fix(PROJ-35): OCR Boot-Resume drosselt nach Queue-Kapazitaet

cmd/archivmail/main.go

@@ -187,20 +187,46 @@ func main() {
 			"pdftotext", ts.HasPdftotext, "tesseract", ts.HasTesseract, "pdftoppm", ts.HasPdftoppm)
 	}
 
-	// Boot-resume: re-enqueue all mails still marked ocr_status='pending'.
+	// Boot-resume: re-enqueue mails still marked ocr_status='pending'.
+	// The worker queue holds 1000 slots; we refill it from the DB whenever
+	// it falls below half capacity. Each iteration only fetches as many
+	// pending mails as currently fit, so nothing is dropped.
+	// The worker updates ocr_status to done/failed/skipped, so subsequent
+	// queries only return genuinely outstanding jobs.
 	go func() {
 		ctx := context.Background()
-		pending, err := mailStore.GetPendingOCRMails(ctx, nil, 5000)
-		if err != nil {
-			logger.Warn("ocr boot-resume: query failed", "err", err)
-			return
-		}
-		if len(pending) == 0 {
-			return
-		}
-		logger.Info("ocr boot-resume: re-enqueueing pending jobs", "count", len(pending))
-		for _, m := range pending {
-			ocrWorker.Submit(m.ID, m.TenantID)
+		queueCap := 1000 // matches ocr.Options.QueueSize above
+		processed := 0
+		idleRounds := 0
+		for {
+			free := queueCap - ocrWorker.QueueLen()
+			if free < queueCap/2 {
+				time.Sleep(2 * time.Second)
+				continue
+			}
+			pending, err := mailStore.GetPendingOCRMails(ctx, nil, free)
+			if err != nil {
+				logger.Warn("ocr boot-resume: query failed", "err", err)
+				return
+			}
+			if len(pending) == 0 {
+				idleRounds++
+				if idleRounds >= 3 {
+					if processed > 0 {
+						logger.Info("ocr boot-resume: backlog drained", "total", processed)
+					}
+					return
+				}
+				time.Sleep(10 * time.Second)
+				continue
+			}
+			idleRounds = 0
+			for _, m := range pending {
+				ocrWorker.Submit(m.ID, m.TenantID)
+			}
+			processed += len(pending)
+			logger.Info("ocr boot-resume: enqueued batch",
+				"batch", len(pending), "total_so_far", processed)
 		}
 	}()