Random 'Unknown error' thrown by Uppy

Hi all,

I can’t get a local reproduction of this problem so I didn’t want to create a bug request but I’m really struggling with an “Unknown error” that gets thrown by Uppy. I’d say it’s working well 99% of the time but we are occasionally getting reports of uploads failing and I can see them in our logs.

Looking at the codebase there are only two instances where a “Unknown error” string gets used:
https://github.com/search?q=repo%3Atransloadit%2Fuppy%20"unknown%20error"&type=code

We are using the aws-s3 package (v3.6.2).

Attached is a screengrab of the logs I do have but does anyone have any hints or ideas about how I can try to debug this further?

The calls to upload to AWS all seem to continue in the background but the UI shows
“Failed to upload” and never recovers.

Many thanks!

Hi, sorry to hear that. Unfortunately without a way to reproduce it’s hard for us to take a look.

@iblack10 did you ever figure out what could be happening here? I’m running into the same issue when uploading 100 or so 10-20 MB files. I am able to successfully recreate this pretty consistently if that could be helpful in diagnosing.

I haven’t been able to resolve the issue. I was never able to get recreate it myself and only had some logs and user reports to go on. If you can provide the Uppy team with a reproducible example that could definitely help them investigate!

I ran into an [Uppy] [%hh:mm%] Unknown error. It happened in the following conditions:

  • Mobile Safari 18.3
  • 4G network
  • Single file upload with a total size of 527 MB
  • AWS/S3 multipart upload with 100MB chunk size, not using Companion, but a custom endpoint that provides presigned URLs to the client
  • Data uploaded to Cloudflare R2 (S3-compatible)
  • The file upload was initially estimated to finish in less than 2 minutes (I have my own ETA calculating code based on the progress report)
  • The file upload progress reported by uppy for this file later went backwards at various moments, e.g., from 37 to 10%, 53% to 26%, 75% to 5%, etc (numbers not super accurate), which caused confusion and the impression that the upload was never going to finish. Can this be caused by multiple network errors?
  • The browser window performing the upload was put in the background after sometime during the upload (the user switched to another app), and when the user came back, the error (had?) occurred. It is not clear whether the app switch was necessary to trigger the error at this point, but I have not yet another error occurrence to contradict this hypothesis.