maintainer.py

@@ -127,6 +127,10 @@ def load_state(cfg):
         "pending": None,        # outstanding question awaiting admin reply
         "last_event_check": None,  # max kuma heartbeat id seen (outage history)
         "open_outages": {},     # monitor_id -> open DOWN window, cross-cycle
+        "since_heartbeat": {    # rolling tally flushed into the daily heartbeat
+            "recovered": {},    # monitor name -> self-recovered blip count
+            "incidents": {},    # monitor name -> sustained-outage count
+        },
     }
 
 
@@ -603,10 +607,37 @@ def report_recovered_outages(cfg, state, recovered):
     relay_result(cfg, state, "Outage post-mortem", result)
 
 
+def _bump(counts, names):
+    for n in names:
+        counts[n] = counts.get(n, 0) + 1
+
+
+def _fmt_name_counts(counts):
+    return ", ".join(f"{n} ×{c}" for n, c in sorted(counts.items()))
+
+
+def heartbeat_since_summary(tally):
+    """One-line 'past 24h' rollup for the daily heartbeat, so an all-green
+    snapshot still says what happened between heartbeats."""
+    rec, inc = tally.get("recovered", {}), tally.get("incidents", {})
+    n_rec, n_inc = sum(rec.values()), sum(inc.values())
+    if not n_rec and not n_inc:
+        return "Past 24h: no downtime — nothing needed operational support."
+    parts = []
+    if n_inc:
+        parts.append(f"{n_inc} sustained outage(s) ({_fmt_name_counts(inc)})")
+    if n_rec:
+        parts.append(f"{n_rec} self-recovered blip(s) ({_fmt_name_counts(rec)})")
+    return "Past 24h: " + "; ".join(parts) + "."
+
+
 def health_cycle(cfg, state):
     log("Health cycle: reading Uptime Kuma…")
     monitors = read_kuma_monitors(cfg)
     streak = state.setdefault("down_streak", {})
+    tally = state.setdefault("since_heartbeat", {})
+    tally.setdefault("recovered", {})
+    tally.setdefault("incidents", {})
 
     down_now = []
     for m in monitors:
@@ -620,6 +651,13 @@ def health_cycle(cfg, state):
     confirmed = [m for m in down_now
                  if streak.get(m["name"], 0) >= cfg["down_confirmations"]]
 
+    # Count each sustained outage once — the cycle its streak first crosses the
+    # confirmation threshold — so an outage spanning many cycles isn't tallied
+    # repeatedly in the daily rollup.
+    _bump(tally["incidents"],
+          [m["name"] for m in confirmed
+           if streak.get(m["name"]) == cfg["down_confirmations"]])
+
     log(f"  {len(monitors)} tracked, {len(down_now)} down, "
         f"{len(confirmed)} confirmed down")
 
@@ -632,6 +670,7 @@ def health_cycle(cfg, state):
         state["last_event_check"] = watermark
         if recovered:
             log(f"  {len(recovered)} self-recovered outage(s) since last check")
+            _bump(tally["recovered"], [o["name"] for o in recovered])
             report_recovered_outages(cfg, state, recovered)
     except Exception as e:
         log(f"WARN: outage-history check failed: {e}")
@@ -643,9 +682,11 @@ def health_cycle(cfg, state):
             up = sum(1 for m in monitors if m.get("status") == KUMA_UP)
             matrix_send(
                 cfg,
-                f"✅ **Daily health** — {up}/{len(monitors)} tracked services up.",
+                f"✅ **Daily health** — {up}/{len(monitors)} tracked services up.\n"
+                f"{heartbeat_since_summary(tally)}",
             )
             state["last_heartbeat"] = time.time()
+            tally["recovered"], tally["incidents"] = {}, {}
         return
 
     if cfg["autonomy"] == "notify":