Running the migrations in a systemd execStartPre was a mistake. The
migrations can be pretty long to run and easily time-out.
Moving this to a proper oneshot service solves this issue and makes
this fits better the systemd execution model. We can now easily filter
the migrations logs.
The fully-qualified name would certainly be a lot here, but `with` can
still be unclear even with narrow scope. A short `let` adds clarity
without significantly increasing verbosity.
This was incorrectly getting `lib.version` which is e.g.
`"24.11pre-git"`, but should have been the ZFS package version. However,
the condition, at least per the comment, is reversed and should be
instead `versionOlder cfgZfs.package.version "2.2.0"`. However, the
entire premise seems to be incorrect, as ZFS 2.2.6 includes the spl
module. Since the previous condition here was effectively always true,
it would initially seem the best move is to remove the conditional
altogether and always include the spl kmod. However, going back to
4360a87c45 where this condition was added,
the intent appears to be that spl was no longer needed here in
the-pre-release ZFS (long since in all supported versions), due to it
being merged into ZFS mainline. Given that intent and that our boot
tests on all versions succeed without including it in the initrd, remove
it.
These can be either an integer or a range.
Range options are necessary for `FREE_LIMIT` to take effect when used in
conjunction with `TIMELINE_LIMIT_*`.
It is currently tied to `services.avahi.enable` which might not be
desirable.
With this change it is possible to disable the service with
`services.printing.browsed.enable = false`
Factor out part of the provisioning script into a
wait-until-service-is-ready script, and put it unconditionally in
front of ExecStartPost=, so that services that depend on influxdb2 are
not started until influxdb2 responds to requests.
Fixes https://github.com/NixOS/nixpkgs/issues/317017 ("Scrutiny tries to start before influxdb has started")
When `diskImage = null`, the root fs is a tmpfs instead of
`/dev/vda`. Thus, it doesn't have to wait for virtio modules to load
before being mounted. The root fs is a dependency of shared
directories by nature of being their parent directory. Without
depending on `/dev/vda`, these shared directories may attempt to mount
without virtio modules being loaded.
The package has been updated to 0.4 which will result in an auto-migration of the config. This updates our config to match the new expected format. Assertions have been added to warn users that they need to migrate their configuration.
In preparation for the deprecation of `stdenv.isX`.
These shorthands are not conducive to cross-compilation because they
hide the platforms.
Darwin might get cross-compilation for which the continued usage of `stdenv.isDarwin` will get in the way
One example of why this is bad and especially affects compiler packages
https://www.github.com/NixOS/nixpkgs/pull/343059
There are too many files to go through manually but a treewide should
get users thinking when they see a `hostPlatform.isX` in a place where it
doesn't make sense.
```
fd --type f "\.nix" | xargs sd --fixed-strings "stdenv.is" "stdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "stdenv'.is" "stdenv'.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "clangStdenv.is" "clangStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "gccStdenv.is" "gccStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "stdenvNoCC.is" "stdenvNoCC.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "inherit (stdenv) is" "inherit (stdenv.hostPlatform) is"
fd --type f "\.nix" | xargs sd --fixed-strings "buildStdenv.is" "buildStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "effectiveStdenv.is" "effectiveStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "originalStdenv.is" "originalStdenv.hostPlatform.is"
```
Move replaceRuntimeDependencies to the replaceDependencies namespace,
where the structure is more consistent with the replaceDependencies
function. This makes space for wiring up cutoffPackages as an option
too.
By default, the system's initrd is excluded. The replacement process does not
work properly anyway due to the structure of the initrd (the files being copied
into it, and it being compressed). In the worst case (which has been observed
to actually occur in practice), a store path makes it into the incompressible
parts of the archive, checksums are broken, and the system won't boot.
Instead of iterating over all replacements and applying them one by one,
use the newly introduced replaceDependencies function to apply them all
at once for replaceRuntimeDependencies. The advantages are twofold in
case there are multiple replacements:
* Performance is significantly improved, because there is only one pass
over the closure to be made.
* Correctness is improved, because replaceDependencies also replaces
dependencies of the replacements themselves if applicable.
Fixes: https://github.com/NixOS/nixpkgs/issues/4336
4b836fb680 added `pkgs.grub2_efi` to `environment.systemPackages` so that it would be in the Nix store and available for install. But `pkgs.grub2` is already in the list. This causes the various paths of the two GRUB2 versions to collide. To fix this, put `pkgs.grub2_efi` into `system.extraDependencies` instead. This should achieve the same effect of adding the second GRUB2 version to the Nix store without the paths colliding in the environment.
To reproduce the problem, execute `nix-build nixos -I nixos-config=nixos/modules/installer/cd-dvd/iso-image.nix -A config.system.build.isoImage` and look for messages like
```
warning: collision between `/nix/store/9jk1p9n5dl431lcm4w9p6x6x8a00dm0q-grub-2.12/bin/grub-install' and `/nix/store/809l0i6aydg4zhn3kqf723brjyp2qm8h-grub-2.12/bin/grub-install'
```
Adding custom plugins causes the `vim` command to be a wrapper script
running `vim -u ...`, which makes it not load the default ~/.vimrc.
(This is analogous to #177375 about neovim.)
As of Vim 9, the syntax-highlighting portion of the nix plugin is
upstream; the full plugin is only needed for indentation etc. (see also
e261eb152b). So, using regular pkgs.vim
works around this behavior/bug and causes any ~/.vimrc to get loaded,
without regressing the syntax highlighting support that motivated the
change being reverted here.
This reverts commit 0b5a0cbc69.
I'm maintaining the associated packages. So it makes sense to add myself
to their modules as well.
Signed-off-by: Felix Singer <felixsinger@posteo.net>
The 1.x iteration of globalprotect-openconnect is no longer being
developed. Remove related components from nixpkgs.
Signed-off-by: Rahul Rameshbabu <sergeantsagara@protonmail.com>
It thinks we want to expand the `*` regex expressions inside the `sed`
commands. We do not.
Signed-off-by: Fernando Rodrigues <alpha@sigmasquadron.net>
tpm2.target was functionally useless without these services and this
generator. When systemd-cryptsetup-generator creates
systemd-cryptsetup@.service units, they are ordered after
systemd-tpm2-setup-early.service, not tpm2.target. These services are
themselves ordered after tpm2.target.
Note: The systemd-tpm2-setup(-early) services will serve no *function*
under a normal NixOS system at the moment. Because of their
ConditionSecurity=measured-uki, they will always be skipped, unless
you are building an appliance with the system.build.uki feature. Thus,
these are enabled solely for their systemd unit ordering properties.
This module provides some abstraction for a multi-stage build to create
a dm-verity protected NixOS repart image.
The opinionated approach realized by this module is to first create an
immutable, verity-protected nix store partition, then embed the root
hash of the corresponding verity hash partition in a UKI, that is then
injected into the ESP of the resulting image.
The UKI can then precisely identify the corresponding data from which
the entire system is bootstrapped.
The module comes with a script that checks the UKI used in the final
image corresponds to the intermediate image created in the first step.
This is necessary to notice incompatible substitutions of
non-reproducible store paths, for example when working with distributed
builds, or when offline-signing the UKI.
The assertion message should include the `nixpkgs.config` value, however
it currently includes the entire `nixpkgs.config` _option_.
This means the type, declarations, definitions, etc were all printed.
smartctl_exporter already runs with SupplementaryGroups "disk", which
gives full access to SATA drives, but NVMe devices are owned by
root:root, resulting in no access:
[...] msg="Smartctl open device: /dev/nvme0 failed: Permission denied"
This patch introduces a "smartctl-exporter-access" supplementary
group, and an udev rule with setfacl to give the exporter access to NVMe
drives, without changing the base root:root ownership.
Fixes https://github.com/NixOS/nixpkgs/issues/210041
Only restart `systemd-udevd.service` if udev rules in `/etc/udev/rules`
actually changed. The paths of `services.udev.packages` may change
frequently but the resulting udev rules built by `udevRulesFor`
likely change less often.
This has the added benefit of not adding `services.udev.packages` to
the system closure if they are only used for their udev rules
(issue #308937).
This reverts commit 89eb93dc3f.
It broken setups where /etc/wpa_supplicant.conf is configured
imperatively and reloading of the service on configuration changes.
This reverts commit 2e702d07bb.
Segfaults during evaluation in yet-to-be-determined circumstances.
Investigation is ongoing, reverting for now to be safe.
Upstream issue: https://github.com/NixOS/nix/issues/11547
This reverts commit ac849e5658.
Nix 2.24 segfaults semi-randomly during evaluation on specific configs (?).
This commit prepares for another revert, putting the default back to 2.18.
Upstream issue: https://github.com/NixOS/nix/issues/11547
We mistakenly used a non-existing nginx variable for the X-Forwarded-Proto causing
the well-known redirect to return erroneous Location headers like:
Location: ://dav.example/dav
instead of the correct:
Location: https://dav.example/dav
Follow up to #342584.
Similarly to that PR, it is surprising that software which was installed by the user isn't available to a script run over ssm by default.
When executing commands with ssm, users will now have more predictable access to baked-in software instead of an extremely bare-minimum set currently there.
- Cleans up downstream systemd units in favour of using upstream units.
- Xen 4.18 on Nixpkgs now supports EFI booting, so we have an EFI boot
builder here that runs after systemd-boot-builder.py.
- Add more options for setting up dom0 resource limits.
- Adds options for the declarative configuration of oxenstored.
- Disables the automatic bridge configuration, as it was broken.
- Drops legacy BIOS boot
- Adds an EFI boot entry builder script.
Signed-off-by: Fernando Rodrigues <alpha@sigmasquadron.net>
Co-authored-by: Yaroslav Bolyukin <iam@lach.pw>
It is surprising that software which was installed by the user at AMI
generation time isn't available to a script run over user data by
default.
When authoring user data to execute at startup, users will now have
more predictable access to baked-in software instead of an extremely
bare-minimum set currently there.
The current default, 7, is, to put it nicely, absolutely ridiculous.
On 2.4 GHz, the only channels you should use are 1, 6 and 11, because
every other channel overlaps one of those and causes interference.
On 5/6 GHz, channel 7 does not exist at all.
Also, it's 2024, most things will support automatic channel selection,
and those that don't will likely need extra care to support prehistoric
hardware anyway.
That was... interesting to debug. It took a me a bit of reading C code
until I realized that the realpath cache is internally used for
`file_get_contents`, but not for `file_exists` 🙃
I'm not comfortable on doing the workaround in the module, but I think
it's good to have this documented in the manual.
