What causes "failed to delete Redis key: %w" in Tencent/WeKnora?

Check Redis connectivity (redis-cli ping) and the app's Redis logs for the underlying delErr.

failed to delete Redis key: %w

Error message

failed to delete Redis key: %w

What it means

DeleteWebSearchTempKBState cleans up a temporary knowledge base and its Redis state key. If s.redisClient.Del fails, the function logs a warning and wraps the Redis error as 'failed to delete Redis key: %w'. This signals that cleanup only partially succeeded: the temp KB may be gone but the Redis state entry persists.

Source

Thrown at internal/application/service/web_search_state.go:137

	logger.Infof(ctx, "Cleaning temporary KB for session %s: %s", sessionID, state.KBID)

	// Delete all knowledge items
	for _, kid := range state.KnowledgeIDs {
		if delErr := s.knowledgeService.DeleteKnowledge(ctx, kid); delErr != nil {
			logger.Warnf(ctx, "Failed to delete temp knowledge %s: %v", kid, delErr)
		}
	}

	// Delete the knowledge base
	if delErr := s.knowledgeBaseService.DeleteKnowledgeBase(ctx, state.KBID); delErr != nil {
		logger.Warnf(ctx, "Failed to delete temp knowledge base %s: %v", state.KBID, delErr)
	}

	// Delete the Redis key
	if delErr := s.redisClient.Del(ctx, stateKey).Err(); delErr != nil {
		logger.Warnf(ctx, "Failed to delete Redis key %s: %v", stateKey, delErr)
		return fmt.Errorf("failed to delete Redis key: %w", delErr)
	}

	logger.Infof(ctx, "Successfully cleaned up temporary KB for session %s", sessionID)
	return nil
}

View on GitHub (pinned to 988cbb0330)

Solutions

Check Redis connectivity (redis-cli ping) and the app's Redis logs for the underlying delErr.
Retry the delete; Del failures here are usually transient connection issues.
Verify the Redis user's ACL permits DELETE on the stateKey pattern.
If the key may already be gone, treat redis.Nil/'key not found' as success and only fail on real errors.
As a fallback, rely on TTL expiry for orphaned state keys and alert on repeated cleanup failures.

Example fix

// before
if delErr := s.redisClient.Del(ctx, stateKey).Err(); delErr != nil {
	return fmt.Errorf("failed to delete Redis key: %w", delErr)
}
// after
if delErr := s.redisClient.Del(ctx, stateKey).Err(); delErr != nil && delErr != redis.Nil {
	return fmt.Errorf("failed to delete Redis key: %w", delErr)
}

Defensive patterns

Strategy: try-catch

Validate before calling

if err := s.redisClient.Ping(ctx).Err(); err != nil {
	// degrade: skip Redis cleanup and rely on TTL expiry
}

Try / catch

if err := svc.DeleteWebSearchTempKBState(ctx, sessionID); err != nil {
	if strings.Contains(err.Error(), "failed to delete Redis key") {
		logger.Warnf(ctx, "state cleanup deferred; TTL will expire key: %v", err)
		return nil // or queue a retry
	}
	return err
}

Prevention

Set a TTL on state keys at write time so failed deletes self-heal.
Monitor Redis health and alerts so cleanup windows don't coincide with outages.
Treat redis.Nil as success in delete paths.
Grant the app's Redis user DELETE permission via ACL.
Retry transient Del failures with backoff before surfacing an error.

When it happens

Trigger: Calling DeleteWebSearchTempKBState when Redis is down, connection dropped, the key is on a read-only replica, or the client lacks DELETE permission on the key (e.g. ACL restrictions or a protected prefix).

Common situations: Redis restarted or evicted connections mid-cleanup; network partition between app and Redis; wrong Redis database/ACL in staging vs production; key expired between read and delete causing unexpected client errors.

Related errors

AI-assisted analysis of Tencent/WeKnora@988cbb0330 (2026-09-02). Data as JSON: /api/errors/e42cd0917f14848c. Report an issue: GitHub.