504 Gateway Time-out in Inference Server Endpoints

Inference Endpoints on the Hub
1

On calling a large Whisper model, with a certain longer audio file, it seems to give 504 Gateway Time-out on a POST request to the API.
I had used the model on Spaces as well, and it had worked quite well, when being tested on multiple larger audio files.
Point to note is that I started getting this error from 19th December 2023, it was running quite smoothly before that.

2

Hi @deepsync , Thanks for reporting. We’ll take a look and get back to you soon.

3

We’re having the same problem that started yesterday at 5am Eastern Time.

4

Thank you @michellehbn
Looking forward to the solution. Also, if possible, it would be great if you can let us know the cause of it.

5

exactly the same problem here. In fact, if I try to send short audios I don’t get the 504 error. Audios longer than 5 minutes are giving 504

6

Hi @feliperafael @deepsync @rexmadden, Thanks for bearing with us while we investigated. We’ve applied a fix and this should now be working as expected. Please let us know though if that’s not the case.

1 Like
7

I haven’t had any more issues over here. Thanks for the quick support.