Initial commit - Horux Despachos NL

apps/api/src/services/sat/sweep-stale-jobs.service.ts

@@ -0,0 +1,98 @@
+import { prisma } from '../../config/database.js';
+
+export interface SweepResult {
+  pendingFound: number;
+  runningFound: number;
+  pendingMarked: number;
+  runningMarked: number;
+  entries: Array<{
+    id: string;
+    tenantId: string;
+    kind: 'pending-stale' | 'running-stale';
+    ageHours: number;
+  }>;
+}
+
+/**
+ * Watchdog para jobs `sat_sync_jobs` stale.
+ *
+ * Categorías:
+ *   1. `pending` con `nextRetryAt` > pendingHours atrás. El cron horario
+ *      `retryTimedOutJobs` normalmente los retoma, pero si no arranca
+ *      (dev, caída, reinicio largo) el job queda colgado y bloquea el
+ *      lock para nuevos syncs del mismo (tenant, contribuyente).
+ *
+ *   2. `running` con `startedAt` > runningHours atrás. Un sync inicial
+ *      típico termina en <2h; si lleva >runningHours es casi seguro
+ *      huérfano de un proceso que murió. La solicitud SAT ya expiró.
+ *
+ * Marca ambos como `failed` con `errorMessage` descriptivo. Idempotente
+ * (volver a correrlo no reabre los ya-marcados-failed).
+ *
+ * - `apply=false` (default): dry-run, no toca BD.
+ * - `pendingHours`/`runningHours`: thresholds (default 12h / 4h).
+ */
+export async function sweepStaleSatJobs(params: {
+  apply: boolean;
+  pendingHours?: number;
+  runningHours?: number;
+} = { apply: false }): Promise<SweepResult> {
+  const pendingHours = params.pendingHours ?? 12;
+  const runningHours = params.runningHours ?? 4;
+  const now = new Date();
+  const pendingCutoff = new Date(now.getTime() - pendingHours * 3600 * 1000);
+  const runningCutoff = new Date(now.getTime() - runningHours * 3600 * 1000);
+
+  const stalePending = await prisma.satSyncJob.findMany({
+    where: { status: 'pending', nextRetryAt: { lt: pendingCutoff } },
+    orderBy: { createdAt: 'asc' },
+  });
+  const staleRunning = await prisma.satSyncJob.findMany({
+    where: { status: 'running', startedAt: { lt: runningCutoff } },
+    orderBy: { createdAt: 'asc' },
+  });
+
+  const result: SweepResult = {
+    pendingFound: stalePending.length,
+    runningFound: staleRunning.length,
+    pendingMarked: 0,
+    runningMarked: 0,
+    entries: [],
+  };
+
+  for (const j of stalePending) {
+    const ageHours = Math.round((now.getTime() - (j.nextRetryAt ?? j.createdAt).getTime()) / 3_600_000);
+    result.entries.push({ id: j.id, tenantId: j.tenantId, kind: 'pending-stale', ageHours });
+  }
+  for (const j of staleRunning) {
+    const ageHours = Math.round((now.getTime() - (j.startedAt ?? j.createdAt).getTime()) / 3_600_000);
+    result.entries.push({ id: j.id, tenantId: j.tenantId, kind: 'running-stale', ageHours });
+  }
+
+  if (!params.apply) return result;
+
+  for (const j of stalePending) {
+    await prisma.satSyncJob.update({
+      where: { id: j.id },
+      data: {
+        status: 'failed',
+        completedAt: now,
+        errorMessage: `Abandoned by watchdog: pending with nextRetryAt ${j.nextRetryAt?.toISOString()} > ${pendingHours}h in the past. Retry cron didn't pick it up.`,
+      },
+    });
+    result.pendingMarked++;
+  }
+  for (const j of staleRunning) {
+    await prisma.satSyncJob.update({
+      where: { id: j.id },
+      data: {
+        status: 'failed',
+        completedAt: now,
+        errorMessage: `Abandoned by watchdog: running with startedAt ${j.startedAt?.toISOString()} > ${runningHours}h (process crash / orphan). SAT request is lost; re-launch manually.`,
+      },
+    });
+    result.runningMarked++;
+  }
+
+  return result;
+}