Tencent/WeKnora · error

rerank call failed: %w

Error message

rerank call failed: %w

What it means

rerankScores calls the configured rerank model to rescore retrieved passages; any error returned by rerankModel.Rerank is wrapped as 'rerank call failed: %w'. This is a downstream-model invocation failure, not a validation error — the search results exist but the reranking step could not complete.

Source

Thrown at internal/agent/tools/knowledge_search.go:695

	return meta, nil
}

// rerankScores scores the candidates with the configured rerank model and
// returns the raw relevance scores, leaving threshold filtering and composite
// scoring to the caller.
func (t *KnowledgeSearchTool) rerankScores(
	ctx context.Context,
	query string,
	results []*searchResultWithMeta,
) ([]rerank.RankResult, error) {
	passages := make([]string, len(results))
	for i, result := range results {
		passages[i] = t.getEnrichedPassage(ctx, result.SearchResult)
	}

	rerankResp, err := t.rerankModel.Rerank(ctx, query, passages)
	if err != nil {
		return nil, fmt.Errorf("rerank call failed: %w", err)
	}
	return rerankResp, nil
}

func (t *KnowledgeSearchTool) rerankThreshold() float64 {
	if t.config != nil && t.config.Conversation != nil && t.config.Conversation.RerankThreshold > 0 {
		return t.config.Conversation.RerankThreshold
	}
	return 0.3
}

const agentRerankFallbackMinScore = 0.15

func filterRerankRankResults(
	rankResults []rerank.RankResult,
	threshold float64,
	preserveTop bool,
) []rerank.RankResult {

View on GitHub (pinned to 988cbb0330)

Solutions

1. Inspect the wrapped cause (%w) via errors.Unwrap or %v of the returned error for the provider-level reason
2. Verify rerank provider credentials and endpoint configuration
3. Check provider status/rate limits; retry with backoff on transient failures
4. Shrink the number/size of passages sent to the reranker
5. Disable rerank or add a fallback path that returns un-reranked results when reranking fails

Example fix

// before
rerankResp, err := t.rerankModel.Rerank(ctx, query, passages)
if err != nil { return nil, fmt.Errorf("rerank call failed: %w", err) }
// after
rerankResp, err := t.rerankModel.Rerank(ctx, query, passages)
if err != nil {
    logger.Warnf(ctx, "rerank failed, falling back to unranked results: %v", err)
    return results, nil // graceful degradation
}

if rerankEnabled && (rerankEndpoint == "" || rerankAPIKey == "") {
    return errors.New("rerank configured but endpoint/key missing")
}

res, err := tool.Execute(ctx, input)
if err != nil && strings.Contains(err.Error(), "rerank call failed") {
    log.Printf("rerank unavailable: %v", err) // error wraps provider cause
    return executeWithoutRerank(ctx, input) // graceful degradation
}

When it happens

Trigger: t.rerankModel.Rerank returns an error: rerank provider HTTP failure, auth rejection, timeout, oversized payload (too many/long passages), or rerank model misconfigured while rerank is enabled.

Common situations: Rerank API key expired or missing; provider outage or rate limiting; passage list exceeds provider limits; network egress blocked in the deployment environment.

Related errors

• rerank model is not configured: please set rerank_model_id o
• no search targets available
• no queries provided
• failed to do bulk: %w
• failed to delete by query: %w

AI-assisted analysis of Tencent/WeKnora@988cbb0330 (2026-09-02). Data as JSON: /api/errors/0835918a5aee3259. Report an issue: GitHub.