This is useful for adding extra functionality or defaults to _every_
nixos evaluation.
My use case is overriding behaviour for all nixos tests, for example
setting packageOverrides to newer versions and changing some default
dependencies/settings.
By making this accessible through an environment variable, this can now
be fully accomplished externally. No more need to fork
nixos/nixpkgs (which becomes a maintenance burden), just use the channel
instead and plug in via this envvar.
The warning was displayed whenever services.virtualboxHost.enable was
true, but if people were to enable hardening, they'd still get that
annoying message.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Creates unnecessary cruft in the root users home directory, which we
really don't need. Except the log, but therefore we now cat the log to
stderr and the private temporary directory is cleaned up afterwards.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The current options for the XServer produce a huge amount of log messages. The
server produces around 70-80 messages per minute. The most messages look like
this:
display-manager-start[1846]: GetModeLine - scrn: 0 clock: 75200
display-manager-start[1846]: GetModeLine - hdsp: 1366 hbeg: 1414 hend: 1478 httl: 1582
display-manager-start[1846]: vdsp: 768 vbeg: 772 vend: 779 vttl: 792 flags: 9
Since theses messages aren't very useful, I propose to remove the `-logverbose`
and `-verbose` options from the XServer arguments.
This should display a big fat warning that people can hardly miss until
we have fixed the issues with the host-only-interfaces that persist when
hardining is enabled.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Essentially adds two more VirtualBox VMs to the test and also increases
the memory size of the qemu VM to 768 MB to make sure we don't run out
of memory too soon.
We're testing whether those two VMs can talk to either each other
(currently via ICMP only) or to/from the host via TCP/IP.
Also, this restructures the VM test a bit, so that we now pass in a
custom stage2Init script that has access to the store via a private
mount over the /nix/store that's already in the initrd. The reason why
this is a private mount is that we don't want to shadow the Nix store of
the initrd, essentially breaking cleanup functionality after the custom
stage 2 script (currently this is only "poweroff -f").
Note that setting the hostname inside the VirtualBox VM is *not* for
additional fanciness but to produce a different store path for the VM
image, so that VirtualBox doesn't bail out when trying to use an image
which is already attached to another VM.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Hardening mode in VirtualBox is quite restrictive and on some systems it
could make sense to disable hardening mode, especially while we still
have issues with hostonly networking and other issues[TM] we don't know
or haven't tested yet.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
We're going to create more than one VirtualBox VM, so let's dynamically
generate subs specific to a particular VirtualBox VM, merging everything
into the testScript and machine expressions.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
It turns out that installing therubytracer, with dependency on old v8, even
when using source libv8 version is problematic.
(see
http://stackoverflow.com/questions/21666379/problems-installing-gitlab-on-odroid-v8-lib-not-available).
But wait, rails does not even need therubytracer, just any kind of javascript
server side execution framework like nodejs. Well just use that, as also
suggested from different internet sources (look link above), it works just
fine.
Currently it pretty much tests starting up virtual machines and just
shutting down afterwards, but for both VBoxManage and the VirtualBox
GUI.
This helps catching errors in hardened mode, however we still need to
test whether networking works the way intended (and I fear that this is
broken at the moment).
The VirtualBox VM is _not_ using hardware virtualization support (thus
we use system = "i686-linux", because x86_64 has no emulation support),
because we're already within a qemu VM, which means it's going to be
slow as hell (that's why I've written own subs just for testing
startup/shutdown/whatnot with respective timeouts).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
We only need to have setuid-root wrappers for VBox{Headless,SDL} and
VirtualBox, otherwise VBoxManage will run as root and NOT drop
privileges!
Fixes#5283.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
I had to make several adjustments to make it work with nixos:
* Replace relative config file lookups with ENV variable.
* Modify gitlab-shell to not clear then environment when running
pre-receive.
* Modify gitlab-shell to write some environment variables into
the .authorized_keys file to make sure gitlab-shell reads the
correct config file.
* Log unicorn output to syslog.
I tried various ways of adding a syslog package but the bundler would
not pick them up. Please fix in a better way if possible.
* Gitlab-runner program wrapper.
This is useful to run e.g. backups etc. with the correct
environment set up.
Since we're using HTTPS for the binary cache (introduced in faf0797) by
default, the binary cache should also be available during installation.
The file that is defined in SSL_CERT_FILE outside of the chroot is
copied over to /tmp/ca-cert.crt inside the chroot, so we have an
absolute path we can reference during nixos-install. However, this might
end up with the file not being cleaned up properly from outside of the
store, but neither would be /tmp/root so the cleanup issue needs to be
solved in another place (or commit to be more exact).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The current nixos module for VirtualBox unconditionally configures a vboxnet0
network interface at boot. This may be undesired, especially when the user wants
to manage network interfaces in a centralized manner.
- Create container nixos profile
- Create lxc-container nixos config using container nixos profile
- Docker nixos image, use nixos profile for its base config
The default configuration installed the Bitstream Vera fonts, but DejaVu
is a superior replacement, and the default Fontconfig settings need it
now for the generic faces monospace, sans-serif, and serif.
Details:
* The option `fonts.fontconfig.ultimate.enable` can be used to disable
the fontconfig-ultimate configuration.
* The user-configurable options provided by fontconfig-ultimate are
exposed in the NixOS module: `allowBitmaps` (default: true),
`allowType1` (default: false), `useEmbeddedBitmaps` (default: false),
`forceAutohint` (default: false), `renderMonoTTFAsBitmap` (default:
false).
* Upstream provides three substitution modes for substituting TrueType
fonts for Type 1 fonts (which do not render well). The default,
"free", substitutes free fonts for Type 1 fonts. The option "ms"
substitutions Microsoft fonts for Type 1 fonts. The option "combi"
uses a combination of Microsoft and free fonts. Substitutions can also
be disabled.
* All 21 of the Infinality rendering modes supported by fontconfig-ultimate
or by the original Infinality distribution can be selected through
`fonts.fontconfig.ultimate.rendering`. The default is the medium style
provided by fontconfig-ultimate. Any of the modes may be customized,
or Infinality rendering can be disabled entirely.
Details:
* The option `fonts.enableFontConfig` has (finally) been renamed
`fonts.fontconfig.enable`.
* Configurations are loaded in this order: first the Fontconfig-upstream
configuration is loaded, then the NixOS-specific font directories are
set, the system-wide default configuration is loaded, and finally the
user configuration is loaded (if enabled).
* The NixOS options `fonts.fontconfig.defaultFonts.monospace`,
`fonts.fontconfig.defaultFonts.sansSerif` and
`fonts.fontconfig.defaultFonts.serif` are added to allow setting the
default system-wide font used for these generic faces. The defaults
are the appropriate faces from the DejaVu collection because of their
comprehensive Unicode coverage, clean rendering, and excellent
legibility.
* The NixOS option `fonts.fontconfig.antialias` can be used to disable
antialiasing (it is enabled by default).
* The options `fonts.fontconfig.subpixel.rgba` and
`fonts.fontconfig.subpixel.lcdfilter` control the system-wide default
settings for subpixel order and LCD filtering algorithm,
respectively.
* `fonts.fontconfig.hinting.enable` can be used to disable TrueType font
hinting (it is enabled by default).
`fonts.fontconfig.hinting.autohint` controls the FreeType autohinter.
`fonts.fontconfig.hinting.style` controls the hint style; it is "full"
by default.
* User configurations can be disabled system-wide by setting
`fonts.fontconfig.includeUserConf = false`. They are enabled by
default so users can set Fontconfig options in the desktop environment
of their choice.
This overhauls the Tor module in a few ways:
- Uses systemd service files, including hardening/config checks
- Removed old privoxy support; users should use the Tor Browser
instead.
- Remove 'fast' circuit/SOCKS port; most users don't care (and it adds
added complexity and confusion)
- Added support for bandwidth accounting
- Removed old relay listenAddress option; taken over by portSpec
- Formatting, description, code cleanups.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Rather than trying to override the 'torsocks' executable in $PATH, the
new module instead properly configures `/etc/tor/torsocks.conf` and puts
the normal `torsocks` executable in $PATH so it can work out of the box.
As a bonus, I think this module actually works now, because the torsocks
configuration has changed a lot from when this was written, it seems...
Signed-off-by: Austin Seipp <aseipp@pobox.com>
'torify' now ships with the tor bundle itself; and using torsocks is
recommended over tsocks (torify will use torsocks automatically.)
Signed-off-by: Austin Seipp <aseipp@pobox.com>
From http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/
You disable the assignment of fixed names, so that the unpredictable
kernel names are used again. For this, simply mask udev's rule file for
the default policy: ln -s /dev/null
/etc/udev/rules.d/80-net-setup-link.rules (since v209: this file was
called 80-net-name-slot.rules in release v197 through v208)
This patch should be reverted if either:
- systemd fixes the multi-swapon issue.
https://bugs.freedesktop.org/show_bug.cgi?id=86930
- If we disable the autogeneration of swap and vfat units within
systemd.
From http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/
You disable the assignment of fixed names, so that the unpredictable
kernel names are used again. For this, simply mask udev's rule file for
the default policy: ln -s /dev/null
/etc/udev/rules.d/80-net-setup-link.rules (since v209: this file was
called 80-net-name-slot.rules in release v197 through v208)
Following the discussion NixOS#5021:
- obsolete the nix.proxy option
- add the networking.proxy option
- open a default no_proxy environment variable
- add a rsync option
- Manual tests ok.
- Automatic tests ok.
Amended by lethalman to simplify the option descriptions.
Using primusrun will work as expected in a multilib environment. Even if the initial program
executes a antoehr program of the another architecture. Assuming the program does not modify
LD_LIBRARY_PATH inappropriately.
This does not update virtualgl for seemless multilib. I was unable to get a mixed 64/32 bit
environment to work with VirtualGL. The mechanism VirtualGL uses to inject the fake GL library would
fail if both 32bit and 64 bit libraries were in the environment. Instead the bumblebee package
creates a optirun32 executable that can be used to run a 32bit executable with optimus on a 64 bit
host. This is not created if the host is 32bit.
For my usage, gaming under wine, the primusrun executable works as expected regardless of
32bit/64bit.
VirtualBox with hardening support requires the main binaries to be
setuid root. Using VBOX_WITH_RUNPATH, we ensure that the RPATHs are
pointing to the libexec directory and we also need to unset
VBOX_WITH_ORIGIN to make sure that the build system is actually setting
those RPATHs.
The hardened.patch implements two things:
* Set the binary directory to the setuid-wrappers dir so that
VboxSVC calls them instead of the binaries from the store path. The
reason behind this is because nothing in the Nix store can have the
setuid flag.
* Excempt /nix/store from the group permission check, because while it
is group-writeable indeed it also has the sticky bit set (and also
the whole store is mounted read-only on most NixOS systems), so we're
checking on that as well.
Right now, the hardened.patch uses /nix/store and /var/setuid-wrappers
directly, so someone would ever want to change those on a NixOS system,
please provide a patch to set those paths on build time. However, for
simplicity, it's best to do it when we _really_ need it.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
We will simply rename the previous module and add a warning whenever the
module is included directly, pointing the user to the right option and
also enable it as well (in case somebody has missed the option and is
wondering why VirtualBox doesn't work anymore).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Especially new users could be confused by this, so we're now marking
services.virtualbox.enable as obsolete and defaulting to
services.virtualboxGuest.enable instead. I believe this now makes it
clear, that this option is for guest additions only.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This is needed when /etc/resolv.conf is being overriden by networkd
and other configurations. If the file is destroyed by an environment
activation then it must be rebuilt so that applications which interface
with /etc/resolv.conf directly don't break.
There currently are collisions between the main CUPS package and the
filters package, which are:
* $storepath/share/cups/banners/classified
* $storepath/share/cups/banners/confidential
* $storepath/share/cups/banners/secret
* $storepath/share/cups/banners/standard
* $storepath/share/cups/banners/topsecret
* $storepath/share/cups/banners/unclassified
* $storepath/share/cups/data/testprint
And they actually have different content, so let's ignore those for now
until we have a better fix.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Of course, this could be done via packageOverrides, but this is more
explicit and makes it possible to run the tests with various Chromium
overrides.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Currently, the test is only for testing the user namespace sandbox and
even that isn't very representative, because we're running the tests as
root.
But apart from that, we should have functionality for opening/closing
windows and the main goal here is to get them as deterministic as
possible, because Chromium usually isn't very nice to chained xdotool
keystrokes.
And of course, the most important "test" we have here: We know at least
whether Chromium works _at_all_.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The NixOS manual says modules have the following signature:
{ config, lib, pkgs, ... }:
But our generated configuration.nix file lacks the 'lib' part. Add it.
The host id value gets generated by reading a 32-bit value from
/dev/urandom.
This makes programs that rely on a correct host id more reliable.
It also makes using ZFS more seamless, as you don't need to configure
the hostId manually; instead, it becomes part of your config from the
moment you install NixOS.
The old boot.spl.hostid option was not working correctly due to an
upstream bug.
Instead, now we will create the /etc/hostid file so that all applications
(including the ZFS kernel modules, ZFS user-space applications and other
unrelated programs) pick-up the same system-wide host id. Note that glibc
(and by extension, the `hostid` program) also respect the host id configured in
/etc/hostid, if it exists.
The hostid option is now mandatory when using ZFS because otherwise, ZFS will
require you to force-import your ZFS pools if you want to use them, which is
undesirable because it disables some of the checks that ZFS does to make sure it
is safe to import a ZFS pool.
The /etc/hostid file must also exist when booting the initrd, before the SPL
kernel module is loaded, so that ZFS picks up the hostid correctly.
The complexity in creating the /etc/hostid file is due to having to
write the host ID as a 32-bit binary value, taking into account the
endianness of the machine, while using only shell commands and/or simple
utilities (to avoid exploding the size of the initrd).
It turns out that the upstream systemd services that import ZFS pools contain
serious bugs. The first major problem is that importing pools fails if there
are no pools to import. The second major problem is that if a pool ends up in
/etc/zfs/zpool.cache but it disappears from the system (e.g. if you
reboot but during the reboot you unplug your ZFS-formatted USB pen drive),
then the import service will always fail and it will be impossible to get rid
of the pool from the cache (unless you manually delete the cache).
Also, the upstream service would always import all available ZFS pools every
boot, which may not be what is desired in some cases.
This commit will solve these problems in the following ways:
1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of
issues, and also does not play well with NixOS's philosophy of
reproducible configurations. Instead, on every boot NixOS will try to import
the set of pools that are specified in its configuration. This is also the
direction that upstream is moving towards.
2. Instead of trying to import all ZFS pools, only import those that are
actually necessary. NixOS will automatically determine these from the
config.fileSystems.* option. Also, the user can import any additional
pools every boot by adding them to the config.boot.zfs.extraPools
option, but this is only necessary if their filesystems are not
specified in config.fileSystems.*.
3. Added options to configure if ZFS should force-import ZFS pools. This may
currently be necessary, especially if your pools have not been correctly
imported with a proper host id configuration (which is probably true for 99% of
current NixOS ZFS users). Once host id configuration becomes mandatory when
using ZFS in NixOS and we are sure that most users have updated their
configurations and rebooted at least once, we should disable force-import by
default. Probably, this shouldn't be done before the next stable release.
WARNING: This commit may change the order in which your non-ZFS vs ZFS
filesystems are mounted. To avoid this problem (now or in the future)
it is recommended that you set the 'mountpoint' property of your ZFS
filesystems to 'legacy', and that you manage them using
config.fileSystems, just like any other non-ZFS filesystem is usually
managed in NixOS.
Also remove custom zfs services from NixOS. This makes NixOS more aligned with
upstream.
More importantly, it prepares the way for NixOS to use ZED (the ZFS event
daemon). This service will automatically be enabled but it is not possible to
configure it via configuration.nix yet.
The dnscrypt-proxy service relays regular DNS queries to
a DNSCrypt enabled upstream resolver.
The traffic between the client and the upstream resolver is
encrypted and authenticated, which may mitigate the risk of
MITM attacks and third-party snooping (assuming a trustworthy
upstream).
Though dnscrypt-proxy can run as a standalone DNS client,
the recommended setup is to use it as a forwarder for a
caching DNS client.
To use dnscrypt-proxy as a forwarder for dnsmasq, do
```nix
{
# ...
networking.nameservers = [ "127.0.0.1" ];
networking.dhcpcd.extraConfig = "nohook resolv.conf";
services.dnscrypt-proxy.enable = true;
services.dnscrypt-proxy.localAddress = "127.0.0.1";
services.dnscrypt-proxy.port = 40;
services.dnsmasq.enable = true;
services.dnsmasq.extraConfig = ''
no-resolv
server=127.0.0.1#40
listen-address=127.0.0.1
'';
# ...
}
```
Perl seems to write the file in latin1 independent of the actual input
encoding. This can corrupt the "description" field of /etc/passwd. By
setting "binmode" to ":utf8" Perl can be forced to write UTF-8. Ideally
the program would simply read/write the fields by value without any
changes in encoding. However, assuming/enforcing UTF-8 is a lot better
than using an obsolete coding like latin1.
