API requests take critically long to complete, using cloudflare DNS with Proxy
Hey, we’re hosting onedollarstats.com, a SvelteKit SPA, on Cloudflare Pages, and the site itself loads fast overall.
However, when users open the dashboard, some API requests take critically long to complete. We’ve reproduced this issue ourselves and confirmed it’s real. Here’s what it looks like: two screenshots attached.
Approximately 22 users have encountered this issue for last 5 days.
A few things to clarify:
- We use Cloudflare strictly as a DNS provider for our API — not as an API host.
- We have Cloudflare DNS proxy enabled.
- When checking the Cloudflare dashboard, we don’t see any slow requests or issues reflected there.
- On our side, we can’t find a reason for these delays.
Frankly, we don’t want to learn about problems from users — I need to see them before they do. Transparency is really important to us.
So a couple of questions:
1. Where can I actually view detailed request logs or performance metrics for proxied traffic (especially for API endpoints)?
2. Would using Cloudflare Argo help improve or stabilize this situation?
3. Is there any guidance for debugging this kind of issue through the Cloudflare dashboard?
We really want to stay with Cloudflare, but if we can’t get better visibility, we may have to disable the proxy, which we’d prefer to avoid.
2 Replies
You're right, "stalled" can mean that the request is waiting for other requests to complete. I'll send additional screenshots showing not only the main request but also the preflight OPTIONS request, which took 50 seconds and eventually failed. As you understand, it never reached our API, which leads us to believe that the preflight request failed during Cloudflare proxying. Why can a preflight request sometimes get blocked?
We figured out the issue was on our server side. But we still have a question: is there a way to spot critically slow requests in the Cloudflare dashboard? If yes, where exactly can we find that?