Tencent/WeKnora · error

model is not active

Error message

model is not active

What it means

GetEmbeddingModelForTenant only serves models whose status is exactly types.ModelStatusActive. Unlike GetModelByID, it does not distinguish downloading/failed — any non-active status returns errors.New("model is not active"), because a cross-tenant embedder must be fully ready to guarantee vector compatibility. It can also be transient (still downloading) or permanent (download failed/disabled).

Source

Thrown at internal/application/service/model.go:539

	// Fetch model from repository using the specified tenant ID
	model, err := s.repo.GetByID(ctx, tenantID, modelId)
	if err != nil {
		logger.ErrorWithFields(ctx, err, map[string]interface{}{
			"model_id":  modelId,
			"tenant_id": tenantID,
		})
		return nil, err
	}

	if model == nil {
		logger.Error(ctx, "Model not found for specified tenant")
		return nil, ErrModelNotFound
	}

	if model.Status != types.ModelStatusActive {
		logger.Errorf(ctx, "Model is not active, status: %s", model.Status)
		return nil, errors.New("model is not active")
	}

	logger.Infof(ctx, "Getting cross-tenant embedding model: %s, source: %s, tenant: %d", model.Name, model.Source, tenantID)

	appID, appSecret := s.resolveWeKnoraCloudCredentials(ctx, &model.Parameters)

	embedder, err := embedding.NewEmbedder(embedding.ConfigFromModel(model, appID, appSecret), s.pooler, s.ollamaService)
	if err != nil {
		logger.ErrorWithFields(ctx, err, map[string]interface{}{
			"model_id":   model.ID,
			"model_name": model.Name,
			"tenant_id":  tenantID,
		})
		return nil, err
	}

	logger.Info(ctx, "Cross-tenant embedding model initialized successfully")
	return embedder, nil

View on GitHub (pinned to 988cbb0330)

Solutions

1. Check the source-tenant model's status (admin UI or GetModelByID) to see whether it is downloading or failed/disabled, and wait for it to become active or fix the underlying download failure.
2. If the model was intentionally disabled, re-enable it or update the shared knowledge base to use a different active embedding model.
3. Re-trigger the model download if it is in a failed state, then retry once status is active.
4. Add a retry-with-backoff for the transient downloading case, and alert on permanently non-active models referenced by active cross-tenant shares.

Example fix

// before
embedder, err := modelService.GetEmbeddingModelForTenant(ctx, modelID, sourceTenantID)
// after
for i := 0; i < 12; i++ {
    embedder, err = modelService.GetEmbeddingModelForTenant(ctx, modelID, sourceTenantID)
    if err == nil { break }
    time.Sleep(15 * time.Second) // transient "model is not active" while downloading
}

m, err := svc.GetModelByID(ctx, modelID)
if err == nil && m.Status != types.ModelStatusActive {
    // wait or repair before cross-tenant use
}

func isActiveModel(m *types.Model) bool { return m != nil && m.Status == types.ModelStatusActive }

embedder, err := svc.GetEmbeddingModelForTenant(ctx, modelID, tenantID)
if err != nil {
    if strings.Contains(err.Error(), "model is not active") {
        // check whether transient (downloading) or permanent (failed/disabled)
    }
    return nil, err
}

When it happens

Trigger: Calling GetEmbeddingModelForTenant for a model whose Status is anything other than ModelStatusActive — still downloading, download failed, disabled, or deleted — typically right after provisioning a model or after it was disabled in the target tenant.

Common situations: Cross-tenant KB sharing references a source-tenant embedding model that was just created and has not finished downloading; the model was disabled/deactivated by an admin after sharing was set up; the model row failed download earlier; status changed between when the share was created and when it is used.

Related errors

• model ID cannot be empty
• model is currently downloading
• model download failed
• abnormal model status
• unknown credential field:

AI-assisted analysis of Tencent/WeKnora@988cbb0330 (2026-09-02). Data as JSON: /api/errors/e893a0e6cf51ca5f. Report an issue: GitHub.