| 13 Aug 2021 |
 @kraftnix:matrix.org | i'm digging around in digga trying to figure out if I can hook in from devos and modify the testScripts | 00:44:11 |
 David Arnold (blaggacao) | In reply to @kraftnix:matrix.org
i'm digging around in digga trying to figure out if I can hook in from devos and modify the testScripts we might kill the all profile tests and replace them with a default example in devos (the template / reference implementation) | 00:53:27 |
| @kraftnix:matrix.org | sounds like a plan tbh, although allProfilesTests could be an enable-able option? it is somewhat useful as a check for laptops/single disk systems/simple configs | 00:54:16 |
 @gtrunsec:matrix.org | I like the idea of customProfileTest , but there seem to be some problems with its logic. It should load the activated suites of the current host instead of merged in allProfilesTest. | 00:54:41 |
| @kraftnix:matrix.org | i couldn't see any way to short-circuit the allProfilesTest currently. Although I have a workaround which makes them all pass on my systems (by relaxing a condition), but it's not ideal, it's not like the test really "passes" in the semantic way but gets me over this hurdle. | 00:55:05 |
| David Arnold (blaggacao) | the implementation is so simple,though:
testScript = ''
${host.config.networking.hostName}.systemctl("is-system-running --wait")
'';
| 00:55:50 |
| David Arnold (blaggacao) | * the implementation is so simple,though:
{
testScript = ''
${host.config.networking.hostName}.systemctl("is-system-running --wait")
'';
}
| 00:56:14 |
| @kraftnix:matrix.org | a quick fix would be to pass allProfiles per host through a seive so you could at least disable them | 00:56:15 |
| @kraftnix:matrix.org | yeah a default like host.wait_for("multi-user.target") or "default.target" might be better | 00:56:54 |
| David Arnold (blaggacao) | In reply to @gtrunsec:matrix.org
I like the idea of customProfileTest , but there seem to be some problems with its logic. It should load the activated suites of the current host instead of merged in allProfilesTest. oh, it does neither... | 00:57:11 |
| David Arnold (blaggacao) | (or I made a mistake 🤣) | 00:57:27 |
| @kraftnix:matrix.org | I had some nasty case before where the test never timed out :/ | 00:57:28 |
| David Arnold (blaggacao) | Nice trick, that Las shared over there:a | 01:39:39 |
| David Arnold (blaggacao) | $ mount /dev/disk/by-label/nixos /mnt
$ mkdir -p /mnt/boot && mount /dev/disk/by-label/boot /mnt/boot # UEFI only
$ swapon /dev/$your_swap_partition
# add some extra space to the store, it's running on a tmpfs on your RAM
$ mkdir -p /mnt/tmpstore/{work,store}
$ mount -t overlay overlay -olowerdir=/nix/store,upperdir=/mnt/tmpstore/store,workdir=/mnt/tmpstore/work /nix/store
| 01:39:50 |
| David Arnold (blaggacao) | I'll put it in the docs... | 01:40:25 |
| David Arnold (blaggacao) | And I also was reminded, that good old bud can't work on a read-only flake root. | 01:41:41 |
| David Arnold (blaggacao) | https://devos.divnix.com/start/bootstrapping.html#mount-partitions | 01:46:57 |
| David Arnold (blaggacao) | Does anybody know a smart way to make the baked in flake writable? maybe using overlays? | 01:58:40 |
| David Arnold (blaggacao) | https://github.com/divnix/digga/blob/dbc88a75a2e2dbdeb0b482d1522f215aa68325e5/modules/bootstrap-iso.nix#L17-L20 | 01:59:27 |
| @kraftnix:matrix.org | In reply to @blaggacao:matrix.org
$ mount /dev/disk/by-label/nixos /mnt
$ mkdir -p /mnt/boot && mount /dev/disk/by-label/boot /mnt/boot # UEFI only
$ swapon /dev/$your_swap_partition
# add some extra space to the store, it's running on a tmpfs on your RAM
$ mkdir -p /mnt/tmpstore/{work,store}
$ mount -t overlay overlay -olowerdir=/nix/store,upperdir=/mnt/tmpstore/store,workdir=/mnt/tmpstore/work /nix/store
that's neat, i have actually run into the tmpfs storage issue before when trying to do a 0 -> full system build during bootstrap | 02:50:05 |
| @kraftnix:matrix.org | In reply to @blaggacao:matrix.org
Does anybody know a smart way to make the baked in flake writable? maybe using overlays? in the context of bootstrapIso my hack is just to rsync it the user home dir, edit + build there | 02:50:52 |
| David Arnold (blaggacao) | I still want to do copy --to ssh://[fe08::47%eno1]/mnt though... | 02:51:11 |
| David Arnold (blaggacao) | but the path part of the uri is not parsed in nix ... 😥 | 02:51:46 |
| David Arnold (blaggacao) | * I still want to do nix copy --to ssh://[fe08::47%eno1]/mnt though... | 02:52:03 |
| @kraftnix:matrix.org | ahhh | 02:52:07 |
| David Arnold (blaggacao) | it seems to be a hard wall to hit | 02:52:53 |
| @kraftnix:matrix.org | given that nix doesn't seem to support that, the cleanest hack i can think of (although i'm not sure whether this should be in deploy/yeet) is to run an sshfs to the install target at /mnt/nix/store and then nix copy --to file:///deploy-host/path/to/sshfs 😐️ | 02:59:07 |
| @kraftnix:matrix.org | * given that nix doesn't seem to support that, the cleanest hack i can think of (although i'm not sure whether this should be in deploy/yeet) is to run an sshfs to the install target at /mnt/nix/store and then nix copy --to file:///deploy-host/path/to/sshfs 😐️ | 03:00:35 |
| David Arnold (blaggacao) | ssh://localhost?store=$NIX_STORE_DIR&remote-store=$remoteRoot?store=$NIX_STORE_DIR&real=$remoteRoot$NIX_STORE_DIR what is this???? | 03:51:54 |
| David Arnold (blaggacao) | It certainle doesn't seem that draft uri RFC spec | 03:53:05 |
