From 04cf74ccd34ab8833d570e977549cff6402ff1ad Mon Sep 17 00:00:00 2001
From: Mathieu Moran <mathieumoran1@gmail.com>
Date: Sun, 9 Aug 2026 15:44:19 -0400
Subject: [PATCH] Fix flaky hotspot->home WiFi reconnect in netwatch

Real-hardware finding: netwatch repeatedly reported 'home WiFi not found'
4 times over 13min while sitting right next to the router. Root cause:
a plain 'nmcli con up home-wifi' straight out of AP mode relies on NM's
scan cache for wlan0, which is stale/empty coming out of AP mode (the
radio can't scan while acting as an AP). A single con-up attempt didn't
wait for a fresh scan to land before giving up.

Fix: explicitly drop the AP connection first to free the radio, force a
fresh scan with a settle delay, and retry the connect attempt (longer
30s timeout) up to 2 times before conceding and restoring the hotspot.
Bumped the systemd unit's TimeoutStartSec to 180s to give the retry loop
enough headroom (was defaulting to 90s).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---
 deploy/yaris-netwatch.service |  3 +++
 deploy/yaris-netwatch.sh      | 27 +++++++++++++++++++++++++--
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/deploy/yaris-netwatch.service b/deploy/yaris-netwatch.service
index 65cb664..eb96acf 100644
--- a/deploy/yaris-netwatch.service
+++ b/deploy/yaris-netwatch.service
@@ -5,4 +5,7 @@ Wants=NetworkManager.service
 
 [Service]
 Type=oneshot
+# Retry loop (down AP, rescan, retry connect up to 2x) can take ~80s worst
+# case - default 90s systemd start timeout was too tight a margin.
+TimeoutStartSec=180
 ExecStart=/usr/local/bin/yaris-netwatch.sh
diff --git a/deploy/yaris-netwatch.sh b/deploy/yaris-netwatch.sh
index 9eb3ca8..a068d99 100755
--- a/deploy/yaris-netwatch.sh
+++ b/deploy/yaris-netwatch.sh
@@ -39,9 +39,32 @@ fi
 
 logger -t "$LOG_TAG" "wlan0 is on hotspot, checking for home WiFi ($HOME_CON)..."
 
-if nmcli --wait 20 con up "$HOME_CON" >"$LAST_LOG" 2>&1; then
+# Real-hardware finding: a plain "nmcli con up home-wifi" straight out of AP
+# mode reliably reports "not found" even when actually in range, repeatedly
+# (confirmed: 4 failures in a row over 13min while sitting right next to the
+# router). NM's scan cache for wlan0 is stale/empty coming out of AP mode
+# (the radio literally can't scan while acting as an AP - see comment above),
+# and a single con-up attempt doesn't wait for a fresh scan to land before
+# giving up. Fix: explicitly drop the AP connection first to free the radio,
+# force a fresh scan and give it a few seconds to populate, then retry the
+# connect attempt with a longer timeout up to 2 times before conceding and
+# restoring the hotspot.
+nmcli con down "$AP_CON" >"$LAST_LOG" 2>&1
+
+connected=false
+for attempt in 1 2; do
+    nmcli device wifi rescan ifname "$IFACE" >>"$LAST_LOG" 2>&1
+    sleep 5
+    if nmcli --wait 30 con up "$HOME_CON" >>"$LAST_LOG" 2>&1; then
+        connected=true
+        break
+    fi
+    logger -t "$LOG_TAG" "home WiFi attempt $attempt failed, retrying..."
+done
+
+if [ "$connected" = true ]; then
     logger -t "$LOG_TAG" "home WiFi found, switched to '$HOME_CON'"
 else
-    logger -t "$LOG_TAG" "home WiFi not found, restoring hotspot"
+    logger -t "$LOG_TAG" "home WiFi not found after retries, restoring hotspot"
     nmcli con up "$AP_CON" >>"$LAST_LOG" 2>&1
 fi