- 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>
Using zfs.latestCompatibleLinuxPackages can result in downgrades to the kernel on a system, potentially causing breakage.
This breakage may not be apparent during build and switch, but only after attempting to reboot into the updated generation.
By forcing users to explicitly manage their kernel version, we can ensure that the breakage will be apparent at build time instead.
Also recommends the usage of sudo's -E flag if --use-remote-sudo cannot
be used. This should still be discouraged IMO, as it means Nix may write
root-owned files to the user's home directory.
Signed-off-by: Fernando Rodrigues <alpha@sigmasquadron.net>
After a discussion on Matrix, it has become clear that building as root
is discouraged, and the (inappropriately named) --use-remote-sudo flag
should be enouraged as the de-facto way to selectively escalate to root
after a system build has finished.
Signed-off-by: Fernando Rodrigues <alpha@sigmasquadron.net>
Since `connectionStringFile` reads the file and puts it into the
invocation of the exporter, it's part of the cmdline and thus
effectively world-readable.
Added a new `connectionEnvFile` which is supposed to be an environment
file of the form
PGBOUNCER_EXPORTER_CONNECTION_STRING=...
that will be added to the systemd service. The exporter will read the
connection string from that value.
These take up 2 GiB every time anything in the minimal installer
changes, or up to 4 GiB per day. We already stopped building Amazon
images in 9426d90c67. Meaningful
installer changes are rare enough, and the couple of days it takes
for them to trickle down to the large channel acceptable enough,
that this is mostly a waste of space.
This should buy enough slack to build `stdenv` on `staging` without
contributing to cache size growth.
The Rust `switch-to-configuration-ng` rewrite was carefully written
to be compatible with the original Perl script, has been checked
against NixOS VM tests, and has been available on an opt‐in basis
for testing for the 24.05 release cycle.
The next step towards replacing the Perl script entirely is to
switch it on by default so that we can get real‐world testing from
a much greater number of users. Maintaining two implementations in
parallel is becoming a burden; we are having to adjust the systemd
service activation behaviour slightly to fix a long‐standing bug,
and backporting the changes to the Perl script is an unpleasant
process. We will do it anyway to ensure that the Rust and Perl
implementations keep parity with each other throughout the 24.11
release cycle, but we think the time has come to flip the switch.
Taking this step now will give us two to three months to test this in
the wild before the 24.11 release and gain confidence that there are
no regressions. If any non‐trivial problems arise before the final
release, we will revert to the Perl implementation by default. Doing
this switch ASAP will help to disentangle any problems that might
arise from the Rust implementation from problems that arise from the
systemd service activation changes, or the upcoming switch to using
systemd in stage 1 by default.
The main concern that was raised about replacing the Perl script in the
PR that added `switch-to-configuration-ng` was that it is currently
possible to run NixOS on systems that cannot natively host a Rust
compiler. This does not apply to any platforms that have official
support from NixOS, and as far as I know we do not know of any such
systems with users that are not cross‐compiling anyway.
My understanding is that these systems are already broken by default
anyway, as `systemd.shutdownRamfs.enable` is on by default and uses
`make-initrd-ng`, which is also written in Rust. Switching the default
while keeping the Perl implementation around will give us at least
an entire release cycle to find out if there are any users that will
be affected by this and decide what to do about it if so.
There is currently one known inconsistency between
the Perl and Rust implementations, as documented in
<https://github.com/NixOS/nixpkgs/issues/312297>; the Rust
implementation has more accurate handling of failed systemd units.
We slightly adjust the semantics of `system.switch.enable{,Ng}` to
not conflict with each other, so that `system.switch.enableNg` is
on by default, but turning off `system.switch.enable` still results
in no `switch-to-configuration` implementation being used. This
won’t break the configuration of anyone who already opted in to
`system.switch.enableNg` and is probably how the option should have
worked to begin with.
ReiserFS has not been actively maintained for many years. It has been
marked as obsolete since Linux 6.6, and is scheduled for removal
in 2025. A warning is logged informing users of this every time a
ReiserFS file system is mounted. It suffers from unfixable issues
like the year 2038 problem.
JFS is a slightly more ambiguous case. It also has not been actively
maintained for years; even in 2008 questions were being raised
about its maintenance state and IBM’s commitment to it, and some
enterprise distributions were opting not to ship support for it as
a result. It will [indefinitely postpone journal writes], leading
to data loss over potentially arbitrary amounts of time. Kernel
developers [considered marking it as deprecated] last year, but
no concrete decision was made. There have been [occasional fixes]
to the code since then, but even the developer of much of those was
not opposed to deprecating it.
[considered marking it as deprecated]: https://lore.kernel.org/lkml/Y8DvK281ii6yPRcW@infradead.org/
[indefinitely postpone journal writes]: https://www.usenix.org/legacy/events/usenix05/tech/general/full_papers/prabhakaran/prabhakaran.pdf
[occasional fixes]: https://www.phoronix.com/news/JFS-Linux-6.7-Improvements
Regardless of whether JFS should be removed from the kernel, with all
the implications for existing installations that entails, I think
it’s safe to say that no new Linux installation should be using
either of these file systems, and that it’s a waste of space and
potential footgun to be shipping support for them on our standard
installation media. We’re lagging behind other distributions on
this decision; neither is supported by Fedora’s installation media.
(It also just so happens that `jfsutils` is the one remaining package
in the minimal installer ISO that has reproducibility issues, due to
some cursed toolchain bug, but I’m not trying to Goodhart’s law
this or anything. I just think we shouldn’t be shipping it anyway.)
