ATTN HF STAFF: Space stuck building indefinitely

Hello,

Can someone at HF please help fix this space which is stuck perpetually during building? I have tried restarting, factory rebuild, pausing, and new commits, without any impact. The build logs do not show anything.

This is a very frustrating problem. Please help!

Thank you,

Edward

1 Like

I made this change on my end and it did get to the startup process. Try merging it.

@john6666 I merged it, but it’s still stuck in the building phase. Thanks for the idea though.

I see it is working on my end. I think this is a continuation of a problem I had over a month ago where the build would get stuck on a space-by-space basis.
There is no solution, but a workaround is to create a new space and re-upload all the source code.
Renaming and rebuilding often doesn’t solve the problem.
Perhaps something is buggy with the VM units assigned to Spaces.

It would be great to have a response from HF staff. I’ve encountered this problem before, as have others. Having to recreate a space repository should not be necessary!

@not-lain Some of HF’s Spaces have been having a problem for several months now with getting stuck forever during builds. It seems to be occurring on a per-space placement basis, not on a per-source-code basis.
The frequency of this problem has decreased from a period of time, but there is no solution when this happens.

2 Likes

Hi @ejschwartz
I have opened a report about your space problem in ejschwartz/dirty-ghidra · 🚩 Report: Not working

Thanks! I did not think anyone except the space owner looked at those.

reports are sent to huggingface staff to check if there are either something broken or violation so they are pinged when a new report is made.
Well it is a weekend now so we can’t expect fast replies, hope it gets fixed soon.

1 Like

@not-lain It’s a space addition to the perpetual build stack problem. I’m sure quite a few people don’t realize this is a malfunction.

Still no response… :frowning:

1 Like

If there was a higher version of Factory rebuild, it would be possible for individuals to solve the problem, but there isn’t…:sob:

Dealing with issues related to Spaces as a whole takes time anyway. Maybe it’s a problem with the server’s VM configuration or something, but from that point on, we don’t know.
It must be hard to figure out the cause, and it’s not like one or two people are in charge of it.
If it’s urgent, you can contact them on Discord and they may respond to you individually.

Thanks, I posted there. :crossed_fingers: Yeah, it seems like Factory rebuild should ideally fix the problem.

1 Like

I was asked on Discord to send an email website@huggingface.co and I did.

They replied with:

Hi Ed,

Thanks for contacting us. This Space was flagged for using a proxy which infringes on our content policy, specifically:

Proxies that are primarily designed to bypass restrictions imposed by the original service provider

The proxy will need to be removed in order to continue using the Space - let us know once you’ve done that and we can help with next steps!

The only problem is that my space is not using a proxy :rofl: Hopefully we can sort that out.

1 Like

Let’s clone the code with git and upload it to the new space. If you leave your current space, you can see your commit history there.
When the original space is restored, you can backport it.
But this is bad know how…:innocent:

I’m hopeful to get the root cause addressed/fixed.

Why was it misidentified as a proxy? Could it be the code or settings in the space, or is the HF configuration buggy?
That’s where we’d start.
Otherwise, it’s going to be a repeat of being frozen again.

Edit:
I read the code again, and there are no elements related to network access, not even proxies…:crazy_face:

The only thing I found suspicious was the “examles” folder.
I have also experienced false positives with a huge safetensors file, and depending on the contents of the files in the “examles” folder, it could have caused the false positive.

I’m seeing this, not sure what’s triggering it. Maybe there was something suspicious at some point in the commit history? The message does state that it was triggered on October 8th. Perhaps the idea to do a clean push (wihtout the history) to a different repo could be worth trying. Otherwise, not sure if @mcpotato may have any pointers.

Screenshot 2024-10-18 at 11.54.57

1 Like

Thanks, is there no way to override the detection? There is definitely nothing related to socks5 in the repo.

Is there someone at HF who is the main contact for these types of problems?

I’ll make a clone for now just to see what happens…

Edit: Clone is here. I copied the whole git history for now just to see if the problem happens again…

In the case of your space, it doesn’t seem to be like this, and I think it’s some kind of HF malfunction.