feat: stop installing the NUT client on new Proxmox nodes

The rack moved from a USB-attached APC on monitor-02 to a UniFi UPS 2U Pro,
a network device running its own NUT server. Cluster nodes no longer run
NUT at all, so this section installed a client that could never connect.

Two reasons it is gone rather than repointed:

  - The UPS allow-lists NUT client IPs and caps at ~3, and requires
    credentials. Only monitor-02 is allow-listed, and it cannot relay:
    dummy-ups repeater mode on NUT 2.8.1 (Debian trixie) cannot
    authenticate upstream, as `authconf` is 2.8.6+.

  - The per-node upsmon was harmful anyway. Its only action was
    SHUTDOWNCMD "/sbin/poweroff", which skips the graceful VM/LXC
    shutdown and the Ceph protective flags that ups-shutdown.sh sets.

Replaced with a startup banner making the real requirement explicit: a new
node is only covered once it is added to the NODES array in ups-shutdown.sh.
That dependency was previously implicit and easy to miss.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

freshStart.sh

@@ -479,72 +479,47 @@ if [[ ! -s /etc/hwmonDaemon/.env ]] || grep -q "your_api_key_here" /etc/hwmonDae
 fi
 
 # =============================================================================
-# NUT UPS Client — connects to monitor-02 (10.10.10.9) for UPS awareness
+# NUT UPS Client — intentionally NOT installed
 # =============================================================================
+# Cluster nodes no longer run NUT. Do not re-add this.
+#
+# The rack is on a UniFi UPS 2U Pro, a network device running its own NUT
+# server. Two things follow from that:
+#
+#   1. Its NUT server allow-lists client IPs and caps out at ~3, and it
+#      requires credentials. Only monitor-02 (10.10.10.9) is allow-listed.
+#      monitor-02 cannot relay to other nodes either: dummy-ups repeater mode
+#      on NUT 2.8.1 (the version in Debian trixie) cannot authenticate
+#      upstream, since `authconf` is 2.8.6+.
+#
+#   2. More importantly, the per-node upsmon was actively harmful. Its only
+#      action was SHUTDOWNCMD "/sbin/poweroff", which skips the graceful
+#      VM/LXC shutdown and the Ceph protective flags (noout, nobackfill,
+#      norecover, pause). A node powering itself off mid-outage is worse than
+#      one that waits.
+#
+# All shutdown is orchestrated over SSH from monitor-02 by ups-shutdown.sh.
+# A new node is covered automatically once it is added to the NODES array in
+# that script:
+#
+#   https://code.lotusguild.org/LotusGuild/nut-ups  →  ups-shutdown.sh
+#
+# *** Add new UPS-backed Proxmox nodes to that NODES array, or they will not
+# *** be shut down during a power outage.
 if [[ "$PLATFORM" == "pve" ]]; then
-    echo "Installing NUT UPS monitoring client..."
-
-    apt-get install -y nut-client
-
-    mkdir -p /etc/nut
-
-    # Fetch the NUT password from the server
-    NUT_ENV_URL="http://10.10.10.63:3000/LotusGuild/nut-ups/raw/branch/main/upsmon-client.conf.template"
-
-    # Fetch the NUT password from monitor-02 (server holds the canonical password)
-    # We pull the client conf template and substitute the password inline.
-    # The password must be set manually or fetched here if you store it securely.
     echo ""
     echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-    echo "  NUT UPS Client Configuration"
+    echo "  UPS shutdown coverage — ACTION REQUIRED"
+    echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
+    echo "  This node does NOT run a NUT client (by design)."
+    echo "  Power-loss shutdown is orchestrated over SSH from monitor-02."
+    echo ""
+    echo "  To cover this node, add it to the NODES array in ups-shutdown.sh:"
+    echo "      repo: LotusGuild/nut-ups"
+    echo "      line: NODES=( \"<ip>:<hostname>\" ... )"
+    echo ""
+    echo "  Also confirm monitor-02 can SSH here as root without a password:"
+    echo "      ssh -o BatchMode=yes root@$(hostname -I | awk '{print $1}') hostname"
     echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-    echo "  Enter the NUT password (from /etc/nut/nut.env on monitor-02):"
-    read -r -s -p "  NUT_PASSWORD: " NUT_CLIENT_PASS
     echo ""
-
-    if [[ -n "$NUT_CLIENT_PASS" ]]; then
-        # Write nut.conf (client mode)
-        cat > /etc/nut/nut.conf << 'NUTEOF'
-MODE=netclient
-NUTEOF
-
-        # Fetch upsmon-client.conf.template and substitute password
-        if curl --max-time 15 --retry 2 -s "$NUT_ENV_URL" -o /tmp/upsmon-client.conf.template; then
-            sed "s/NUT_PASSWORD_PLACEHOLDER/${NUT_CLIENT_PASS}/g" \
-                /tmp/upsmon-client.conf.template > /etc/nut/upsmon.conf
-            rm -f /tmp/upsmon-client.conf.template
-        else
-            # Fallback: write inline
-            cat > /etc/nut/upsmon.conf << UPSMONEOF
-MONITOR apcups@10.10.10.9 1 upsmon_remote ${NUT_CLIENT_PASS} slave
-MINSUPPLIES    1
-SHUTDOWNCMD    "/sbin/poweroff"
-POLLFREQ       5
-POLLFREQALERT  2
-RBWARNTIME     43200
-NOCOMMWARNTIME 300
-FINALDELAY     5
-NOTIFYFLAG ONLINE   SYSLOG+WALL
-NOTIFYFLAG ONBATT   SYSLOG+WALL
-NOTIFYFLAG LOWBATT  SYSLOG+WALL
-NOTIFYFLAG FSD      SYSLOG+WALL
-NOTIFYFLAG SHUTDOWN SYSLOG+WALL
-NOTIFYFLAG NOCOMM   SYSLOG+WALL
-UPSMONEOF
-        fi
-
-        chown root:nut /etc/nut/upsmon.conf 2>/dev/null || true
-        chmod 640 /etc/nut/upsmon.conf
-
-        systemctl enable nut-monitor
-        if systemctl start nut-monitor 2>/dev/null; then
-            echo "✓ NUT client connected to UPS monitor on monitor-02"
-        else
-            echo "⚠️  NUT client installed but could not connect yet (monitor-02 may not be running NUT server)"
-            echo "   It will connect automatically once nut-server is running on monitor-02."
-        fi
-    else
-        echo "⚠️  NUT password not provided — skipping NUT client setup."
-        echo "   Run manually: apt install nut-client && configure /etc/nut/upsmon.conf"
-    fi
 fi