External USB drive solely for passthrough to a VM/docker

[Ppriorfl]

I’m a current unraid user but the parity situation with no bitrot correction scares me. I have been watching TN Scale for a year now and think it’s safe to try.

I understand ZFS doesn’t play well with external USB drives and would not do that for my main storage pool.

However specifically I have a lot of drives filled to the brim (99% full) with Chia plots running on my unraid server. It seems foolish to try to limit vdev use to 80% and use these drives - don’t need parity or any protection, if one dies so be it.

So I’m wondering if it’s possible at all to hook up USB drives externally NOT in a vdev and just pass them through to a specific container. They would only be accessed by whatever chia client I would be running (machinaris I think is available).

I guess the question is, does everything have to be ZFS or can truenas access external drives stored in other file system and pass that data to a docker?

Appreciate anyone who might take the time to answer. Thank you.

 

[Ppriorfl]

Nobody??

 

[sretalla]

At present (as is mentioned in many other threads), the SCALE implementation of KVM is done in a way where you can't really use USB device passthrough.

What you can do is pass an entire PCIe device into a VM though, so if you have a USB controller that's suitably available for that, then you have a method of connection.

You can also do that in CORE too.

 

[Ppriorfl]

Thank you. I assume the limit of not passing into a VM would also apply to docker containers as well? I think getting a pcie usb controller to a hub and then into external drives would work for my use case.

 

[sretalla]

I assume the limit of not passing into a VM would also apply to docker containers as well?

I think that is actually an option now... at least for TrueCharts apps...

I haven't tried it personally, so can't report on how well it works, but you could get support over on TrueCharts' Discord if it doesn't. I found them to be quick and very helpful so far.