apps/api/src/jobs/prefetch-worker.service.ts

@@ -59,6 +59,17 @@ const WORKER_RATE_MAX = Number(process.env.PREFETCH_RATE_MAX) || 20;
  */
 const PCAT_PACE_MS = Number(process.env.PREFETCH_PCAT_DELAY_MS) || 15_000;
 
+// ── Phase-1 residue exclusion ──
+/**
+ * Skip a zero-parts vehicle once this many backfill attempts have completed with
+ * still-zero parts (genuine upstream-data gaps — e.g. a VIN in no catalog).
+ * Re-picking these every wave wastes capacity and starves the Phase-2 rescan.
+ */
+const NORESULT_MAX_ATTEMPTS = Number(process.env.PREFETCH_NORESULT_MAX) || 2;
+/** TTL for the no-result counter — excluded vehicles retry after this, so a later
+ *  catalog fix eventually re-fills them (default 7 days). */
+const NORESULT_TTL_S = (Number(process.env.PREFETCH_NORESULT_TTL_DAYS) || 7) * 86_400;
+
 @Injectable()
 export class PrefetchWorkerService implements OnModuleInit, OnModuleDestroy {
   private readonly logger = new Logger(PrefetchWorkerService.name);
@@ -200,6 +211,9 @@ export class PrefetchWorkerService implements OnModuleInit, OnModuleDestroy {
 
     if (topCategories.length === 0) {
       this.logger.log(`[prefetch] No categories for vehicle=${vehicleId}`);
+      // No categories after a self-seed attempt = no catalog data for this VIN.
+      // Count it as residue so Phase-1 stops re-picking it every wave.
+      await this.markNoResult(vehicleId);
       await updateProgress(this.redis, vehicleId, {
         status: "completed",
         total: 0,
@@ -375,6 +389,10 @@ export class PrefetchWorkerService implements OnModuleInit, OnModuleDestroy {
     const tryPick = async (v: { id: string; source: string | null }): Promise<void> => {
       if (picked.length >= BACKFILL_BATCH_SIZE || seen.has(v.id) || !v.source) return;
       if (await this.redis.exists(`prefetch:scheduled:${v.id}`)) return; // already in flight
+      // Skip exhausted residue: vehicles whose prefetch keeps finishing with zero
+      // parts (no catalog data). They'd otherwise be re-picked every wave forever.
+      const noResult = await this.redis.get(this.noResultKey(v.id));
+      if (noResult && Number(noResult) >= NORESULT_MAX_ATTEMPTS) return;
       seen.add(v.id);
       picked.push({ id: v.id, source: v.source });
     };
@@ -553,6 +571,15 @@ export class PrefetchWorkerService implements OnModuleInit, OnModuleDestroy {
 
     if (isFinished) {
       this.logger.log(`[prefetch] Completed all jobs for vehicle=${vehicleId}`);
+      // Genuine residue: the whole chain finished but the vehicle still has no
+      // parts (all leaves empty / no catalog data). Count it so Phase-1 stops
+      // re-picking it every wave.
+      const [hasPart] = await this.db
+        .select({ id: parts.id })
+        .from(parts)
+        .where(eq(parts.vehicleId, vehicleId))
+        .limit(1);
+      if (!hasPart) await this.markNoResult(vehicleId);
       // Clean up Redis keys — data is in PostgreSQL now
       await this.redis.del(`prefetch:scheduled:${vehicleId}`);
       await this.redis.del(`prefetch:progress:${vehicleId}`);
@@ -566,4 +593,16 @@ export class PrefetchWorkerService implements OnModuleInit, OnModuleDestroy {
       errors: (progress.errors || 0) + 1,
     });
   }
+
+  private noResultKey(vehicleId: string): string {
+    return `prefetch:noresult:${vehicleId}`;
+  }
+
+  /** Record that a backfill attempt finished with the vehicle still at zero parts. */
+  private async markNoResult(vehicleId: string): Promise<void> {
+    const key = this.noResultKey(vehicleId);
+    const n = await this.redis.incr(key);
+    await this.redis.expire(key, NORESULT_TTL_S);
+    this.logger.debug(`[prefetch] no-result #${n} for vehicle=${vehicleId}`);
+  }
 }