R2 files not accessible in all regions

For Workers & Pages, what is the name of the domain?

dlss-swapper-downloads.beeradmoore.com

What is the error message?

There is no specific error message, just access to files either times out or it doesn’t

What is the issue or error you’re encountering

Small number of users attempting to download files in R2 receiving timeouts

What steps have you taken to resolve the issue?

I am unsure what I can do as it is a R2 network issue

What are the steps to reproduce the issue?

  1. Attempt to download a file from my bucket: https://dlss-swapper-downloads.beeradmoore.com/dlss/nvngx_dlss_v1.0.0.0.zip
  2. If it downloads you do not have the problem, if the browser loads endlessly you are having the problem

I can’t seem to edit so here is some additional information.

I see there is current network issues in Brazil, some of my users are located in the UK. Using a VPN fixes network issues for them (other sites load fine for them).

Status page says “R2 - Degraded Performance” but there is no information if this is to do with the AWS CLI tool issues, or Brazil issue, or other issues, nor is there any report on it being known and worked on (unless I missed that on the page)

We have a similar issue right now with 3 customers, all suspiciously located in the UK. Did you ever figure anything out?

Unfortunately not fully.

I added a network tester in my app and changed my timeout from 20sec to 30min. I also changed to use HTTP/2 instead of 1.1 but I doubt this has any difference.

The test first gets headers before it downloads the rest of the requests content. Some users are reporting ~35sec to just get headers, which is wild. Download of the 30mb (I think that’s how big it is, I can confirm later) test file took about 1min 30sec after that. Slow-ish connections, but the <1kb headers should still have been instant.

Users are also reporting download of the same file in their browser works fine. This could be cloudflare blocking my app but not browsers, but it could also be “user reported or worked” but didn’t also say it took 2mins to work.

Users report using a VPN often fixes their issue. Which then wouldn’t disguise the app itself, just the edge nodes they are hitting.

I also had a test that would use the user agent of a browser, this also did not appear to work.

None of makes sense, and being free users I don’t know of we’ll get any support.

I am now having to offer mirrors in my app so people can load from alternative sources :cry:

Thanks for the additional details. I think we may just need to find a different, more reliable place to host our zip. We just chose this since it was free— I guess you get what you pay for!
-Glen

Also, fwiw, Cloudflare did report this London-specific issue on their status earlier in March, but marked it as resolved: