Installing Jina on Truenas

[MaxOutStudio]

Hello, people of this community! I recently came across an amazing A.I called DALL-E Flow by Jina A.I, in the description of the Dall-E Flow Repository on GitHub they mentioned in the description that we can deploy our own server, is it possible to deploy the server on Truenas? If so how do I deploy it?


DALL-E Flow: https://github.com/jina-ai/dalle-flow#server

(Yes I know the instructions are there but they don't explain it for TrueNAS, I'm hoping there is a way to install it... I hope you can help me and if so thanks!)

I'm using TrueNAS-12.0-U8.1

 

[Volts]

I wouldn't bother attempting this on TrueNAS CORE. I assume there are many Linux-specific dependencies.

There's a dockerfile. It might be feasible on SCALE if the hardware requirements are met.

 

[Patrick M. Hausen]

I wouldn't bother attempting this on TrueNAS CORE. I assume there are many Linux-specific dependencies.

Create Ubuntu VM on TN CORE.
Install Docker in Ubuntu VM.
Follow instructions in github repo.

If you happen to already run CORE and all your other requirements are met by the server you have, I don't see why one should switch platforms to run a piece of Linux specific software.

Two instances of Atlassian Confluence here each running in a Ubuntu VM on my TN CORE 13.

HTH,
Patrick

 

[Volts]

Does bhyve support GPU passthrough?

If this was about Jira I'd agree.
This Jina-ai AI overlord thing wants a thicc GPU.

 

[Patrick M. Hausen]

Does bhyve support GPU passthrough?

If you are lucky and have just the right hardware - yes.

 

[Volts]

It would be fun to see that working.

I'm mostly just reluctant to recommend something that isn't supported in the UI.

 

[Patrick M. Hausen]

I'm mostly just reluctant to recommend something that isn't supported in the UI.

Oh it is - sort of. You need to set one tunable in the UI and then you can configure the VM accordingly, also in the UI. There are even success stories here on the forum.

Only caveat is that the system is picky about the mainboard/GPU configuration etc.

E.g. what if that graphics card is also your main console and the kernel configures it as pass-through instead of for its own use? Crash or will a serial console work instead? Does the mainboard have a serial port at all or some IPMI based serial over LAN? If not can you have two graphics cards in that mainboard?

Like that.