From f8119f37866d88629c76fe660516a9345a8ecf3f Mon Sep 17 00:00:00 2001
From: Gmer4Lfe <gmer4lfe@gmail.com>
Date: Sat, 22 Aug 2026 00:20:40 -0400
Subject: [PATCH] Give the test harness a stop that cannot strand its iptables
 rule, and clean up the dry-run state copy

---
 Fallback/fallback.sh      | 11 +++++++-
 Fallback/fallback_test.sh | 57 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 67 insertions(+), 1 deletion(-)

diff --git a/Fallback/fallback.sh b/Fallback/fallback.sh
index eebec5e..98ef2a0 100755
--- a/Fallback/fallback.sh
+++ b/Fallback/fallback.sh
@@ -928,7 +928,16 @@ echo "  $ICON_NET  Remote IP: $REMOTE_SERVER"
 echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
 
 FALLBACK_RUNNING=true
-trap 'FALLBACK_RUNNING=false; warn "Fallback received shutdown signal — stopping cleanly"; exit 0' \
+
+# The dry-run state copy is per-PID and would otherwise accumulate one file per preview run.
+# EXIT as well as the signals, because a dry run is usually ended with Ctrl-C or --stop but can
+# also just fall out of the loop.
+dryrun_state_cleanup() {
+    [[ "$DRY_RUN" == true && "$FALLBACK_STATE_FILE" == *".dryrun."* ]] && rm -f "$FALLBACK_STATE_FILE"
+    return 0
+}
+trap 'dryrun_state_cleanup' EXIT
+trap 'FALLBACK_RUNNING=false; warn "Fallback received shutdown signal — stopping cleanly"; dryrun_state_cleanup; exit 0' \
     SIGTERM SIGINT
 
 while [[ "$FALLBACK_RUNNING" == true ]]; do
diff --git a/Fallback/fallback_test.sh b/Fallback/fallback_test.sh
index cb5d9f3..fe3368b 100755
--- a/Fallback/fallback_test.sh
+++ b/Fallback/fallback_test.sh
@@ -118,6 +118,11 @@
 # fallback_test.sh --log
 #     Verbose output on every check in every phase.
 #
+# fallback_test.sh --stop
+#     Stop a running test. SIGTERM only — never SIGKILL, because only this script's EXIT
+#     trap removes the iptables DROP rule it installed. Also sweeps a rule stranded by an
+#     earlier SIGKILL or power cut.
+#
 # ==============================================================================================
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
@@ -126,6 +131,58 @@ source "$SCRIPT_DIR/../load_config.sh"
 
 parse_args "$@"
 
+# ── Stop mode — runs before acquire_lock so we can target the holding instance ────────────────
+#
+# SIGTERM ONLY, and deliberately no SIGKILL escalation — the opposite of fallback.sh --stop.
+# A running test holds an iptables DROP rule against the partner, and the only thing that removes
+# it is this script's own EXIT trap. SIGKILL does not run traps, so force-killing a test strands
+# the rule: the partner stays invisible, fallback.sh reads that as a permanent outage and holds
+# FALLBACK indefinitely. A test that will not die is a worse outcome than a test still running,
+# so this reports the stranded rule and the command to clear it rather than causing one.
+if [[ " ${PARSED_ARGS[*]:-} " == *" --stop "* ]]; then
+    LOCKFILE="${LOCK_DIR}/fallback_test.lock"
+    if [[ ! -f "$LOCKFILE" ]]; then
+        log "No fallback_test.sh lock found — not running"
+        exit 0
+    fi
+    lock_content=$(cat "$LOCKFILE" 2>/dev/null)
+    target_pid="${lock_content%%:*}"
+    if [[ -z "$target_pid" ]] || ! kill -0 "$target_pid" 2>/dev/null; then
+        warn "Stale lock — fallback_test.sh not running (PID ${target_pid:-unknown} gone) — clearing"
+        rm -f "$LOCKFILE"
+        # A stale lock is exactly the SIGKILL/power-cut case, so the rule may still be in place.
+        if iptables -C OUTPUT -d "${REMOTE_SERVER:-0.0.0.0}" -j DROP 2>/dev/null; then
+            warn "Stranded iptables DROP rule found for $REMOTE_SERVER — removing"
+            iptables -D OUTPUT -d "$REMOTE_SERVER" -j DROP 2>/dev/null \
+                && warn "Stranded rule removed — remote connectivity restored ✅" \
+                || error "Could not remove stranded rule — run: iptables -D OUTPUT -d $REMOTE_SERVER -j DROP"
+        fi
+        exit 0
+    fi
+    warn "Stopping fallback_test.sh (PID $target_pid) — SIGTERM so its trap clears the iptables rule..."
+    kill -TERM "$target_pid" 2>/dev/null || true
+    waited=0
+    while kill -0 "$target_pid" 2>/dev/null && [[ "$waited" -lt 30 ]]; do
+        sleep 1
+        (( waited++ )) || true
+    done
+    if kill -0 "$target_pid" 2>/dev/null; then
+        error "fallback_test.sh (PID $target_pid) did not exit within 30s"
+        error "NOT force-killing — SIGKILL would strand the iptables DROP rule on $REMOTE_SERVER"
+        error "Wait, or clear manually: iptables -D OUTPUT -d $REMOTE_SERVER -j DROP"
+        exit 1
+    fi
+    warn "Stopped: fallback_test.sh (PID $target_pid) ✅"
+    if iptables -C OUTPUT -d "${REMOTE_SERVER:-0.0.0.0}" -j DROP 2>/dev/null; then
+        error "iptables DROP rule for $REMOTE_SERVER survived the stop — removing"
+        iptables -D OUTPUT -d "$REMOTE_SERVER" -j DROP 2>/dev/null \
+            && warn "Rule removed ✅" || error "Could not remove — run it by hand"
+    else
+        log "No iptables DROP rule remains for $REMOTE_SERVER ✅"
+    fi
+    exit 0
+fi
+
 FALLBACK_SCRIPT="$SCRIPT_DIR/fallback.sh"
 DOCKER_TIMEOUT=15