2026-08-03 10:17:07 +00:00
1 changed files with 13 additions and 1 deletions
@@ -529,7 +529,19 @@ cmd_run() {
        echo "detector.sh: WARN — off-host liveness beacon emit failed (see beacon.sh); the off-host absence check remains the authoritative dead-man." >&2
    fi
    [ "$once" -eq 1 ] && break
-    sleep "$interval"
+    # Close the detector lock fd in the sleep child; otherwise an orphaned sleep
+    # keeps the single-instance flock (fd 9, taken at exec 9> above) alive after
+    # the detector parent dies. The lock is non-blocking (`flock -n`, above), so
+    # for as long as that sleep survives a replacement instance is REFUSED and
+    # exits rather than queueing. This particular hold is BOUNDED by one poll
+    # interval (WAKE_DETECTOR_INTERVAL, default 30s): when the orphaned sleep
+    # exits its copy of fd 9 closes, ending this bounded sleep-child hold. It
+    # does NOT follow that the next start succeeds — other inheritors of fd 9
+    # (the M1 adapter, M2 sink grandchildren) are outside this patch's scope and
+    # can keep holding the flock. The cost this removes is a restart window in
+    # which every supervisor retry fails on the sleep child's account.
+    # `9>&-` closes ONLY the child's copy — the parent's lock is unaffected.
+    sleep "$interval" 9>&-
  done
 }