Google Chrome net:ERR_FAILED on some requests, not others

[paul132]

What is the name of the domain?

https://remx.xyz

What is the error message?

index-Cf_0Bsxn.js:132 POST https://api.remx.xyz/graphql net::ERR_FAILED

What is the issue you’re encountering

Some requests to our API server are failing, other identical requests succeed

What steps have you taken to resolve the issue?

We have added response headers via Cloudflare:

Cache-Control: no-cache, no-store, must-revalidate
Expires: 0
Pragma: no-cache

We have confirmed that the request is not reaching our API Gateway.

The error only seems to happen in Google Chrome (macOS confirmed, windows not confirmed), is not reproducible in Safari, cannot confirm on Mobile but doesn’t seem to be happening.

We have two identical deployments of the application for QA and development testing, neither confirmed with Cloudflare for DNS/proxy and the error does not seem to occur in either environment.

The chrome network debugger shows Provisional Headers on the failed requests, which seem to indicate that the request was either served from the cache (which it shouldn’t) or that the host was not reachable.

What feature, service or problem is this related to?

I don’t know

What are the steps to reproduce the issue?

Load https://remx.xyz in Google Chrome, open network tab and filter on fetch/XHR requests then wait. The site polls for new content periodically, after some amount of time, one or more of the poll requests will fail while others will succeed.

This seems to happen more to some people than others, some not at all.

[zakri]

It appears the over 1.58k POST requests are being flagged with a “Managed Challenge” by Bot Management. You can review the graph by navigate to Security > Events and filter by Host, Path and HTTP Method. This behavior aligns with your current Bot Management settings, where requests identified as “Likely Automated” are configured to trigger a Managed Challenge.

You can navigate to Security > Events in the Cloudflare dashboard and filter the logs based on criteria such as Host, Path, and HTTP Method.

To address this issue, you can modify the action of Likely Automated or create a custom rule for the requests currently being flagged based on the details blocked in the events.

If you require further assistance, please raise the issue for checking it further.