Hello kiasoc5,

this morning I realized that...

Giovanni Biscuolo <g@xelera.eu> writes:

[...]

>> WDYT? any tips/suggestions?
>
> I never tried this,

ehrm: it's false :-O

Actually I'm using a shared store AND "LOCALSTATEDIR/guix/" (usually
/var/guix) between my host OS and all my LXC containers, it's something
I learned by adapting Ludo' and Ricardo notes on installing Guix on a
cluster; please see this message (it was Feb 2019, I forgot I wrote
that) for details and pointers to the relevant documentation:

https://yhetil.org/guix/87h8d8dl6d.fsf@roquette.mug.biscuolo.net/

In your case the "store and state sharing" systems are not running and
at the same time like on a cluster, but from a systemistic point of view
it's (almost) the same thing some of us are already doing in their
setup.

Please pay attention: you can skip the setup of guix-daemon on client
(compute) nodes since... you have no clients :-), but you should keep
guix-daemon updated (as root, sudo -E...) on both dual-boot operating
systems to avoid potential problems due to changes in state data format
and/or API.

Obviously also consider all the caveats I mentioned in my previous
message.

As a side note, you could also consider to switch to Guix System and
keep your foreign distro running as an LXC container sharing store and
state with the host, but you have to be familiar with LXC tooling [1]
AND know how to "convert" a "physical" machine to an LXC container [2],
giving the guests acces to the host GPU [3] for graphical applications
or using remote dektop applications like SPICE or VNC... it's a little
bit complex but pure fun!

[...]

Please keep us updated on your progress!

Happy Hacking! Gio'

[1] see info '(guix) Virtualization services'

[2] this is a sort of howto:
https://askubuntu.com/questions/680608/how-to-migrate-physical-12-04-machine-into-lxc

[3] https://bookstack.swigg.net/books/linux/page/lxc-gpu-access


-- 
Giovanni Biscuolo

Xelera IT Infrastructures