Restarting user@ instances is bad because it causes all user services
(such as ssh-agent.service) to be restarted. Maybe one day we can have
switch-to-configuration restart user units in a fine-grained way, but
for now we should just ignore user systemd instances.
Backport: 14.04
This now provides a handful of different grsecurity kernels for slightly
different 'flavors' of packages. This doesn't change the grsecurity
module to use them just yet, however.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
AppArmor only requires a few patches to the 3.2 and 3.4 kernels in order
to work properly (with the minor catch grsecurity -stable includes the
3.2 patches.) This adds them to the kernel builds by default, removes
features.apparmor (since it's always true) and makes it the default MAC
system.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
"nixos-install --chroot" runs a command (by default a login shell) in
a chroot inside the NixOS installation in /mnt. This might useful for
poking around a new installation.
By setting a line like
MACVLANS="eno1"
in /etc/containers/<name>.conf, the container will get an Ethernet
interface named mv-eno1, which represents an additional MAC address on
the physical eno1 interface. Thus the container has direct access to
the physical network. You can specify multiple interfaces in MACVLANS.
Unfortunately, you can't do this with wireless interfaces.
Note that dhcpcd is disabled in containers by default, so you'll
probably want to set
networking.useDHCP = true;
in the container, or configure a static IP address.
To do: add a containers.* option for this, and a flag for
"nixos-container create".
This allows you to use the Linux kernel's built-in compressed memory as
swap space functionality.
It is recommended to enable only for kernel 3.14 (which is when zram came out of
the staging drivers area) or higher.
Personal information management application that provides integrated mail,
calendaring and address book functionality
https://wiki.gnome.org/Apps/Evolution
This overhauls the Datadog module a bit to be much more useful. In
particular, it adds support for nginx and postgresql monitoring
integrations to dd-agent. These have to exist in separate files under
/etc/dd-agent, so the module just exposes then as separate options. In
the future, more integrations could be added this way.
In the process of doing this, I also had to rename the dd-agent user to
datadog. Note the UIDs did not change, so this is strictly backwards
compatible. The reason for this is to make it easier to create a
'datadog' postgres user with access to pg_stats, as 'dd-agent' typically
isn't a valid username. This allows the out of the box configurations to
be used.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
If /boot is a btrfs subvolume, it will be on a different device than /
but not be at the root from grub's perspective. This should be fixed in
a nicer way by #2449, but that can't go into 14.04.
Dmidecode fails in our EFI test with the error "SMBIOS entry point
missing". But we don't need dmidecode because we have already have
systemd-detect-virt.
NixOS has a pervasive dependency on bash. For instance, the X11
session script sources /etc/profile to get a reasonable
environment. Thus we should not provide an option to disable bash.
Also, enabling zsh no longer sets ‘users.defaultUserShell’ to zsh, to
prevent a collision with bash's definition of the same
option. (Changing the default shell is also something that should be
left to the user.)
Previously all card-specific stuff was scattered across xserver.nix
and opengl.nix, which is ugly. Now it can be kept together in a single
card-specific module. This required the addition of a few internal
options:
- services.xserver.drivers: A list of { name, driverName, modules,
libPath } sets.
- hardware.opengl.package: The OpenGL implementation. Note that there
can be only one OpenGL implementation at a time in a system
configuration (i.e. no dynamic detection).
- hardware.opengl.package32: The 32-bit OpenGL implementation.
Fixes#2379.
The new name was a misnomer because the values really are X11 video
drivers (e.g. ‘cirrus’ or ‘nvidia’), not OpenGL implementations. That
it's also used to set an OpenGL implementation for kmscon is just
confusing overloading.
The Tarsnap module is now far more flexible, allowing individual
archives with individual options to be specified at will, allowing
granular backup schedules, etc.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
This reverts commit 6eaced3582. Doesn't
work very well, e.g. if you actually have the FUSE module loaded. And
in any case it's already fixed in NixOps.
Otherwise, when switching from systemd 203 to 212, you get errors like:
Failed to stop remote-fs.target: Bad message
Failed to stop systemd-udevd-control.socket: Bad message
...
These fail to mount if you don't have the appropriate kernel support,
and this confuses NixOps' ‘check’ command. We should teach NixOps not
to complain about non-essential mount points, but in the meantime it's
better to turn them off.
By default, socat only waits 0.5s for the remote side to finish after
getting EOF on the local side. So don't close the local side, instead
wait for socat to exit when the remote side finishes.
http://hydra.nixos.org/build/10663282
This fixes several problems in the dhcpcd service:
* A segfault during startup, due to a race with udev (dhcpcd would get
an ADD event from udev, causing it to re-add an interface that it
already had, leading to a segfault later on).
* A hang/segfault processing "dhcpcd rebind" (which NixOS calls after
waking up from suspend).
Also, add "lo" to the list of ignored interfaces. It usually ignores
"lo", but apparently not when it gets an ADD event from udev.
Upstream has not been tagging new versions for a long time, but we need
compatibility with newer kernels. The 0.6.2 versions already have a bunch of
backported compatibility patches, but 3.14 kernels need even more.
Also, the git versions have fixed a bunch of crashes and other bugs, so perhaps
we should just bite the bullet and just use recent git versions (as sometimes
upstream recommends, when people run into bugs).
This adds a new "boot.zfs.useGit" boolean option, so that a user can
easily opt into using the git versions.
By enabling ‘services.openssh.startWhenNeeded’, sshd is started
on-demand by systemd using socket activation. This is particularly
useful if you have a zillion containers and don't want to have sshd
running permanently. Note that socket activation is not noticeable
slower, contrary to what the manpage for ‘sshd -i’ says, so we might
want to make this the default one day.
This causes OpenVPN services to reach the "active" state when the VPN
connection is up (i.e., after OpenVPN prints "Initialization Sequence
Completed"). This allows units to be ordered correctly after openvpn-*
units, and makes systemctl present a password prompt:
$ start openvpn-foo
Enter Private Key Password: *************
(I first tried to implement this by calling "systemd-notify --ready"
from the "up" script, but systemd-notify is not reliable.)
This seems to have combined badly with the systemd upgrade, we'll revert
for now and revisit after the 14.04 branch.
This reverts commit ad80532881, reversing
changes made to 1c5d3c7883.
The ability for unprivileged users to mount external media is useful
regardless of the desktop environment. Also, since udisks2 is
activated on-demand, it doesn't add any overhead if you're not using it.
Apparently systemd is now smart enough to figure out predictable names
for QEMU network interfaces. But since our tests expect them to be
named eth0/eth1..., this is not desirable at the moment.
http://hydra.nixos.org/build/10418789
This used to work with systemd-nspawn 203, because it bind-mounted
/etc/resolv.conf (so openresolv couldn't overwrite it). Now it's just
copied, so we need some special handling.
This led to the container test failing, which made no sense
whatsoever, until I realized nix-daemon.socket creates the socket
directory as a side effect, which systemd-nspawn then bind-mounts.
http://hydra.nixos.org/build/10397575
This has some advantages:
* You get ssh-agent regardless of how you logged in. Previously it was
only started for X11 sessions.
* All sessions of a user share the same agent. So if you added a key
on tty1, it will also be available on tty2.
* Systemd will restart ssh-agent if it dies.
* $SSH_AUTH_SOCK now points to the /run/user/<uid> directory, which is
more secure than /tmp.
For bonus points, we should patch ssh-agent to support socket-based
activation...
With ‘systemd.user.units’ and ‘systemd.user.services’, you can specify
units used by per-user systemd instances. For example,
systemd.user.services.foo =
{ description = "foo";
wantedBy = [ "default.target" ];
serviceConfig.ExecStart = "${pkgs.foo}/bin/foo";
};
declares a unit ‘foo.service’ that gets started automatically when the
user systemd instance starts, and is stopped when the user systemd
instance stops.
Note that there is at most one systemd instance per user: it's created
when a user logs in and there is no systemd instance for that user
yet, and it's removed when the user fully logs out (i.e. has no
sessions anymore). So if you're simultaneously logged in via X11 and a
virtual console, you get only one copy of foo.
If you define a unit, and either systemd or a package in
systemd.packages already provides that unit, then we now generate a
file /etc/systemd/system/<unit>.d/overrides.conf. This makes it
possible to use upstream units, while allowing them to be customised
from the NixOS configuration. For instance, the module nix-daemon.nix
now uses the units provided by the Nix package. And all unit
definitions that duplicated upstream systemd units are finally gone.
This makes the baseUnit option unnecessary, so I've removed it.
This allows specifying rules for systemd-tmpfiles.
Also, enable systemd-tmpfiles-clean.timer so that stuff is cleaned up
automatically 15 minutes after boot and every day, *if* you have the
appropriate cleanup rules (which we don't have by default).
mkdir -m will only set the permissions if it *creates* the directory.
Existing directories, with possibly wrong permissions, will not be
updated.
Use explicit chmod so permissions will always be correct.
The preStart snippets (graphite, carbon) try to create directories under
/var/db/. That currently fails because the code is run as user
"graphite". Fix by setting "PermissionsStartOnly = true" so that the
preStart stuff is run as 'root'.
Further:
* graphite-web-0.9.12/bin/build-index.sh needs perl, so add it to PATH.
* Now that preStart runs as root, we must wait with "chown graphite"
until we're done creating files/directories.
* Drop needless check for root (uid 0) before running chown.
This creates static device nodes such as /dev/fuse or
/dev/snd/seq. The kernel modules for these devices will be loaded on
demand when the device node is opened.
This prevents insidious errors once systemd begins handling the unit. If
the unit is loaded at boot, any errors of this nature are logged to the
console before the journal service is running. This makes it very hard
to diagnose the issue. Therefore, this assertion helps guarantee the
mistake is not made.
Note that systemd no longer depends on dbus, so we're rid of the
cyclic dependency problem between systemd and dbus.
This commit incorporates from wkennington's systemd branch
(203dcff45002a63f6be75c65f1017021318cc839,
1f842558a95947261ece66f707bfa24faf5a9d88).
Using pkgs.lib.mkOverride in a sysctl option would throw a bogus error.
Also, if you defined a sysctl multiple times in the same configuration,
only one of the values would be picked up, while the others were silently
discarded.
This patch should fix both issues. If you define a sysctl multiple
times at your highest defined priority level, you will get a proper
error with detailed location information.
This mostly upgrades transmission, and does some very minor touchups on
AppArmor support.
In particular, there is now no need to ever specify the umask as part of
the settings, as it will be mixed in by default (which is essentially
always what you want). Also, the default configuration is now more
sensible: Downloads are put in /var/lib/transmission/Downloads, and
incomplete files are put in /var/lib/transmission/.incomplete - this
also allows easy use of file syncing probrams, like BitTorrent Sync.
Finally, this unconditionally enables the AppArmor profiles for the
daemon, if AppArmor is enabled - rather than letting the user specify
profile support, it's best to default to supporting profiles for daemons
transparently in all places.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Previously, if the currently installed Nix is too old to evaluate
Nixpkgs, then nixos-rebuild would fail and the user had to upgrade Nix
manually. Now, as a fallback, we run ‘nix-store -r’ to obtain a binary
Nix directly from the binary cache.
This allows doing any necessary actions that were not in the installed
nixos-rebuild (such as downloading a new version of Nix). This does
require us to be careful that nixos-rebuild is backwards-compatible
(i.e. can run in any old installation).
The style for IDs dictates that groups/users should have the same ID -
so if a user doesn't have a group or vice versa, then we should skip
that ID.
In this case, we had already assigned grsecurity GID 121, but I
accidentally also assigned Hydra UID 121. Instead, let's assign Hydra
UID 122. And also assign a GID (122) as well.
Luckily nobody was depending on this yet (except me).
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Using pkgs.lib on the spine of module evaluation is problematic
because the pkgs argument depends on the result of module
evaluation. To prevent an infinite recursion, pkgs and some of the
modules are evaluated twice, which is inefficient. Using ‘with lib’
prevents this problem.
Enabling by default on gnome3 as now it's possible to create and use
accounts (tested with telepathy_gabble and gtalk).
At this time, empathy x86-64 fails to build on hydra but I'm unable
to reproduce. Therefore, try disabling the parallel build.
Previously we were setting GRKERNSEC_PROC_USER y, which was a little bit
too strict. It doesn't allow a special group (e.g. the grsecurity group
users) to access /proc information - this requires
GRKERNSEC_PROC_USERGROUP y, and the two are mutually exclusive.
This was also not in line with the default automatic grsecurity
configuration - it actually defaults to USERGROUP (although it has a
default GID of 1001 instead of ours), not USER.
This introduces a new option restrictProcWithGroup - enabled by default
- which turns on GRKERNSEC_PROC_USERGROUP instead. It also turns off
restrictProc by default and makes sure both cannot be enabled.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
This module implements a significant refactoring in grsecurity
configuration for NixOS, making it far more usable by default and much
easier to configure.
- New security.grsecurity NixOS attributes.
- All grsec kernels supported
- Allows default 'auto' grsec configuration, or custom config
- Supports custom kernel options through kernelExtraConfig
- Defaults to high-security - user must choose kernel, server/desktop
mode, and any virtualisation software. That's all.
- kptr_restrict is fixed under grsecurity (it's unwriteable)
- grsecurity patch creation is now significantly abstracted
- only need revision, version, and SHA1
- kernel version requirements are asserted for sanity
- built kernels can have the uname specify the exact grsec version
for development or bug reports. Off by default (requires
`security.grsecurity.config.verboseVersion = true;`)
- grsecurity sysctl support
- By default, disabled.
- For people who enable it, NixOS deploys a 'grsec-lock' systemd
service which runs at startup. You are expected to configure sysctl
through NixOS like you regularly would, which will occur before the
service is started. As a result, changing sysctl settings requires
a reboot.
- New default group: 'grsecurity'
- Root is a member by default
- GRKERNSEC_PROC_GID is implicitly set to the 'grsecurity' GID,
making it possible to easily add users to this group for /proc
access
- AppArmor is now automatically enabled where it wasn't before, despite
implying features.apparmor = true
The most trivial example of enabling grsecurity in your kernel is by
specifying:
security.grsecurity.enable = true;
security.grsecurity.testing = true; # testing 3.13 kernel
security.grsecurity.config.system = "desktop"; # or "server"
This specifies absolutely no virtualisation support. In general, you
probably at least want KVM host support, which is a little more work.
So:
security.grsecurity.enable = true;
security.grsecurity.stable = true; # enable stable 3.2 kernel
security.grsecurity.config = {
system = "server";
priority = "security";
virtualisationConfig = "host";
virtualisationSoftware = "kvm";
hardwareVirtualisation = true;
}
This module has primarily been tested on Hetzner EX40 & VQ7 servers
using NixOps.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
This prevents errors like "Another app is currently holding the
xtables lock" if the firewall and NAT services are starting in
parallel. (Longer term, we should probably move to a single service
for managing the iptables rules.)
