Hi everyone! I ran into a minor problem. We deal with highly sensitive customer data at the company where I currently work, we work on virtual machines that are fully disconnected from the internet. Is there a way for me to download the Orchest package from my online environment and install and dockerize it on the VMs?

thanks @Masab Anees for bringing this up - @Yannick I already told @Masab Anees that maybe a method could be 1. perform a regular Orchest installation in a machine connected to the Internet 2. disconnect the machine from the Internet 3. do

orchest uninstall

4. rerun

orchest install

, see what happens but that's a no-go as well, since @Masab Anees would need to have the files there in the remote VM

Is there a way for me to download the Orchest package from my online environment and install and dockerize it on the VMs?

Hmmm I am not completely sure I understand what you are asking here. Are you trying to install Orchest on a "clean" VM and after the installation you would remove internet access from the VM? Orchest is already fully containerized. Strictly speaking, Orchest can run without internet access but in general I would not recommend it. To run pipelines in Orchest you need to have Docker images on the VM that represent environments (you could pull these onto the VM before removing internet access). Furthermore, the moment you require a new package, e.g.

pandas

, you would need a way to pull in/install that new package.

Agree with @Yannick's perspective. Orchest wasn't designed to run in non-internet connected environments, although as he describes it will work if you first set it up with internet and use it without internet afterwards (but some features that depend on an internet connection like downloading Python packages won't work).

Would you be able to point Orchest to a local pip library? Had to do this for some space stuff previously. Pain to maintain but possible. (I.e., build the image with all packages available from some on-disk source.)

Thanks for jumping in @Kevin Lalli! 💯 Indeed when building environments, you have access to the project directory. Just like you can do

pip install -r requirements.txt

you can also install from some package inside the project directory (e.g.

pip install -e mypackage

)

Thank you folks for the response. Apologize, if my statement was unclear. The VM I'm talking about doesn't have internet access, and I can't switch it ON because the admin has disabled all of the essential services. We already have a jupyter lab container running in the VM with all of the essential packages installed. Anyhow thank you for clarifying the statement, @Rick Lamers.