the build-time check is not safe (e.g. doesn't protect from bad users or nomissingok
paths missing), so add a new unit for configuration switch time check
Now the service no longer starts immediately,
check if the config we generated makes sense as soon as possible.
The check isn't perfect because logrotate --debug wants to check
users required, there are two problems:
- /etc/passwd and /etc/group are sandboxed and we don't have
visibility of system users
- the check phase runs as nixbld which cannot su to other users
and logrotate fails on this
Until these two problems can be addressed, users-related checks
are filtered out, it's still much better than no check.
The check can be disabled with services.logrotate.checkConfig
if required
(bird also has a preCheck param, to prepare the environment
before check, but we can add it if it becomes necessary)
Since this makes for very verbose builds, we only show errors:
There is no way to control log level, but logrotate hardcodes
'error:' at common log level, so we can use grep, taking care
to keep error codes
Some manual tests:
───────┬──────────────────────────────────────────
│ File: valid-config.conf
───────┼──────────────────────────────────────────
1 │ missingok
───────┴──────────────────────────────────────────
logrotate --debug ok
grep ok
───────┬──────────────────────────────────────────
│ File: postrotate-no-end.conf
───────┼──────────────────────────────────────────
1 │ missingok
2 │ /file {
3 │ postrotate
4 │ test
5 │ }
───────┴──────────────────────────────────────────
error: postrotate-no-end.conf:prerotate, postrotate or preremove without endscript
───────┬──────────────────────────────────────────
│ File: missing-file.conf
───────┼──────────────────────────────────────────
1 │ "test" { daily }
───────┴──────────────────────────────────────────
error: stat of test failed: No such file or directory
───────┬──────────────────────────────────────────
│ File: unknown-option.conf
───────┼──────────────────────────────────────────
1 │ some syntax error
───────┴──────────────────────────────────────────
logrotate --debug ok
error: unknown-option.conf:1 unknown option 'some' -- ignoring line
───────┬──────────────────────────────────────────
│ File: unknown-user.conf
───────┼──────────────────────────────────────────
1 │ su notauser notagroup
───────┴──────────────────────────────────────────
error: unknown-user.conf:1 unknown user 'notauser'
In particular note that logrotate would not error on unknown option
(it just ignores the line) but this change makes the check fail.
using freeform is the new standard way of using modules and should replace
extraConfig.
In particular, this will allow us to place a condition on mails
having pkgs.logrotate depend on mailutils brings in quite a bit of dependencies
through mailutil itself and recursive dependency to guile when most people
do not need it.
Remove mailutils dependency from the package, and conditionally add it to the
service if the user specify the mail option either at top level or in a path
Fixes#162001
This accomplishes multiple things:
- Allows us to start systemd without stage-2-init.sh. This was not
possible before because the environment would have been wrong
- `systemctl daemon-reexec` also changes the environment, giving us
newer tools for the fs packages
- Starts systemd in a fully clean environment, making everything more
consistent and pure
* Change groupId to gid to align with the rest of NixOS modules
* Add a check to the gid option to ensure it is greater than or equal
to 1000
* Use the overridden package for the wrappers
At some point, I'd like to make another attempt at
71f1f4884b ("openssl: stop static binaries referencing libs"), which
was reverted in 195c7da07d. One problem with my previous attempt is
that I moved OpenSSL's libraries to a lib output, but many dependent
packages were hardcoding the out output as the location of the
libraries. This patch fixes every such case I could find in the tree.
It won't have any effect immediately, but will mean these packages
will automatically use an OpenSSL lib output if it is reintroduced in
future.
This patch should cause very few rebuilds, because it shouldn't make
any change at all to most packages I'm touching. The few rebuilds
that are introduced come from when I've changed a package builder not
to use variable names like openssl.out in scripts / substitution
patterns, which would be confusing since they don't hardcode the
output any more.
I started by making the following global replacements:
${pkgs.openssl.out}/lib -> ${lib.getLib pkgs.openssl}/lib
${openssl.out}/lib -> ${lib.getLib openssl}/lib
Then I removed the ".out" suffix when part of the argument to
lib.makeLibraryPath, since that function uses lib.getLib internally.
Then I fixed up cases where openssl was part of the -L flag to the
compiler/linker, since that unambigously is referring to libraries.
Then I manually investigated and fixed the following packages:
- pycurl
- citrix-workspace
- ppp
- wraith
- unbound
- gambit
- acl2
I'm reasonably confindent in my fixes for all of them.
For acl2, since the openssl library paths are manually provided above
anyway, I don't think openssl is required separately as a build input
at all. Removing it doesn't make a difference to the output size, the
file list, or the closure.
I've tested evaluation with the OfBorg meta checks, to protect against
introducing evaluation failures.
Commit 7addb1c0ec disabled this as a
side effect of switching gnome-terminal to gnome-console, but it’s
still useful for gnome-console.
Signed-off-by: Anders Kaseorg <andersk@mit.edu>
Lo and behold, we're finally catching up with Mozillas very own firefox
build in terms of speed.
PGO is an optimization technique in which in a first step we create a
build that supports instrumentation, meaning we can use it to create a
profile of how the browser behaved during usage. Then in a second pass
we create the final build that uses the acquired profiling data to
optimize the browser for the workload it actually received during
profiling.
The downside is that with PGO we now need to build Firefox twice, which
increases the build time from around 20 minutes to roughly 50 minutes.
In the Speedometer 2.0 benchmark multiple tests could see a
responsiveness improvemeant around 20-25%, which makes the increased
build time well worth it.
Sadly this benefit seems limited to x86_64-linux, builds on
aarch64-linux get stuck during profiling and I haven't found out why.
Finally, after a long time, we can say:
Closes: #76484
Supersedes: #129503
We can perform most of the mkdir/ln/rm using systemd-tmpfiles
instead which cleans up the script.
/bin and /home are created by their activation script snippets
usbfs is deprecated and unused.
hwclock seems to be automatically executed by systemd on startup.
The mkswap to prevent hibernation cycles seems to be executed by systemd
as well since the provided regression tests succeeds.
Before this patch, services.dendrite.environmentFile is used for
secrets and environment variable substitution only happens when this
option is used.
systemd-247 provides a mechanism called LoadCredential for secrets and
it is better than environment file. See the section of Environment=
in the manual of systemd.exec for more information.
This patch always substitute environment variables, which enables the
usage of systemd LoadCredential.
* With the upgrade to waydroid to 1.2.0, dependencies that previously
were shipped in the service's path have been moved to the waydroid
package.
* Make sure /var/lib/misc exists when starting waydroid. As required
by dnsmasq
This patch allows creation of files like
/etc/systemd/system/user-.slice.d/limits.conf with
systemd.units."user-.slice.d/limits.conf" = {
text = ''
[Slice]
CPUAccounting=yes
CPUQuota=50%
'';
};
which previously threw an error
Also renames the systemd-unit-path test to sytsemd-misc, and extends it to
test that `systemd.units` can handle directories. In this case we make
sure that resource limits specified in user slices apply.
While doing kernel development with the nixos testing infrastructure, it
is useful to quickly compile the kernel on the side and boot it in the testing VM.
This patch allows overriding the kernel through the environment
variable. For example, the following:
$ NIXPKGS_QEMU_KERNEL_testvm=$LINUX_SRC/arch/x86/boot/bzImage $VM/bin/nixos-test-driver
runs testvm testing VM with the fresh kernel compiled in $LINUX_SRC.
Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
GNOME Shell 42 switched an icon for the accessibility menu in the top panel
to one from gnome-control-center instead of a legacy one from adwaita-icon-theme:
https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2155
Let’s add that dependency to the systemd unit since installing it
through `environment.systemPackages` is not enough due to environment isolation.
The tests complained:
/nix/store/nm3nf5y4hzgmy00lw5s6ls68j38y84y0-gjs-1.72.0-installedTests/libexec/installed-tests/gjs/scripts/testCommandLineModules.sh: line 90: gjs-console: command not found
But they still passed.
Sendto device selection is removed in gnome-bluetooth 42.
I decide it is not worth to maintain a legacy gnome-bluetooth
package for the contract.
This should also be broken in elementary OS 7.
The current wrapper only includes vim, gvim and the man pages
(optionally). This rewrite distinguishes two scenarios, which I expect
cover the majority of use cases:
- standalone mode, when `name != "vim"`, means the user already has a
vim in scope and only wants to add a customized version with a
different name. In this case we only include wrappers for `/bin/*vim`.
- non-standalone mode, when `name == "vim"`, means the user expects a
normal vim package that uses the specified configuration. In this case
we include everything in the original derivation, with wrappers for
all the executables that accept a vimrc.
Browser Integration requires setgid and setuid programs, which needs to be done in the system configuration.
This is cleaner than the ad-hoc ways we have to set things up for platforms without a global configuration file.
* nixos/earlyoom: bring the module up to date
Removes deprecated option `ignoreOOMScoreAdjust`, introduces `killHook`
as a replacement for `notificationsCommand`, and adds an `extraArgs`
option for things not covered by the module.
* nixos/earlyoom: add nixos test
* nixos/earlyoom: add reportInterval
Allows setting the interval for logging a memory report. Defaults to
3600 following upstream
(https://github.com/rfjakob/earlyoom/blob/master/earlyoom.default#L5)
to avoid flooding logs.
* nixos/earlyoom: add free{Mem,Swap}KillThreshold
Fixes https://github.com/NixOS/nixpkgs/issues/83504
When a NixOS system uses flakes, i.e., /etc/nixos/flake.nix exists, it
is impossible to use nixos-rebuild to build a pre-flake
configuration.nix. Of course, one can directly use nix command to
build the configuration, but not everybody remembers the correct nix
options to do that.
With the new option, it is possible to build a pre-flake configuration
with command like this:
nixos-rebuild build-vm -I nixos-config=./vm.nix --no-flake
The option might be useful for people following older pre-flake
tutorials on a flake-based system.
In https://github.com/NixOS/nixpkgs/pull/142747, the implementation
behind Machine.execute() has been changed to pipe all the command's
output into base64 on the guest machine.
Unfortunately this means that base64 is blocking until stdout is closed,
which in turn means that we now need to make sure that whenever we run a
program in background via "&" we also need to make sure to close stdout,
which we do by redirecting stdout to stderr.
Signed-off-by: aszlig <aszlig@nix.build>
This adds the option `networking.wg-quick.interfaces.<name>.autostart`, which defaults to `true`, which is the previous behavior. With this option set to `false`, the systemd-unit will no longer be set to `wantedBy = [ "multi-user.target" ]` and therefore the tunnel has to be enabled/disabled via `systemctl start/stop wg-quick-<name>`.
Co-authored-by: pennae <82953136+pennae@users.noreply.github.com>
Naively deduplicate VLANs in the python driver for NixOS tests. The
current implementation accidentally works, since the VLan class mutates
the environment. On construction it sets QEMU_VDE_SOCKET_${id} and this
environment variable gets overwritten once a second VLAN with the same
id is constructed. Because the NIC flags passed to qemu just use the
QEMU_VDE_SOCKET_${id} environment variable, this implicitly chooses a
single vde_switch process for each VLAN.
However, this leads to unusable vde_switch processes being spawned in
each test run and as a side effect makes it impossible to access the
correct VLan objects in the interactive test driver. It also makes it
remarkably hard to understand why the current implementation ever
worked.
