Fixes race conditions like this:
> systemd[1]: Started prometheus-kea-exporter.service.
> kea-exporter[927]: Listening on http://0.0.0.0:9547
> kea-exporter[927]: Socket at /run/kea/dhcp4.sock does not exist. Is Kea running?
> systemd[1]: prometheus-kea-exporter.service: Main process exited, code=exited, status=1/FAILURE
It doesn't make sense to have a default value for this that's
incompatible with the default locate implementation. It means that
just doing services.locate.enable = true; generates a warning, even if
you don't care about pruning anything. So only use the default prune
list if the locate implementation supports it (i.e., isn't findutils).
If `services.tor.client.enable` is set to false (the default), the `SOCKSPort` option is not added to the torrc file but since Tor defaults to listening on port 9050 when the option is not specified, the tor client is not actually disabled. To fix this, simply set `SOCKSPort` to 0, which disables the client.
Use `mkForce` to prevent potentially two different `SOCKSPort` options in the torrc file, with one of them being 0 as this would cause Tor to fail to start. When `services.tor.client.enable` is set to false, this should always be disabled.
When `services.resolved.enable` is set to true, the file /etc/resolv.conf becomes a symlink to /etc/static/resolv.conf, which is a symlink to /run/systemd/resolve/stub-resolv.conf. Without this commit, tor does not have access to this file thanks to systemd confinement. This results in the following warning when tor starts:
```
[warn] Unable to stat resolver configuration in '/etc/resolv.conf': No such file or directory
[warn] Could not read your DNS config from '/etc/resolv.conf' - please investigate your DNS configuration. This is possibly a problem. Meanwhile, falling back to local DNS at 127.0.0.1.
```
To fix this, simply allow read-only access to the file when resolved is in use.
According to https://github.com/NixOS/nixpkgs/pull/161818#discussion_r824820462, the symlink may also point to /run/systemd/resolve/resolv.conf, so allow that as well.
Use a quoted heredoc to inject installBootLoader safely into the script,
and restore the previous invocation of `system` with a single argument so
that shell commands keep working.
Without this fix, evaluating a NixOS configuration with Tomcat enabled and the
default settings results in the following evaluation error:
Failed assertions:
- users.users.tomcat.group is unset. This used to default to
nogroup, but this is unsafe. For example you can create a group
for this user with:
users.users.tomcat.group = "tomcat";
users.groups.tomcat = {};
As a novice to using this module, I found the existing description to be
quite misleading. It does not at all disable pulling from the registry,
it just loads some image archive that may or may not be related to the
container you're specifying. I had thought there was extra magic behind
this option, but it's just a `docker load`. You need foreknowledge of
the contents of the archive so that whatever it contained is actually
used to run the container.
I've reworded the description to hopefully make this behavior clearer.
it's really easy to accidentally write the wrong systemd Exec* directive, ones
that works most of the time but fails when users include systemd metacharacters
in arguments that are interpolated into an Exec* directive. add a few functions
analogous to escapeShellArg{,s} and some documentation on how and when to use them.
This bug is so obscure and unlikely that I was honestly not able to
properly write a test for it. What happens is that we are calling
handleModifiedUnit() with $unitsToStart=\%unitsToRestart. We do this to
make sure that the unit is stopped before it's started again which is
not possible by regular means because the stop phase is already done
when calling the activation script.
recordUnit() still gets $startListFile, however which is the wrong file.
The bug would be triggered if an activation script requests a service
restart for a service that has `stopIfChanged = true` and
switch-to-configuration is killed before the restart phase was run. If
the script is run again, but the activation script is not requesting
more restarts, the unit would be started instead of restarted.
The current logic assumes that everything that isn't a derivation is a
store path, but it can also be something that's *coercible* to a store
path, like a flake input.
Unnecessary uses of `lib.toDerivation` result in errors in pure evaluation
mode when `builtins.storePath` is disabled.
Also document what a `package` is.
We spent a whole afternoon debugging this, because upstream has very
bad software quality and the error messages were incredibly
misleading.
So let’s document it for the sanity of other people.
Btw, I think the implementation of our module is pretty brittle,
especially the part about diffing tokens to check whether they
changed. We should rather just request a new builder registration
every time, it’s not that much overhead, and always set `replace` so
it is idempotent.
Follow-up on https://github.com/NixOS/nixpkgs/pull/161426.
Explain why having legacy iptables rules installed can lead to confusing
firewall behaviour, and provide some guidance on how to fix this.
Some systems should not be rebooted at just any time. If the upgrade process takes too long, for instance because of a
slow internet connection, or if the upgrade service is ran during production hours, we want to allow to define a window
outside of which a reboot will not be performed.
The system will then reboot on the next run of the upgrade service which finishes inside the reboot window.
E.g. we can run the update service twice per week, once during the night and once during the day, but reboots are only
allowed during the night. By doing so, a system that is usually shut down during the night will still receive updates
and systems that are turned on 24/7 can be rebooted outside of production hours.
Co-authored-by: Silvan Mosberger <github@infinisil.com>
zsh-autosuggestions supports having fallback strategies expressed
through the ZSH_AUTOSUGGEST_STRATEGY array. For example,
`ZSH_AUTOSUGGEST_STRATEGY=(history completion)`. We should also support
this.
For systems without internet connections, it doesn't make sense to
require the existence of an /etc/resolv.conf file to disable
resolvconf, so let's expose networking.resolveconf.enable as a public
option that can be set to false.
The cntr sometimes hangs until the 10-hour hydra limit. This behaviour
appears to be an edge-case related to the type of TTY in which the cntr
command runs during test execution. We can work around this by running
the command as a background job.
I additionally added a wait_for_open_port to fix nondeterministic test
failures I observed after fixing the hanging issue.
In issue #157787 @martined wrote:
Trying to use confinement on packages providing their systemd units
with systemd.packages, for example mpd, fails with the following
error:
system-units> ln: failed to create symbolic link
'/nix/store/...-system-units/mpd.service': File exists
This is because systemd-confinement and mpd both provide a mpd.service
file through systemd.packages. (mpd got updated that way recently to
use upstream's service file)
To address this, we now place the unit file containing the bind-mounted
paths of the Nix closure into a drop-in directory instead of using the
name of a unit file directly.
This does come with the implication that the options set in the drop-in
directory won't apply if the main unit file is missing. In practice
however this should not happen for two reasons:
* The systemd-confinement module already sets additional options via
systemd.services and thus we should get a main unit file
* In the unlikely event that we don't get a main unit file regardless
of the previous point, the unit would be a no-op even if the options
of the drop-in directory would apply
Another thing to consider is the order in which those options are
merged, since systemd loads the files from the drop-in directory in
alphabetical order. So given that we have confinement.conf and
overrides.conf, the confinement options are loaded before the NixOS
overrides.
Since we're only setting the BindReadOnlyPaths option, the order isn't
that important since all those paths are merged anyway and we still
don't lose the ability to reset the option since overrides.conf comes
afterwards.
Fixes: https://github.com/NixOS/nixpkgs/issues/157787
Signed-off-by: aszlig <aszlig@nix.build>
This type correctly merges multiple option types together while also
annotating them with file information. In a future commit this will be
used for `_module.freeformType`
Makes all options rendered in the manual throw an error if they don't
have a type specified.
This is a follow-up to #76184
Co-Authored-By: Silvan Mosberger <contact@infinisil.com>
This fixes the following issues with the database provisioning script
included in the services.keycloak module:
- It lacked permission to access the DB password file specified in the
module option 'services.keycloak.database.passwordFile'.
- It prevented Keycloak from starting after the second time if the user
chose MySQL for the database.
Update version to 1.4.231.
Build 231 points to a specific commit from the 1.4.x branch adding many
fixes and improvements. Since this version is an unofficial release, add
an unstable prefix to the version string in Nixpkgs.
Signed-off-by: Felix Singer <felixsinger@posteo.net>
Signed-off-by: Franz Pletz <fpletz@fnordicwalking.de>
Currently the test-watch.service gets started in a loop as long as
/testpath exists, so `rm /testpath /testpath-modified` runs into a race
condition where if the service was just getting activated, it will
create /testpath-modified and make the test fail.
This is fixed by making the service RemainAfterExit so that it only
starts once, and stopping it manually after we remove /testpath.
logrotate.timer is enough for rotating logs. Enabling logrotate.service would
make the service start on every configuration switch, leading to tests failure when
logrotate is enabled.
Also update test to make sure the timer is active and runs the service
on date change.
https://github.com/ipfs/fs-repo-migrations/releases/tag/v2.0.2
This is pretty much a complete rewrite of the ipfs-migrator package.
In version 2.0.0 a major change was made to the way the migrator works. Before, there was one binary that contained every migration. Now every migration has its own binary. If fs-repo-migrations can't find a required binary in the PATH, it will download it off the internet. To prevent that, build every migration individually, symlink them all into one package and then wrap fs-repo-migrations so it finds the package with all the migrations.
The change to the IPFS NixOS module and the IPFS package is needed because without explicitly specifying a repo version to migrate to, fs-repo-migrations will query the internet to find the latest version. This fails in the sandbox, for example when testing the ipfs passthru tests.
While it may seem like the repoVersion and IPFS version are in sync and the code could be simplified, this is not the case. See https://github.com/ipfs/fs-repo-migrations#when-should-i-migrate for a table with the IPFS versions and corresponding repo versions.
Go 1.17 breaks the migrations, so use Go 1.16 instead. This is also the Go version used in their CI, see 3dc218e300/.github/workflows/test.yml (L4). See https://github.com/ipfs/fs-repo-migrations/pull/140#issuecomment-982715907 for a previous mention of this issue. The issue manifests itself when doing anything with a migration, for example `fs-repo-11-to-12 --help`:
```
panic: qtls.ClientHelloInfo doesn't match
goroutine 1 [running]:
github.com/marten-seemann/qtls-go1-15.init.0()
github.com/marten-seemann/qtls-go1-15@v0.1.1/unsafe.go:20 +0x132
```
Also add myself as a maintainer for this package.
This fixes the test failure discovered in https://github.com/NixOS/nixpkgs/pull/160914.
See https://github.com/ipfs/fs-repo-migrations/issues/148 to read some of my struggles with updating this package.
The argument parser used by snapserver behaves differntly for optional
arguments with existing defaults. In such cases, the standalone argument
name is a valid input and a following value is interpreted as a
positional argument. Therefore the argument and the value must be
provided as a single argument seperated by equals sign.
The test was looking at the wrong interface and relying on silly
behaviour by the dummy driver, which autocreated a `dummy0` interface on
modprobe.
Fix this by making it look at the actual `foo` interface that the test
creates.
With the previous change that enabled error propagation through
`inherit_errexit`, the script would fail if `errexit` was set, but
`inherit_errexit` was not. This is due to `shopt -p` exiting with an
error if the option is disabled. To work around this, use the exit
code instead of the text value returned by `shopt -p`.
Fixes#160869.
This change probably wasn't documented sufficiently in the release
notes, neither the fact systemd stopped using iptables on its own in
case of nf_tables support.
Fixes#156041.
Previously the bonding driver would create an initial `bond0` interface
when it was loaded. If the network management integration used that
interface and did not recreate it, it was stuck to the default
`balance-rr` mode.
Deploying systemds modprobe.d configuration sets `max_bonds=0`, so we
don't run into that issue anymore.
Hence we now make sure that we can indeed create `bond0` with `802.3ad`
(LACP), which is a non default mode.
Overriding can now happen using module options, which is preferred
because it is more discoverable and doesn't require knowledge of
overrides in the first place.
While the documentation said to set this to null, in case an imperative
config was supposed to be used, this was not possible with the typing in
place.
As `nix-daemon.service` does not make use of `ExecStop`, we prefer
to keep the socket up and available. This is important for machines
that run Nix-based services, such as automated build, test, and deploy
services, that expect the daemon socket to be available at all times.
See committed inline comment for further explanation.
As explained in vpnc's Makefile, a vpnc with OpenSSL support is
non-redistributable. The option to enable OpenSSL support, which is
disabled by default, is even called OPENSSL_GPL_VIOLATION — something
that was conveniently hidden by the strange way the option was set in
the previous version of this package.
Not entirely sure when it got broken this time, but when creating a VM
network with `nixos-build-vms(8)`, there are should be the following scripts:
* `$out/bin/nixos-test-driver` which drops into an interactive shell to
interactively perform test steps.
* `$out/bin/nixos-run-vms` which non-interactively starts the VMs from
the network so that one can manually play around in the VM.
The latter also starts an interactive shell for a while now which means
that it does the exact same thing as `nixos-test-driver` which is not
its purpose.
