a month ago
On 6 October 2026, a harmless synthetic 1,000,000-byte file was accepted through the bucket Files upload screen at about 15:01 UTC. It appeared in the file list, with usage showing 1 MB.
Two attempts to use the file's Download action around 15:03–15:06 UTC returned no file. In both attempts, the browser's download listener was waiting before the click; it timed out after 29,995 ms and 44,997 ms.
A public Microsoft download completed in the same cloud browser in one attempt at about 15:12 UTC, so download capture was working for another source.
Is there a known issue with the bucket Files Download action, or a supported remedy we should use?
3 Replies
Status changed to Awaiting Railway Response Railway • about 2 months ago
a month ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • about 2 months ago
a month ago
Mind sharing which Cloud Browser you were using?
And just to confirm, you're talking about the Bucket's built in file Upload and Download options, right?
To answer your questions, there's no known issue with a bucket's Download action (or at least none that I know of). As for a remedy: when you download a file Railway creates a presigned URL valid for 1 hour for the resource you're trying to download and downloads it. You can pretty easily create the same thing to download files by using an s3 client, or just use an s3 CLI to download them, this still uses the bucket to download directly and so you won't get any egress charges.
a month ago
We used OpenAI's dot/ChatGPT cloud browser. Yes, this was the bucket's built-in Upload and Download controls on the Files tab. The 1,000,000-byte nonprivate sample uploaded and appeared in the list. Two fresh Download clicks in that same short session produced no captured download or local file, while an official Microsoft public download worked in the same browser.
Is there a documented dashboard route that produces a normal browser download without exporting bucket access keys? Or is there a specific nonsensitive browser behavior we can report to help diagnose this?