        ):
+            return server._evaluate_domain_checklist(domain, external_ip)
+
+    def test_loopback_dns_step_is_ok_not_error(self):
+        result = self._eval("cloud.example.com", "203.0.113.10", ["127.0.0.1"])
+        dns_step = next(s for s in result["domain_check_steps"] if s["step"] == 2)
+        self.assertEqual(dns_step["status"], "ok")
+        self.assertNotIn("mismatch", dns_step.get("detail", "").lower())
+
+    def test_loopback_domain_status_is_local_override(self):
+        result = self._eval("cloud.example.com", "203.0.113.10", ["127.0.0.1"])
+        self.assertEqual(result["domain_status"]["status"], "local_override")
+
+    def test_loopback_has_no_issues_when_reachable(self):
+        result = self._eval(
+            "cloud.example.com", "203.0.113.10", ["127.0.0.1"],
+            reachable_result={"reachable": True, "status_code": 200},
+        )
+        self.assertFalse(result["has_issues"])
+
+    def test_loopback_has_issues_when_caddy_unreachable(self):
+        """A loopback override with Caddy down should still report an issue."""
+        result = self._eval(
+            "cloud.example.com", "203.0.113.10", ["127.0.0.1"],
+            reachable_result={"reachable": False, "error": "connection refused"},
+        )
+        self.assertTrue(result["has_issues"])
+
+    def test_ipv6_loopback_no_issue(self):
+        result = self._eval("cloud.example.com", "203.0.113.10", ["::1"])
+        self.assertEqual(result["domain_status"]["status"], "local_override")
+        self.assertFalse(result["has_issues"])
+
+    def test_genuine_mismatch_still_reports_error(self):
+        result = self._eval("cloud.example.com", "203.0.113.10", ["198.51.100.1"])
+        self.assertEqual(result["domain_status"]["status"], "dns_mismatch")
+        self.assertTrue(result["has_issues"])
+
+    def test_correct_public_dns_still_reports_ok(self):
+        result = self._eval("cloud.example.com", "203.0.113.10", ["203.0.113.10"])
+        self.assertEqual(result["domain_status"]["status"], "connected")
+        self.assertFalse(result["has_issues"])
+
+    def test_no_domain_has_issues(self):
+        result = self._eval(None, "203.0.113.10", [])
+        self.assertTrue(result["has_issues"])
+
+
+# ===========================================================================
+# api_services — composite health with loopback
+# ===========================================================================
+
+class TestApiServicesLoopbackHealth(unittest.IsolatedAsyncioTestCase):
+
+    async def _get_health(self, resolved_addrs, cached_reachable):
+        """Return the health value for a single domain-requiring service."""
+        service_cfg = {
+            "services": [
+                {"unit": "caddy.service", "icon": "nextcloud", "enabled": True, "type": "system"}
+            ]
+        }
+        with (
+            patch.object(server, "load_config", return_value=service_cfg),
+            patch.object(server, "_read_hub_overrides", return_value=({}, None, None)),
+            patch.object(server.sysctl, "is_active", return_value="active"),
+            patch.dict(server.SERVICE_DOMAIN_MAP, {"caddy.service": "nextcloud"}, clear=False),
+            patch("builtins.open", mock_open(read_data="cloud.example.com\n")),
+            patch.object(server, "_resolve_all_addresses", return_value=resolved_addrs),
+            patch.object(server, "_is_domain_reachable_cached", return_value=cached_reachable),
+            patch.object(server, "_get_listening_ports",
+                         return_value={"tcp": {80, 443}, "udp": set()}),
+            patch.object(server, "_get_firewall_allowed_ports",
+                         return_value={"tcp": set(), "udp": set()}),
+            patch.object(server, "_cached_external_ip", "203.0.113.10"),
+        ):
+            results = await server.api_services()
+
+        return results[0]["health"]
+
+    async def test_loopback_and_reachable_is_healthy(self):
+        """Loopback override + Caddy reachable → healthy, not needs_attention."""
+        health = await self._get_health(["127.0.0.1"], cached_reachable=True)
+        self.assertEqual(health, "healthy")
+
+    async def test_loopback_and_caddy_down_is_needs_attention(self):
+        """Loopback override + Caddy unreachable → needs_attention (genuine issue)."""
+        health = await self._get_health(["127.0.0.1"], cached_reachable=False)
+        self.assertEqual(health, "needs_attention")
+
+    async def test_correct_dns_and_reachable_is_healthy(self):
+        health = await self._get_health(["203.0.113.10"], cached_reachable=True)
+        self.assertEqual(health, "healthy")
+
+    async def test_dns_mismatch_is_needs_attention(self):
+        health = await self._get_health(["198.51.100.1"], cached_reachable=True)
+        self.assertEqual(health, "needs_attention")
+
+
+# ===========================================================================
+# api_domains_check — loopback detection
+# ===========================================================================
+
+class TestApiDomainsCheckLoopback(unittest.IsolatedAsyncioTestCase):
+
+    async def _check(self, resolved_addrs, external_ip="203.0.113.10"):
+        with (
+            patch.object(server, "_resolve_all_addresses", return_value=resolved_addrs),
+            patch.object(server, "_cached_external_ip", external_ip),
+        ):
+            result = await server.api_domains_check(
+                MagicMock(domains=["cloud.example.com"])
+            )
+        return result["domains"][0]
+
+    async def test_loopback_ipv4_returns_local_override(self):
+        result = await self._check(["127.0.0.1"])
+        self.assertEqual(result["status"], "local_override")
+
+    async def test_loopback_ipv6_returns_local_override(self):
+        result = await self._check(["::1"])
+        self.assertEqual(result["status"], "local_override")
+
+    async def test_correct_dns_returns_connected(self):
+        result = await self._check(["203.0.113.10"])
+        self.assertEqual(result["status"], "connected")
+
+    async def test_mismatch_returns_dns_mismatch(self):
+        result = await self._check(["198.51.100.1"])
+        self.assertEqual(result["status"], "dns_mismatch")
+
+    async def test_no_resolution_returns_unresolvable(self):
+        result = await self._check([])
+        self.assertEqual(result["status"], "unresolvable")
+
+
+if __name__ == "__main__":
+    unittest.main()

app/tests/test_manual_backup_workflow.py

@@ -0,0 +1,41 @@
+import unittest
+from pathlib import Path
+
+
+REPO_ROOT = Path(__file__).resolve().parents[2]
+BACKUP_SCRIPT = REPO_ROOT / "app" / "sovran_systemsos_web" / "scripts" / "sovran-hub-backup.sh"
+SERVER_FILE = REPO_ROOT / "app" / "sovran_systemsos_web" / "server.py"
+SUPPORT_JS = REPO_ROOT / "app" / "sovran_systemsos_web" / "static" / "js" / "support.js"
+
+
+class ManualBackupWorkflowTests(unittest.TestCase):
+    def test_backup_script_uses_tar_archives_with_checksums_and_exclusions(self):
+        source = BACKUP_SCRIPT.read_text()
+
+        self.assertIn("tar \\", source)
+        self.assertIn("--create", source)
+        self.assertIn("--one-file-system", source)
+        self.assertIn("sha256sum", source)
+        self.assertIn("export_postgresql_dumps", source)
+        self.assertIn("export_mariadb_dumps", source)
+        self.assertIn("export_lnd_scb_if_possible", source)
+        self.assertIn("--exclude='var/lib/bitcoind'", source)
+        self.assertIn("--exclude='var/lib/electrs'", source)
+        self.assertIn("--exclude='var/lib/lnd'", source)
+        self.assertIn("set_status \"RUNNING\"", source)
+        self.assertIn("set_status \"SUCCESS\"", source)
+        self.assertIn("set_status \"FAILED\"", source)
+        self.assertNotIn("rsync -a", source)
+
+    def test_backend_and_frontend_use_explicit_backup_terminal_states(self):
+        server_source = SERVER_FILE.read_text()
+        support_source = SUPPORT_JS.read_text()
+
+        self.assertIn("_write_backup_status, \"RUNNING\"", server_source)
+        self.assertIn("_monitor_backup_subprocess", server_source)
+        self.assertIn("asyncio.create_task(_monitor_backup_subprocess(proc))", server_source)
+        self.assertIn("result === \"success\" || result === \"failed\"", support_source)
+
+
+if __name__ == "__main__":
+    unittest.main()

app/tests/test_rdp_module_boot_setup.py

@@ -0,0 +1,122 @@
+import unittest
+from pathlib import Path
+
+
+RDP_NIX = Path(__file__).resolve().parents[2] / "modules" / "rdp.nix"
+USERNAME_READ = "USERNAME=\"$(tr -d '\\n' < \"$USERNAME_FILE\")\""
+USERNAME_LENGTH_GUARD = "if [ \"''${#USERNAME}\" -gt 32 ]; then"
+SHORT_PASSWORD_GUARD = 'if [ "\'\'${#PASSWORD}" -lt 8 ]; then'
+
+
+def _section(source: str, start: str, end: str) -> str:
+    start_idx = source.find(start)
+    if start_idx == -1:
+        raise AssertionError(f"Expected section start not found: {start!r}")
+    end_idx = source.find(end, start_idx)
+    if end_idx == -1:
+        raise AssertionError(f"Expected section end not found: {end!r}")
+    return source[start_idx:end_idx]
+
+
+class RdpModuleBootSetupTests(unittest.TestCase):
+    def setUp(self):
+        self.source = RDP_NIX.read_text()
+        self.gnome_service = _section(
+            self.source,
+            "systemd.services.gnome-remote-desktop = {",
+            "systemd.tmpfiles.rules = [",
+        )
+        self.setup_service = _section(
+            self.source,
+            "systemd.services.gnome-remote-desktop-setup = {",
+            "};\n}",
+        )
+
+    def test_does_not_redeclare_gnome_remote_desktop_user(self):
+        self.assertNotIn("users.users.gnome-remote-desktop", self.source)
+        self.assertNotIn("createHome = true;", self.source)
+
+    def test_main_service_requires_setup_before_starting(self):
+        self.assertIn('wantedBy = [ "graphical.target" ];', self.gnome_service)
+        self.assertIn('after = [ "gnome-remote-desktop-setup.service" ];', self.gnome_service)
+        self.assertIn('requires = [ "gnome-remote-desktop-setup.service" ];', self.gnome_service)
+
+    def test_setup_waits_for_configuration_service_and_bounded_timeout(self):
+        self.assertIn('wantedBy = [ "graphical.target" ];', self.setup_service)
+        self.assertIn('before = [ "gnome-remote-desktop.service" ];', self.setup_service)
+        self.assertIn('"dbus.service"', self.setup_service)
+        self.assertIn('"gnome-remote-desktop-configuration.service"', self.setup_service)
+        self.assertNotIn("RemainAfterExit", self.setup_service)
+        self.assertIn('TimeoutStartSec = "2min";', self.setup_service)
+        self.assertIn('timeout --kill-after=5s 10s', self.setup_service)
+        self.assertIn('echo "grdctl command timed out: $*" >&2', self.setup_service)
+        self.assertIn('echo "grdctl command failed (exit $rc): $*" >&2', self.setup_service)
+
+    def test_setup_runs_grdctl_directly_as_root(self):
+        # The oneshot service runs as root; grdctl --system is called directly.
+        # GRD 50.x invokes pkexec internally, but the call itself is plain
+        # grdctl --system, not a manual pkexec invocation.
+        self.assertIn('grdctl --system "$@"', self.setup_service)
+        self.assertNotIn("runuser", self.setup_service)
+        self.assertNotIn("sudo", self.setup_service)
+        # No direct Nix-store pkexec invocation (pkgs.polkit}/bin/pkexec).
+        self.assertNotIn("pkgs.polkit}/bin/pkexec", self.setup_service)
+
+    def test_privilege_escalation_packages_absent_from_setup_path(self):
+        self.assertNotIn("pkgs.polkit", self.setup_service)
+        self.assertNotIn("pkgs.util-linux", self.setup_service)
+
+    def test_run_wrappers_bin_prepended_to_path(self):
+        # /run/wrappers/bin must be prepended to PATH before any grdctl_system
+        # invocation so that grdctl --system resolves the NixOS setuid pkexec.
+        path_export = 'export PATH="/run/wrappers/bin:$PATH"'
+        grdctl_marker = "grdctl_system"
+        script = self.setup_service
+        path_idx = script.find(path_export)
+        grdctl_idx = script.find(grdctl_marker)
+        self.assertGreater(path_idx, -1, f"{path_export!r} not found in setup script")
+        self.assertGreater(
+            grdctl_idx, path_idx,
+            "PATH export must appear before the first grdctl_system usage",
+        )
+
+    def test_pkexec_preflight_check(self):
+        # A preflight must confirm /run/wrappers/bin/pkexec is executable
+        # with a clear error message before any GRD configuration changes.
+        self.assertIn("test -x /run/wrappers/bin/pkexec", self.setup_service)
+        self.assertIn(
+            "/run/wrappers/bin/pkexec is absent or not executable",
+            self.setup_service,
+        )
+
+    def test_hub_files_are_the_source_of_truth_for_username_and_password(self):
+        self.assertIn('DEFAULT_USERNAME="sovran"', self.setup_service)
+        self.assertIn('if [ ! -f "$USERNAME_FILE" ]; then', self.setup_service)
+        self.assertIn(USERNAME_READ, self.setup_service)
+        self.assertIn(USERNAME_LENGTH_GUARD, self.setup_service)
+        self.assertIn('case "$USERNAME" in', self.setup_service)
+        self.assertIn('[A-Za-z_][A-Za-z0-9_-]*)', self.setup_service)
+        self.assertIn("RDP username is too long (''${#USERNAME} characters, maximum 32)", self.setup_service)
+        self.assertIn("RDP username must start with a letter or underscore and contain only letters, numbers, underscores, and hyphens", self.setup_service)
+        self.assertIn('if [ ! -f "$PASSWORD_FILE" ]; then', self.setup_service)
+        self.assertIn("tr -d '\\n'", self.setup_service)
+        self.assertIn('"$PASSWORD_FILE"', self.setup_service)
+        self.assertIn(SHORT_PASSWORD_GUARD, self.setup_service)
+        self.assertIn("RDP password is too short (''${#PASSWORD} characters, minimum 8)", self.setup_service)
+        self.assertIn('grdctl_system rdp set-credentials "$USERNAME" "$PASSWORD"', self.setup_service)
+        self.assertNotIn('grdctl --system rdp set-credentials sovran "$PASSWORD"', self.setup_service)
+
+    def test_secure_permissions_are_enforced_for_state_and_secret_files(self):
+        self.assertIn('"d /var/lib/gnome-remote-desktop/tls 0700', self.source)
+        self.assertIn("chmod 700", self.setup_service)
+        self.assertIn('chmod 600 "$USERNAME_FILE"', self.setup_service)
+        self.assertIn('chmod 600 "$PASSWORD_FILE"', self.setup_service)
+        self.assertIn('chmod 600 "$CRED_FILE"', self.setup_service)
+        self.assertIn('chmod 600 "$TLS_DIR/rdp-tls.key"', self.setup_service)
+        self.assertIn('chmod 644 "$TLS_DIR/rdp-tls.crt"', self.setup_service)
+        self.assertIn('LOCAL_IP="$(hostname -I | awk \'{print $1}\')"', self.setup_service)
+        self.assertIn('LOCAL_IP="127.0.0.1"', self.setup_service)
+
+
+if __name__ == "__main__":
+    unittest.main()

flake.lock

@@ -5,11 +5,11 @@
         "nixpkgs": "nixpkgs"
       },
       "locked": {
-        "lastModified": 1783086783,
-        "narHash": "sha256-NxXpNF/9tq2nI+SxFHUxjro3u11SF3l4vs7bawdMKkQ=",
+        "lastModified": 1783519926,
+        "narHash": "sha256-2zwAN4lNitHFrHVnRZG3YcvpdtWOoF0cOBstxMeB1KI=",
         "owner": "emmanuelrosa",
         "repo": "btc-clients-nix",
-        "rev": "4f6d07cae877ef58f0fbc9e731c99800ddb80859",
+        "rev": "731a1e11c2fefb14f0aa4b1f03cfa85c19c28d71",
         "type": "github"
       },
       "original": {
@@ -139,11 +139,11 @@
     },
     "nixpkgs-stable": {
       "locked": {
-        "lastModified": 1782999065,
-        "narHash": "sha256-5Dgj5+pIQYZKrXUGaLCk7CKfN3MmpwIhO94++WVxvng=",
+        "lastModified": 1783856661,
+        "narHash": "sha256-ZGP04e+Q6WyQJGA9ZvI5CL6+heGQldbAG9U1T9NGvmU=",
         "owner": "nixos",
         "repo": "nixpkgs",
-        "rev": "80d591ed473cfc46329932c2aadac9b435342c7c",
+        "rev": "569d578509928497eddc3fdbf94a799027050be4",
         "type": "github"
       },
       "original": {
@@ -187,11 +187,11 @@
     },
     "nixpkgs_3": {
       "locked": {
-        "lastModified": 1782959384,
-        "narHash": "sha256-xnJJk+ct+D2+wdRxj1wk36w5zV9RVESwRqcklPdt3fM=",
+        "lastModified": 1783776592,
+        "narHash": "sha256-UgCQzxeWI75XM8G+hPrPh+MKzEPjG3SpAj7dtqSbksA=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "65179426c83bb3f6bc14898b42ea1c6f01d374b0",
+        "rev": "e7a3ca8092b61ff85b6a45bf863ea2b2d6a661b3",
         "type": "github"
       },
       "original": {
@@ -203,11 +203,11 @@
     },
     "nixpkgs_4": {
       "locked": {
-        "lastModified": 1782948114,
-        "narHash": "sha256-AXmz9ho4Lud5CsbrZsuSVwpQZ4o5FgZ1chxBn5cJ8+0=",
+        "lastModified": 1783791668,
+        "narHash": "sha256-zbcZ1dmBTPfJ7Mlqh/yLEPGpgJnwuv4Xr1xucy2WqMA=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "9e92285f211dad236540fd617d7e30e0b99bc0e1",
+        "rev": "716c7a2664ca8325617b8a7fbb609273f2c4cae7",
         "type": "github"
       },
       "original": {
@@ -224,11 +224,11 @@
         "systems": "systems_2"
       },
       "locked": {
-        "lastModified": 1783173302,
-        "narHash": "sha256-nlnOw/zsD2H2NHSZ5oNWwcjuM17vipyAapfXsO78GjY=",
+        "lastModified": 1783941741,
+        "narHash": "sha256-F+3M1IZrJa920cx2/k2AMKqedEodxLF7COJVkLJwUBo=",
         "owner": "nix-community",
         "repo": "nixvim",
-        "rev": "a402fdf2a1ef297d8ea7c95b90d6af0dbe90ab11",
+        "rev": "e6715f01d9f56f07a27a01386b85ae22b06f0705",
         "type": "github"
       },
       "original": {

iso/installer.py

@@ -1169,7 +1169,7 @@ class InstallerWindow(Adw.ApplicationWindow):
         btn_box = Gtk.Box(orientation=Gtk.Orientation.HORIZONTAL, spacing=0)
         btn_box.set_halign(Gtk.Align.CENTER)
         btn_box.set_margin_bottom(32)
-        reboot_btn = Gtk.Button(label="I Have Written Down My Password — Reboot Now")
+        reboot_btn = Gtk.Button(label="I Have Written Down My Password — Restart Entire System")
         reboot_btn.add_css_class("suggested-action")
         reboot_btn.add_css_class("pill")
         reboot_btn.connect("clicked", lambda b: subprocess.run(["sudo", "reboot"]))

modules/core/local-domain-loopback.nix

@@ -0,0 +1,171 @@
+{ config, pkgs, lib, ... }:
+
+# ── Server-local domain loopback overrides ────────────────────────────────────
+#
+# Some routers (especially newer ISP-provided devices) do not support NAT
+# loopback (hairpin NAT).  When a request originates on this computer and
+# targets a public domain name that resolves to the router's WAN address, the
+# router may refuse to loop the connection back in — causing Nextcloud, WordPress
+# background jobs, and other server-side callbacks to fail even when the service
+# is fully operational from the internet.
+#
+# This module installs a one-shot systemd service,
+# ``sovran-hosts-update.service``, that reads the configured service domains
+# from ``/var/lib/domains/`` at boot (and whenever triggered by the Hub after a
+# domain is saved) and writes ``127.0.0.1`` entries for them into a dedicated
+# Sovran-managed block in ``/etc/hosts``.
+#
+# With those entries in place:
+#   • Requests originating on this computer resolve the public domain name to
+#     127.0.0.1, reach Caddy directly, and never touch the router.
+#   • Caddy still receives the correct public hostname via TLS SNI so virtual-
+#     host routing and certificate validation continue to work.
+#   • The Sovran Hub can verify Caddy reachability locally without needing NAT
+#     loopback.
+#
+# Limitation: this does not help other devices on your home network (phones,
+# laptops).  Those devices resolve domains via the router's DNS and still depend
+# on NAT loopback (or require manual router DNS overrides).  For now, only
+# server-originated requests benefit from this override.
+#
+# On NixOS, /etc/hosts is normally a symlink into the Nix store and is
+# regenerated by the system activation script.  The ``system.activationScripts``
+# hook below converts it to a writable file each time the system is activated
+# (i.e. after every ``nixos-rebuild switch``) and then injects the Sovran block.
+# The same wrapped Nix-store executable is reused by both the activation hook
+# and the ``sovran-hosts-update.service`` unit, ensuring a deterministic runtime
+# PATH in every execution context.
+
+let
+  # ── Wrapped Nix-store executable ──────────────────────────────────────────
+  # Built with pkgs.writeShellApplication so that all required runtime tools
+  # (awk, grep, coreutils) are declared explicitly and injected into PATH by
+  # Nix.  Both the systemd service and the activation hook reference this same
+  # store-path executable — there is no second raw script body.
+  hostsUpdateScript = pkgs.writeShellApplication {
+    name = "sovran-hosts-update";
+
+    # Declare every external command the script calls.  These packages are
+    # added to the script's runtime PATH by writeShellApplication; nothing from
+    # the system PATH is relied upon.
+    runtimeInputs = [
+      pkgs.coreutils   # readlink, cp, mv, chmod, mktemp, rm, tr, head
+      pkgs.gawk        # awk
+      pkgs.gnugrep     # grep
+    ];
+
+    text = ''
+      # Regenerate the Sovran-managed loopback block in /etc/hosts.
+      # Safe to run multiple times — idempotent.
+
+      DOMAINS_DIR="/var/lib/domains"
+      HOSTS_FILE="/etc/hosts"
+      BEGIN_MARKER="# Sovran managed begin — server-local loopback overrides"
+      END_MARKER="# Sovran managed end"
+
+      # ── Step 1: ensure /etc/hosts is a regular writable file ──────────────
+      # On NixOS /etc/hosts starts as a symlink to the Nix store.  We replace
+      # it with a copy so we can append our block without touching the store.
+      if [ -L "$HOSTS_FILE" ]; then
+        TARGET=$(readlink -f "$HOSTS_FILE")
+        cp --no-preserve=all "$TARGET" "$HOSTS_FILE.sovran-tmp"
+        mv "$HOSTS_FILE.sovran-tmp" "$HOSTS_FILE"
+        chmod 644 "$HOSTS_FILE"
+      fi
+
+      # ── Step 2: remove any existing Sovran block ──────────────────────────
+      # Use a temp file so the operation is atomic.
+      # awk -v passes marker strings safely without shell interpolation.
+      TMP=$(mktemp "$HOSTS_FILE.XXXXXX")
+      trap 'rm -f "$TMP"' EXIT
+      awk -v begin="$BEGIN_MARKER" -v end="$END_MARKER" '
+        $0 == begin { skip=1; next }
+        $0 == end   { skip=0; next }
+        !skip
+      ' "$HOSTS_FILE" > "$TMP"
+
+      # ── Step 3: collect valid configured service domains ──────────────────
+      # NOTE: The hostname validation regex below must stay in sync with
+      # _SAFE_DOMAIN_RE in app/sovran_systemsos_web/server.py.
+      ENTRIES=""
+      for KEY in matrix wordpress nextcloud btcpayserver vaultwarden haven element-calling; do
+        FILE="$DOMAINS_DIR/$KEY"
+        [ -f "$FILE" ] || continue
+        # Read the domain value (strip all whitespace, limit to 253 chars)
+        DOMAIN=$(tr -d '[:space:]' < "$FILE" | head -c 253)
+        [ -z "$DOMAIN" ] && continue
+        # Validate: must match a reasonable hostname pattern (no injection)
+        if ! printf '%s' "$DOMAIN" | grep -qE \
+          '^[a-zA-Z0-9]([a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(\.[a-zA-Z0-9]([a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)+$'; then
+          echo "sovran-hosts-update: skipping invalid domain value for $KEY: $DOMAIN" >&2
+          continue
+        fi
+        ENTRIES="$ENTRIES
+127.0.0.1 $DOMAIN
+::1       $DOMAIN"
+      done
+
+      # ── Step 4: append the Sovran block if there are any entries ──────────
+      if [ -n "$ENTRIES" ]; then
+        {
+          printf '\n%s\n' "$BEGIN_MARKER"
+          printf '%s\n' "# These entries route configured service domains to local Caddy."
+          printf '%s\n' "# They are managed automatically — do not edit this block."
+          printf '%s\n' "$ENTRIES"
+          printf '%s\n' "$END_MARKER"
+        } >> "$TMP"
+      fi
+
+      # ── Step 5: atomically replace /etc/hosts ─────────────────────────────
+      mv "$TMP" "$HOSTS_FILE"
+      chmod 644 "$HOSTS_FILE"
+    '';
+  };
+
+in
+{
+  # ── /etc/sovran-hosts-update.sh — operator discoverability symlink ─────────
+  # Retain the familiar /etc path so administrators can inspect or manually
+  # invoke the helper.  The target is the wrapped Nix-store executable, so
+  # there is no second raw script body to keep in sync.
+  environment.etc."sovran-hosts-update.sh".source = lib.getExe hostsUpdateScript;
+
+  # ── Systemd service ────────────────────────────────────────────────────────
+
+  systemd.services.sovran-hosts-update = {
+    description = "Update /etc/hosts with Sovran server-local loopback overrides";
+    documentation = [ "https://github.com/naturallaw777/sovran-systems" ];
+
+    # Run before Caddy so loopback entries are ready when it starts.
+    before = [
+      "caddy.service"
+      "network-online.target"
+    ];
+    wantedBy = [ "multi-user.target" ];
+
+    serviceConfig = {
+      Type = "oneshot";
+      RemainAfterExit = true;
+      # Point directly at the wrapped Nix-store executable, not the /etc path.
+      ExecStart = lib.getExe hostsUpdateScript;
+    };
+  };
+
+  # ── Activation script (runs after every nixos-rebuild switch) ─────────────
+  # This ensures the loopback block survives rebuilds that restore the /etc/hosts
+  # symlink.  The "users" and "etc" scripts must complete first.
+  # The same wrapped Nix-store executable used by the systemd service is
+  # referenced here, guaranteeing identical runtime dependencies in both
+  # execution contexts.
+
+  system.activationScripts.sovranDomainLoopback = {
+    text = ''
+      if [ -d /var/lib/domains ]; then
+        if ! ${lib.getExe hostsUpdateScript}; then
+          echo "warning: sovran-hosts-update: failed to update /etc/hosts loopback entries" >&2
+        fi
+      fi
+    '';
+    deps = [ "etc" "users" ];
+  };
+}

modules/core/sovran-hub.nix

@@ -388,7 +388,16 @@ in
         pkgs.nftables
         pkgs.iptables
         pkgs.hostname
-      ] ++ lib.optional cfg.services.bitcoin config.services.bitcoind.package;
+        pkgs.coreutils
+        pkgs.findutils
+        pkgs.gnugrep
+        pkgs.gnutar
+        pkgs.util-linux
+      ]
+      ++ lib.optional cfg.services.bitcoin config.services.bitcoind.package
+      ++ lib.optionals cfg.services.bitcoin [ pkgs.lnd ]
+      ++ lib.optionals (cfg.services.nextcloud || cfg.services.synapse) [ config.services.postgresql.package ]
+      ++ lib.optionals config.services.mysql.enable [ config.services.mysql.package ];
     };
 
     systemd.services.sovran-hub-update = {

modules/element-calling.nix

@@ -72,17 +72,30 @@ $MATRIX {
 }
 
 $ELEMENT_CALLING {
-  handle /livekit/jwt/sfu/get {
+  # Route all current lk-jwt-service authorization endpoints to port 8073,
+  # stripping the /livekit/jwt prefix that Caddy adds on the public URL.
+  @lk_jwt path /livekit/jwt/sfu/get* /livekit/jwt/get_token* /livekit/jwt/healthz* /livekit/jwt/sfu_webhook* /livekit/jwt/delegate_delayed_leave*
+  handle @lk_jwt {
     uri strip_prefix /livekit/jwt
     reverse_proxy [::1]:8073 {
       header_up Host {host}
       header_up X-Forwarded-Server {host}
       header_up X-Real-IP {remote_host}
       header_up X-Forwarded-For {remote_host}
+      header_up X-Forwarded-Proto {scheme}
     }
   }
   handle {
-    reverse_proxy localhost:7880
+    reverse_proxy localhost:7880 {
+      header_up Host {host}
+      header_up X-Forwarded-Proto {scheme}
+      header_up X-Forwarded-For {remote_host}
+      header_up X-Real-IP {remote_host}
+      transport http {
+        read_timeout 300s
+        write_timeout 300s
+      }
+    }
   }
 }
 EOF
@@ -94,8 +107,11 @@ EOF
   #   * reads the matrix domain from /var/lib/domains/matrix (never hardcoded)
   #   * copies Caddy's already-issued matrix cert/key into /var/lib/livekit
   #     so LoadCredential can stage them for the (DynamicUser) livekit unit
-  #   * writes a complete LiveKit config (with turn.domain substituted) that the
-  #     overridden ExecStart loads.
+  #   * detects the primary network interface from the IPv4 default route so
+  #     LiveKit only advertises real ICE candidates — not VPN/container/private
+  #     addresses from interfaces like Tailscale or Docker bridges
+  #   * writes a complete LiveKit config (with turn.domain and interface
+  #     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" ];
@@ -109,7 +125,7 @@ EOF
     unitConfig = {
       ConditionPathExists = "/var/lib/domains/element-calling";
     };
-    path = [ pkgs.coreutils pkgs.findutils ];
+    path = [ pkgs.coreutils pkgs.findutils pkgs.iproute2 pkgs.gawk ];
     script = ''
       MATRIX=$(cat /var/lib/domains/matrix)
 
@@ -123,16 +139,37 @@ EOF
       cp "$KEY" /var/lib/livekit/turn.key
       chmod 640 /var/lib/livekit/turn.crt /var/lib/livekit/turn.key
 
-      # Generate the full LiveKit config the daemon will load. turn.domain is
-      # only known at runtime, so it is substituted here. The cert/key paths
-      # point at the LoadCredential-staged copies under /run/credentials.
+      # Detect the primary network interface from the IPv4 default route.
+      # Restricting LiveKit to this single interface prevents it from
+      # advertising VPN/container/private ICE candidates (e.g. Tailscale,
+      # Docker bridges) that remote peers cannot reach, which causes all
+      # ICE negotiation attempts to fail with responsesReceived: 0.
+      IFACE=$(ip -4 route show default | awk '/^default/ { for(i=1;i<=NF;i++) if($i=="dev" && (i+1)<=NF) { print $(i+1); exit } }')
+      if [ -z "$IFACE" ]; then
+        echo "ERROR: Could not detect a default-route network interface from 'ip -4 route show default'." >&2
+        echo "ERROR: Cannot generate a valid LiveKit config without a real interface to bind ICE candidates to." >&2
+        echo "ERROR: Ensure a default IPv4 route is configured, e.g.: ip route add default via <gateway> dev <interface>" >&2
+        echo "ERROR: Inspect the current routing table with: ip -4 route show" >&2
+        exit 1
+      fi
+      echo "Detected primary network interface: $IFACE"
+
+      # Generate the full LiveKit config the daemon will load. turn.domain and
+      # 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.
       cat > /run/livekit/livekit.yaml <<EOF
 port: 7880
 rtc:
   use_external_ip: true
+  skip_external_ip_validation: true
+  tcp_port: 7881
   udp_port: 7882
   port_range_start: 30000
   port_range_end: 40000
+  interfaces:
+    includes:
+      - $IFACE
 room:
   auto_create: false
 turn:
@@ -155,6 +192,8 @@ EOF
     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;
@@ -186,6 +225,9 @@ EOF
 
   networking.firewall.allowedTCPPorts = [ 5349 7881 ];
   networking.firewall.allowedUDPPorts = [ 3478 7882 ];
+  networking.firewall.allowedUDPPortRanges = [
+    { from = 30000; to = 40000; } # LiveKit internal TURN relay range
+  ];
 
   ####### JWT SERVICE RUNTIME CONFIG #######
   systemd.services.lk-jwt-service-runtime-config = {
@@ -204,11 +246,13 @@ EOF
     path = [ pkgs.coreutils ];
     script = ''
       ELEMENT_CALLING=$(cat /var/lib/domains/element-calling)
+      MATRIX=$(cat /var/lib/domains/matrix)
 
       mkdir -p /run/lk-jwt-service
 
       cat > /run/lk-jwt-service/env <<EOF
 LIVEKIT_URL=wss://$ELEMENT_CALLING
+LIVEKIT_FULL_ACCESS_HOMESERVERS=$MATRIX
 EOF
 
       chmod 640 /run/lk-jwt-service/env

modules/modules.nix

@@ -16,6 +16,7 @@
     ./core/remote-deploy.nix
     ./core/no-sleep.nix
     ./core/cpu-performance.nix
+    ./core/local-domain-loopback.nix
 
     # ── Always on (no flag) ───────────────────────────────────
     ./php.nix

modules/rdp.nix

@@ -2,70 +2,104 @@
 
 lib.mkIf config.sovran_systemsOS.features.rdp {
 
-  users.users.gnome-remote-desktop = {
-    isSystemUser = true;
-    group = "gnome-remote-desktop";
-    home = "/var/lib/gnome-remote-desktop";
-    createHome = true;
-  };
-  users.groups.gnome-remote-desktop = {};
-
   # Enable the GNOME Remote Desktop service at the system level
   services.gnome.gnome-remote-desktop.enable = true;
 
   # Open RDP port in the firewall
   networking.firewall.allowedTCPPorts = [ 3389 ];
 
-  # Ensure the service actually starts and waits for setup to complete
+  # Ensure the service only starts after setup succeeds
   systemd.services.gnome-remote-desktop = {
     wantedBy = [ "graphical.target" ];
     after = [ "gnome-remote-desktop-setup.service" ];
-    wants = [ "gnome-remote-desktop-setup.service" ];
+    requires = [ "gnome-remote-desktop-setup.service" ];
   };
 
   systemd.tmpfiles.rules = [
-    "d /var/lib/gnome-remote-desktop 0750 gnome-remote-desktop gnome-remote-desktop -"
-    "d /var/lib/gnome-remote-desktop/.local 0750 gnome-remote-desktop gnome-remote-desktop -"
-    "d /var/lib/gnome-remote-desktop/.local/share 0750 gnome-remote-desktop gnome-remote-desktop -"
-    "d /var/lib/gnome-remote-desktop/.local/share/gnome-remote-desktop 0750 gnome-remote-desktop gnome-remote-desktop -"
+    "d /var/lib/gnome-remote-desktop/.local 0700 gnome-remote-desktop gnome-remote-desktop -"
+    "d /var/lib/gnome-remote-desktop/.local/share 0700 gnome-remote-desktop gnome-remote-desktop -"
+    "d /var/lib/gnome-remote-desktop/.local/share/gnome-remote-desktop 0700 gnome-remote-desktop gnome-remote-desktop -"
+    "d /var/lib/gnome-remote-desktop/tls 0700 gnome-remote-desktop gnome-remote-desktop -"
   ];
 
   systemd.services.gnome-remote-desktop-setup = {
     description = "Configure GNOME Remote Desktop RDP";
-    wantedBy = [ "multi-user.target" ];
+    wantedBy = [ "graphical.target" ];
     before = [ "gnome-remote-desktop.service" ];
-    after = [ "systemd-tmpfiles-setup.service" "network-online.target" ];
-    wants = [ "network-online.target" ];
+    after = [
+      "dbus.service"
+      "systemd-tmpfiles-setup.service"
+      "network-online.target"
+      "gnome-remote-desktop-configuration.service"
+    ];
+    wants = [
+      "network-online.target"
+      "gnome-remote-desktop-configuration.service"
+    ];
     serviceConfig = {
       Type = "oneshot";
-      RemainAfterExit = true;
+      TimeoutStartSec = "2min";
     };
     path = [
-      pkgs.gnome-remote-desktop
-      pkgs.polkit
-      pkgs.openssl
-      pkgs.hostname
+      pkgs.coreutils
       pkgs.gawk
+      pkgs.gnome-remote-desktop
+      pkgs.hostname
+      pkgs.openssl
+      pkgs.systemd
     ];
     script = ''
-      # Ensure directory structure exists
-      mkdir -p /var/lib/gnome-remote-desktop/.local/share/gnome-remote-desktop
-      chown -R gnome-remote-desktop:gnome-remote-desktop /var/lib/gnome-remote-desktop
+      set -euo pipefail
 
-      TLS_DIR="/var/lib/gnome-remote-desktop/tls"
-      CRED_FILE="/var/lib/gnome-remote-desktop/rdp-credentials"
+      # GRD 50.x invokes pkexec internally for every grdctl --system call, even
+      # when the caller is root.  NixOS exposes the required setuid wrapper at
+      # /run/wrappers/bin/pkexec; the Nix-store polkit binary is not setuid and
+      # must not shadow it.  Prepend the wrapper directory so every subsequent
+      # grdctl --system resolves the correct binary.
+      export PATH="/run/wrappers/bin:$PATH"
+
+      STATE_DIR="/var/lib/gnome-remote-desktop"
+      TLS_DIR="$STATE_DIR/tls"
+      USERNAME_FILE="$STATE_DIR/rdp-username"
+      PASSWORD_FILE="$STATE_DIR/rdp-password"
+      CRED_FILE="$STATE_DIR/rdp-credentials"
+      DEFAULT_USERNAME="sovran"
+
+      grdctl_system() {
+        local rc=0
+
+        if timeout --kill-after=5s 10s \
+          grdctl --system "$@"; then
+          return 0
+        else
+          rc=$?
+        fi
+
+        if [ "$rc" -eq 124 ] || [ "$rc" -eq 137 ]; then
+          echo "grdctl command timed out: $*" >&2
+        fi
+        echo "grdctl command failed (exit $rc): $*" >&2
+
+        return "$rc"
+      }
+
+      mkdir -p "$STATE_DIR/.local/share/gnome-remote-desktop" "$TLS_DIR"
+      chown -R gnome-remote-desktop:gnome-remote-desktop "$STATE_DIR"
+      chmod 700 \
+        "$STATE_DIR" \
+        "$STATE_DIR/.local" \
+        "$STATE_DIR/.local/share" \
+        "$STATE_DIR/.local/share/gnome-remote-desktop" \
+        "$TLS_DIR"
 
-      # Regenerate TLS certificate if missing OR if ownership is wrong
-      # (disable/re-enable cycle can break ownership or grdctl state)
       NEED_REGEN=0
       if [ ! -f "$TLS_DIR/rdp-tls.crt" ] || [ ! -f "$TLS_DIR/rdp-tls.key" ]; then
         NEED_REGEN=1
-      elif [ "$(stat -c '%U' "$TLS_DIR/rdp-tls.key" 2>/dev/null)" != "gnome-remote-desktop" ]; then
+      elif [ "$(stat -c '%U:%G' "$TLS_DIR/rdp-tls.key" 2>/dev/null)" != "gnome-remote-desktop:gnome-remote-desktop" ]; then
         NEED_REGEN=1
       fi
 
       if [ "$NEED_REGEN" = "1" ]; then
-        mkdir -p "$TLS_DIR"
         rm -f "$TLS_DIR/rdp-tls.key" "$TLS_DIR/rdp-tls.crt"
         openssl req -x509 -newkey ec -pkeyopt ec_paramgen_curve:prime256v1 \
           -sha256 -nodes -days 3650 \
@@ -75,39 +109,59 @@ lib.mkIf config.sovran_systemsOS.features.rdp {
         echo "Generated new RDP TLS certificate"
       fi
 
-      # Always fix ownership and permissions (handles re-enable after disable)
-      chown -R gnome-remote-desktop:gnome-remote-desktop "$TLS_DIR"
+      chown gnome-remote-desktop:gnome-remote-desktop "$TLS_DIR/rdp-tls.key" "$TLS_DIR/rdp-tls.crt"
       chmod 600 "$TLS_DIR/rdp-tls.key"
       chmod 644 "$TLS_DIR/rdp-tls.crt"
 
-      # Configure TLS certificate
-      grdctl --system rdp set-tls-cert "$TLS_DIR/rdp-tls.crt"
-      grdctl --system rdp set-tls-key "$TLS_DIR/rdp-tls.key"
+      if [ ! -f "$USERNAME_FILE" ]; then
+        printf '%s\n' "$DEFAULT_USERNAME" > "$USERNAME_FILE"
+      fi
+      USERNAME="$(tr -d '\n' < "$USERNAME_FILE")"
+      if [ -z "$USERNAME" ]; then
+        USERNAME="$DEFAULT_USERNAME"
+        printf '%s\n' "$USERNAME" > "$USERNAME_FILE"
+      fi
+      if [ "''${#USERNAME}" -gt 32 ]; then
+        echo "RDP username is too long (''${#USERNAME} characters, maximum 32): $USERNAME from $USERNAME_FILE" >&2
+        exit 1
+      fi
+      case "$USERNAME" in
+        [A-Za-z_][A-Za-z0-9_-]*)
+          ;;
+        *)
+          echo "RDP username must start with a letter or underscore and contain only letters, numbers, underscores, and hyphens: $USERNAME from $USERNAME_FILE" >&2
+          exit 1
+          ;;
+      esac
+      chown gnome-remote-desktop:gnome-remote-desktop "$USERNAME_FILE"
+      chmod 600 "$USERNAME_FILE"
 
-      # Generate password on first boot only
-      PASSWORD=""
-      if [ ! -f /var/lib/gnome-remote-desktop/rdp-password ]; then
-        PASSWORD=$(openssl rand -base64 16)
-        echo "$PASSWORD" > /var/lib/gnome-remote-desktop/rdp-password
-        chmod 600 /var/lib/gnome-remote-desktop/rdp-password
-      else
-        PASSWORD=$(cat /var/lib/gnome-remote-desktop/rdp-password)
+      if [ ! -f "$PASSWORD_FILE" ]; then
+        openssl rand -base64 16 > "$PASSWORD_FILE"
+      fi
+      PASSWORD="$(tr -d '\n' < "$PASSWORD_FILE")"
+      if [ -z "$PASSWORD" ]; then
+        echo "RDP password file is empty: $PASSWORD_FILE" >&2
+        exit 1
+      fi
+      if [ "''${#PASSWORD}" -lt 8 ]; then
+        echo "RDP password is too short (''${#PASSWORD} characters, minimum 8): $PASSWORD_FILE" >&2
+        exit 1
+      fi
+      chown gnome-remote-desktop:gnome-remote-desktop "$PASSWORD_FILE"
+      chmod 600 "$PASSWORD_FILE"
+
+      LOCAL_IP="$(hostname -I | awk '{print $1}')"
+      if [ -z "$LOCAL_IP" ]; then
+        LOCAL_IP="127.0.0.1"
       fi
 
-      # Write username to a separate file for the hub
-      echo "sovran" > /var/lib/gnome-remote-desktop/rdp-username
-      chmod 600 /var/lib/gnome-remote-desktop/rdp-username
-
-      # Get current IP address
-      LOCAL_IP=$(hostname -I | awk '{print $1}')
-
-      # Always rewrite the credentials file with the current IP
       cat > "$CRED_FILE" <<EOF
       ========================================
        GNOME Remote Desktop (RDP) Credentials
       ========================================
 
-       Username: sovran
+       Username: $USERNAME
        Password: $PASSWORD
 
        Connect from any RDP client to:
@@ -116,11 +170,22 @@ lib.mkIf config.sovran_systemsOS.features.rdp {
       ========================================
       EOF
 
+      chown gnome-remote-desktop:gnome-remote-desktop "$CRED_FILE"
       chmod 600 "$CRED_FILE"
 
-      # Enable RDP backend and set credentials
-      grdctl --system rdp enable
-      grdctl --system rdp set-credentials sovran "$PASSWORD"
+      # Preflight: the NixOS setuid pkexec wrapper must be present and executable
+      # before any grdctl --system call.  Absence means the system was booted
+      # without security.wrappers or the wrapper directory is not mounted yet.
+      if ! test -x /run/wrappers/bin/pkexec; then
+        echo "Preflight check failed: /run/wrappers/bin/pkexec is absent or not executable." >&2
+        echo "GNOME Remote Desktop system configuration requires the NixOS setuid pkexec wrapper at /run/wrappers/bin/pkexec." >&2
+        exit 1
+      fi
+
+      grdctl_system rdp enable
+      grdctl_system rdp set-tls-cert "$TLS_DIR/rdp-tls.crt"
+      grdctl_system rdp set-tls-key "$TLS_DIR/rdp-tls.key"
+      grdctl_system rdp set-credentials "$USERNAME" "$PASSWORD"
 
       echo "GNOME Remote Desktop RDP configured successfully"
     '';