FR: Check available disk space before downloading models

[Efreak]

From Discord:

I'd like to request a simple disk usage check for downloading models. I set my instance's disk allocation for an a1111 instance, not for a worker (60gb vs 200-300gb), my runpod spent the night restarting itself because not enough space left for cache. 😦

Tazlin:

Can you create an issue on the worker repo?

One of the things coming soon is the addition of all the model file sizes to the model reference

So we can precompute the disk impact

The size would actually help, I'd like to host as many models as I can on my runpod instances, which means fewer big models like Poison (7gb iirc), and more smaller models. Either way, the ability to decide on it's own when to stop downloading models would be helpful in just letting a worker download as many models as it wants to without worrying about it crashing.

[tazlin]

Haidra-Org/horde-model-reference#23 is related.