feat(system): process heartbeat + /system/health, engine runs business fields (W9-I)

- process_heartbeat / process_heartbeat_sample (0034): api / worker / worker:agent / all
  upsert every 15s (event loop lag last/max, concurrency, running runs, SSE connections, memory)
- GET /system/health (admin): processes with 45s stale flag, SSE, DB latency/size/connections,
  disks (DATA_DIR / library roots / download dirs, cache thresholds), mount sentinel checks
  with IO timeout, engine pending/oldest/24h failure rate, schedules
- GET /system/health/history?hours=: bucketed lag / running / SSE series
- /engine/runs and /engine/runs/{id}: business {kind, refId, title, triggeredBy} joined from
  run input / task / agent_run / tasks_run_meta / outbox; businessKind / businessRefId filters
- docs/runbook/ops.md: field meanings and alert thresholds

Co-Authored-By: Claude Opus 5.5 <[email protected]>

apps/server/src/worker/index.ts

@@ -2,7 +2,8 @@ import { EventType } from "@umt/contracts";
 import { OutboxRelay } from "@umt/tasks";
 import type { Logger } from "pino";
 import { collectOutboxRoutes, tasksLogSink } from "../context.ts";
-import type { EventBusPort } from "../http/sse.ts";
+import { type EventBusPort, sseConnectionCount } from "../http/sse.ts";
+import { type HeartbeatWriter, startHeartbeatWriter } from "../modules/system/heartbeat.ts";
 import type { ServerContext, ServerModule } from "../modules/types.ts";
 
 /** agent 组任务:整段会话占一槽数分钟,与文件复制等 io 任务分池 */
@@ -155,6 +156,34 @@ export async function startWorker({ logger, heartbeatMs, bus, runtime, lagProbe
     }
   }
 
+  // 进程与健康页:每 15s 把心跳/延迟/并发/运行中 run 数写进 process_heartbeat(写失败不影响 worker)
+  let hbWriter: HeartbeatWriter | null = null;
+  if (runtime) {
+    const { ctx } = runtime;
+    const groupTasks = new Set(plan.flatMap((p) => p.versions ?? []));
+    const countAll = plan.some((p) => !p.versions);
+    hbWriter = startHeartbeatWriter({
+      db: ctx.db,
+      logger,
+      kind: group === "split" ? "all" : group === "agent" ? "worker:agent" : "worker",
+      sample: async () => {
+        const running = await ctx.engine.list({ status: "running", limit: 500 }).catch(() => null);
+        const s = probe.stats();
+        return {
+          concurrency: {
+            ...(plan.some((p) => p.label !== "agent") ? { io: runtime.concurrency ?? 4 } : {}),
+            ...(plan.some((p) => p.label === "agent") ? { agent: runtime.agentConcurrency ?? 4 } : {}),
+          },
+          runningRuns: running
+            ? running.filter((r) => r.status === "running" && (countAll || groupTasks.has(r.taskName))).length
+            : null,
+          sseConnections: group === "split" ? sseConnectionCount() : null,
+          extra: { group, lagWarnings: s.loopLagWarnings, probeMaxLagMs: s.maxLoopLagMs },
+        };
+      },
+    });
+  }
+
   log.info(
     {
       heartbeatMs,
@@ -170,6 +199,7 @@ export async function startWorker({ logger, heartbeatMs, bus, runtime, lagProbe
     async stop() {
       clearInterval(timer);
       probe.stop();
+      await hbWriter?.stop();
       for (const m of started.reverse()) {
         try {
           await m.stop?.();