🎈 Sync 2026-10-04 14:38:18

luci-app-homeproxy-pro/root/etc/homeproxy-pro/scripts/update_resources_cron.sh

@@ -0,0 +1,47 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Refresh the four resource lists (china_ip4 / china_ip6 / gfw_list /
+# china_list) and reload the service only when one of them actually moved.
+#
+# Scheduled unconditionally - see hp_sync_resource_cron() in runtime/service.sh
+# for why this is not behind the subscription auto_update switch.
+#
+# Why a reload is needed at all, when the sing-box side reloads itself: the
+# lists feed two consumers. update_resources.sh regenerates the sing-box
+# rule-sets from them, and sing-box watches those files and picks a new one up
+# in place. The firewall reads the same lists directly when firewall_post.ut is
+# rendered into the nft sets, and that only happens on an fw4 reload - so
+# without this the kernel kept routing from the previous list while sing-box
+# had already moved to the new one, and the two halves of the same split
+# disagreed. That window is invisible: mainland traffic still works, it just
+# stops being mainland for whatever the new entries added.
+#
+# update_resources.sh exits 3 when a list is already current, so a day with no
+# upstream movement costs one round of requests and no interruption.
+
+SCRIPTS_DIR="/etc/homeproxy-pro/scripts"
+NAME="homeproxy-pro"
+RUN_DIR="/var/run/$NAME"
+LOG_PATH="$RUN_DIR/$NAME.log"
+
+mkdir -p "$RUN_DIR" 2>/dev/null
+
+log() {
+	printf '%s %s\n' "$(date "+%Y-%m-%d %H:%M:%S")" "$*" >> "$LOG_PATH"
+}
+
+changed=0
+for i in "china_ip4" "china_ip6" "gfw_list" "china_list"; do
+	"$SCRIPTS_DIR"/update_resources.sh "$i"
+	case "$?" in
+	0) changed=1 ;;
+	esac
+done
+
+[ "$changed" = "1" ] || exit 0
+
+log "A resource list changed; reloading $NAME so the nft sets catch up."
+if ! /etc/init.d/$NAME reload >>"$LOG_PATH" 2>&1; then
+	log "Warning: the $NAME reload after a resource update failed; the firewall still uses the previous list."
+fi