Launch timed out, space was not healthy after 30 min (space keeps on building)

Spaces
1

The space keeps on building and after 30 mins it gives runtime error Launch timed out, space was not healthy after 30 min

image
image2472×456 72.7 KB

Have tried restarting and factory rebuilt. Port is already 7860

1 Like
2

image
image828×576 9.17 KB

1 Like
3

Same here.

4

Hi @Anne31415 and @wajid, could you please share more details about your Spaces? The default timeout during runtime is 30 minutes. Our infra expects a service to bind to the exposed port within 30 minutes, otherwise, it throws that error.

5

The same is happening to me. Every space I create from autotrain just never stop building and after 30 minutes I get the error Runtime error Launch timed out, space was not healthy after 30 min Container.

6

hi @MatteoWood ,
Could you please try to increase the startup_duration_timeout param? inside the yaml README.md
and if you keep getting error with the Autotrain template, please create a new issue on the category here

startup_duration_timeout: 1h

7

Hi @radames,
Of course, thank you. I’ll update you in 1 hour then.
I’m not sure it will be enough, though… the training seems to start fine, it’s just ending because the space never reach the running status. In my opinion 30 minutes or 1 hour is not solving the problem unfortunately. Here’s the space: MatteoWood/autotrain-bert-large-uncased-0
In case it is not working, in around 30 hours the fine-tuning should end: would modifing the startup_duration_timeout parameter to something like 40h be a work-around?
Best

8

interesting, probably better start a new thread with @abhishek, startup_duration_timeout only affects the expected time for a service to bind on the exposed port 7860, so maybe is something can be fixed on the autotrain app

9

I confirm the error

Runtime error Launch timed out, space was not healthy after 60 min

10

I’ve now encountered this problem for the third time. None of my spaces were starting or building anymore. I haven’t changed anything in the code, or anything else for that matter, and then a few hours later (without any changes), everything starts working again. These hours are just very frustrating because you never know how long it will be before it starts working again. Right now its working.

I can give you more details about the space - what do you need to know?

11

I am facing the same error and i am using a paid GPU currently so my training just stops due to runtime error after 30 min. Please provide a solution for this because it is still being billed for these 30 min. @radames please provide a solution for this.