It looks like a platform issue if your Space stays paused despite a successful push. Share the build logs along with the Space URL so the Hugging Face team can investigate or reset it.
I have dozens of Spaces that are permanently stuck, with no further explanation or reason. Most of them, IIRC, entered this permafrozen state at or around the same time while the Hub was experiencing some serious outages late last year. The only explanation I can think of that seems somewhat plausible is that something somewhere suffered some serious outages and ultimately data loss or corruption. It’s completely bizarre – the repo itself seems to be fine, all files and commit history seems to have survived, but when you try to restart the instance or even clone the repo into a new space all hell breaks loose and the Hub suddenly doesn’t think the repo exists. The only solution I’ve found is simply to manually download/scrape each file one by one and start on a fresh space/repo.
Considering that these issues (for me at least) has only persisted on spaces using the free tier and set to private, a part of me wants to put on my tinfoil hat on and suspect HF has been aware of the problem for quite some time now and is not in a hurry to fix it, or even initiated intentionally to combat the tens of thousands of spaces ToS-breaking and straight up illegal applications that run on free tier. Tricky situation, and I don’t blame them… providing cloud services were generally seen as scalable business models just five years ago… today, not so much. Again, these are just conjectures, from years of observation and whispers here and there.
My situation may be completely different from yours, but out of curiosity, can you share your requirements.txt? I’ve tried (and failed) to pinpoint a common pattern in these stuck spaces, and I’ve never seen it happen on a new space What happens when you try to push the same repo on a new space?
edit: aaand just saw your screenshots, damn. Definitely different situations here.