<colemickens>
samueldr: btw, from what you know, should I hold out for another rev, or does it make sense to pick up the new 3GB pre-order? (I'm okay with the current state of sw/battery/etc)
<samueldr>
I don't know anything about new revisions
<samueldr>
I'm so out of touch that I don't even know what 1.1 and 1.2 brought :)
<samueldr>
colemickens: I'd go with the updated model, since anyways RAM can't go higher
<samueldr>
oh, I'm so out of touch I assumed braveheart was 1.0
<kahiru>
patagonicus: hey, when you run nixos on the hc2s, do you have /nix on the sdcard is there a way of setting it up differently?
<kloenk>
/nix/store on nfs sounds interesting xP
<patagonicus>
kahiru: My main system currently has it on LVM on a SATA SSD. Needs the uas kernel module added to the kernel and takes a bit at startup before the SATA port is ready, but works fine. For what I eventually plan I'll keep it on the SD card unless that makes them break all the time.
<patagonicus>
I think I read somewhere in the docs that nix store on NFS doesn't work, but I could be wrong.
<kahiru>
it is written in the docs, but I couldn't figure why
<sphalerite>
nix store on nfs should work I think, just shouldn't be shared between multiple hosts
<kahiru>
but maybe there's emphasis on sharing between machines
<kahiru>
dunno
<patagonicus>
I'm tempted to try with two VMs now.
alp has quit [Ping timeout: 272 seconds]
<sphalerite>
what should also work is sharing a read-only nix store, but that means not being able to run nix in its usual capacity (including nixos-rebuild and friends) on the machine where it's mounted
<sphalerite>
That said, mounting a store read-only and then accessing it on the NFS server via --store … could just work
<sphalerite>
now I really want to try this as well, but I have work to do :D
<patagonicus>
A shared read only store would also be nice for VMs - you can just rebuild them on the host.
<patagonicus>
Yeah, same. Few more hours of work then I can keep working on my script for my HC2 deployments.
orivej has quit [Ping timeout: 246 seconds]
orivej_ has joined #nixos-aarch64
stiell has joined #nixos-aarch64
alp has joined #nixos-aarch64
globin has quit [Ping timeout: 260 seconds]
globin has joined #nixos-aarch64
<kloenk>
IIRC the nixos-rebuild build-vm mounts the host store ro into the vm, and then provides an overlay
<kloenk>
The nixos wiki says that nfs store only works on non nixos. Anyone knows why?
alp has quit [Ping timeout: 272 seconds]
orivej_ has quit [Ping timeout: 258 seconds]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-aarch64
cole-h has joined #nixos-aarch64
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-aarch64
<sphalerite>
kloenk: I don't see a particular reason why it wouldn't work on nixos too, with the caveats that either (a) it needs to be unshared, or (b) it needs to be read-only
<sphalerite>
or (c) it must never be garbage-collected
<sphalerite>
kloenk: there might be some issues with early networking as well. But nothing that can't be worked around, I think.
<kloenk>
what is there about garbage collection?
<sphalerite>
kloenk: have a look at /nix/var/nix/profiles and /nix/var/nix/gcroots. These are host-specific, will refer to host-specific locations, and if one host performs a gc it will almost certainly remove a path used by another host if the store is shared between them
<sphalerite>
there might also be issues with concurrent access to nix's sqlite database, not sure on that one.
<kloenk>
ah, of course. that makes sense
<patagonicus>
Hmm. In my script I'm loading the target systems config with "import <nixpkgs/nixos> { configuration = configuration.nix; }". I wonder if I can instantiate that with different hostnames (the four machines will be identical except for a few things like hostnames and ip addresses, but will also need to reference each other's addresses, so I'd
<patagonicus>
preferably just specify everything in one file).
orivej has quit [Ping timeout: 258 seconds]
orivej has joined #nixos-aarch64
<sphalerite>
patagonicus: absolutely. import <nixpkgs/nixos> { configuration = { imports = [ ./configuration.nix ]; networking.hostName = "foo"; } for example
<sphalerite>
for a simple approach
<patagonicus>
Oh. That's way easier than I thought. :D
<sphalerite>
or rather: let mkHost = name: import <nixpkgs/nixos> { configuration = { imports = [ ./configuration.nix ]; networking.hostName = name; }; in map mkHost ["foo" "bar" "baz"]
<sphalerite>
(assuming you want a list)
<patagonicus>
I don't think so, I think I'll just have an argument to my default.nix, which builds the root and boot fs and spits out the script for creating the image with LUKS and everything. Then I can just call it four times for the four machines. Makes it easier to just (re)build a single image.
orivej has quit [Quit: No Ping reply in 180 seconds.]