Tencent/WeKnora · error

failed to index summary chunk: %w

Error message

failed to index summary chunk: %w

What it means

Thrown when retrieveEngine.BatchIndex fails to embed and write the summary chunk into the vector store. The chunk row exists in the DB but is not retrievable; typically embedding-provider or vector-store failure.

Source

Thrown at internal/application/service/knowledge_process.go:1365

			logger.Errorf(ctx, "Failed to get embedding model: %v", err)
			summaryErr = err
			return fmt.Errorf("failed to get embedding model: %w", err)
		}

		indexInfo := []*types.IndexInfo{{
			Content:         summaryChunk.Content,
			SourceID:        summaryChunk.ID,
			SourceType:      types.ChunkSourceType,
			ChunkID:         summaryChunk.ID,
			KnowledgeID:     knowledge.ID,
			KnowledgeBaseID: knowledge.KnowledgeBaseID,
			IsEnabled:       true,
		}}

		if err := retrieveEngine.BatchIndex(ctx, embeddingModel, indexInfo); err != nil {
			logger.Errorf(ctx, "Failed to index summary chunk: %v", err)
			summaryErr = err
			return fmt.Errorf("failed to index summary chunk: %w", err)
		}

		logger.Infof(ctx, "Successfully created and indexed summary chunk for knowledge: %s", payload.KnowledgeID)
		summaryOut["summary_chunk_indexed"] = true
	}

	logger.Infof(ctx, "Successfully generated summary for knowledge: %s", payload.KnowledgeID)
	summaryOut["status"] = "completed"
	return nil
}

// ProcessQuestionGeneration handles async question generation task. It
// dispatches between the batched fan-out path (current: one task per window of
// text chunks, payload.ChunkIDs set) and the legacy whole-knowledge path (kept
// for tasks enqueued before fan-out shipped, no chunk ids). A lone ChunkID
// (from an interim per-chunk build) is treated as a one-element batch.
func (s *knowledgeService) ProcessQuestionGeneration(ctx context.Context, t *asynq.Task) (retErr error) {
	var payload types.QuestionGenerationPayload

View on GitHub (pinned to 988cbb0330)

Solutions

1. Check embedding provider and vector store health/quotas
2. Verify the summary content does not exceed embedding input limits
3. Retry summary regeneration; indexing is re-attempted end-to-end

When it happens

Trigger: Thrown at internal/application/service/knowledge_process.go:1365 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of Tencent/WeKnora@988cbb0330 (2026-09-02). Data as JSON: /api/errors/7d03b9ccd0b68bbb. Report an issue: GitHub.