EC2 instances don't have a console, so it's pointless to start
emergency mode if a mount fails. (This happened to me with an
encrypted filesystem where the key wasn't sent on time using "charon
send-keys".) Better to cross fingers and continue booting.
This is necessary to prevent a race. Udev 197 has a new naming scheme
for network devices, so it will rename (say) eth0 to eno0. This fails
with "error changing net interface name eth0 to eno1: Device or
resource busy" if another process has opened the interface in the
meantime.
This reverts commit 1e741f1572b6793b861e2f9820015475ce339ae0 as it is
unnecessary according to @edolstra, because services.xserver.config from another
module will be merged into the configuration.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This is currently only a very simple implementation which just recurses a list
of heads that get chained together to the right of the corresponding previous
item of the list.
If I forgot about something in the already existing configuration options,
please let me know or if this commit is useless or a duplicate, feel free to
revert. But by looking at implementation before this commit, I only see zaphod
and/or quirky xinerama-like configuration options.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Starting an authenticated root shell is a security hole, so don't do
it by default. The kernel command line parameter
‘initrd.shell_on_fail’ restores the original. (Of course, this only
improves security if you have a password on GRUB to prevent the kernel
command line from being edited by unauthorized users.)
The 'memtest86' package didn't work on any of my machines. 'memtest86plus', on
the other hand, seems to work just fine. Does anyone know why we keep the
seemingly older version around still?
This is especially useful if you want to supply a default XRandR configuration,
where you need multiple "Monitor" sections in order to set properties for
specific CRTCs (if not running in zaphod mode).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The "S|s|single" option is handled by systemd (starting
rescue.target). And the rescue target basically removes the need for
a special debug shell. (Also, there is "systemd.crash_shell=1" for
starting a shell if systemd crashes.)
Also, symlink kbrequest.target to rescue.target as suggested by the
systemd.special manpage. This way, you can start a sulogin rescue
shell by pressing Alt+Up.
Restarting <interface>-cfg.service may cause the interface's IP
addresses to be flushed. If the default gateway goes through that
interface, then the default gateway is deleted. So we need to
restart network-setup.target.
This allows setting the max tcp window size for the route of
the default gateway (usually the internet access).
It works only for non-DHCP configurations by now.
To be honest, it's more like "be less discriminating against USB tablets".
USB tablets usually get autodetected, device name is not necessary and defaulting to a serial touchscreen is a clear discrimination.
Unconditionally remapping buttons is generally not a good idea either.
Old defaults transformed into examples.
During a configuration switch, changed units are stopped in the old
configuration, then started in the new configuration (i.e. after
running the activation script and running "systemctl daemon-reload").
This ensures that services are stopped using the ExecStop/ExecStopPost
commands from the old configuration.
However, for some services it's undesirable to stop them; in
particular dhcpcd, which deconfigures its network interfaces when it
stops. This is dangerous when doing remote upgrades - usually things
go right (especially because the switch script ignores SIGHUP), but
not always (see 9aa69885f0). Likewise,
sshd should be kept running for as long as possible to prevent a
lock-out if the switch fails.
So the new option ‘stopIfChanged = false’ causes "systemctl restart"
to be used instead of "systemctl stop" followed by "systemctl start".
This is only proper for services that don't have stop commands. (And
it might not handle dependencies properly in some cases, but I'm not
sure.)
Running it from systemd rather than cron has several advantages:
systemd ensures that only one instance runs at a time; the GC can be
manually started/stopped; and logging goes to the journal.
We still need cron to start the service at the right time, but
hopefully soon we can get rid of cron entirely (once systemd supports
starting a unit at a specific time).
This is mainly useful for specifying mounts that depend on other
units. For example sshfs or davfs need network (and possibly
nameservices).
While systemd makes a distinction between local and remote
filesystems, this only works for in-kernel filesystems such as
nfs and cifs.
fuse-based filesystems (such as sshfs and davs) are classified as
local, so they fail without networking. By explicitly declaring these
mounts as full systemd units (as opposed to having systemd generate
them automatically from /etc/fstab), dependencies can be specified as
on every other unit.
In the future, we can probably port NixOS' filesystems handling to use
these native systemd.mount units and skip /etc/fstab altogether, but
this probably requires additional changes, such as starting systemd
even earlier during boot (stage 1).
I'm not any good at perl, and I only came up with this after many
slow attempts. Any review welcome.
But until this, memtest was broken, and extraPrepareConfig as well, in grub.
Having all services with DefaultDependencies=yes depend on
local-fs.target is annoying, because some of those services might be
necessary to mount local filesystems. For instance, Charon's
send-keys feature requires sshd to be running in order to receive LUKS
encryption keys, which in turn requires dhcpcd, and so on. So we drop
this dependency (and swap.target as well for consistency). If
services require a specific mount, they should use RequiresMountsFor
in any case.
Charon needs this to include the dynamically generated
/root/.vbox-charon-client-key. (We used
users.extraUsers.root.openssh.authorizedKeys.keyFiles for this, but
that no longer works.)
Instead of the somewhat hacky script that inserted public keys
into the users' .ssh/authorized_keys files, use the AuthorizedKeysFile
configuration directive in sshd_config and generate extra key
files for each user (placed in /etc/authorized_keys.d/).
This has rendered my system unbootable, because I forgot to enable AUTOFS4 in my
custom kernel. In addition to AUTOFS4, this includes (hopefully) all other
kernel features needed by systemd, as listed in the README:
REQUIREMENTS:
Linux kernel >= 2.6.39
with devtmpfs
with cgroups (but it's OK to disable all controllers)
optional but strongly recommended: autofs4, ipv6
Autofs4 is not a requirement here, but in our case it turns out that the system
is not able to boot properly with a LUKS-enabled system (or at least not on _my_
system).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The xsession script runs services that depend on a sane environment. Gpg-agent, for
example, runs the program "pinentry-gtk-2" to obtain the password to unlock GnuPG
and SSH keys. That program will display only gibberish unless $FONTCONFIG_FILE is
configured properly. Instead of configuring these variables explicitly one by one,
we just source /etc/profile, which contains the appropriate @shellInit@ code.
This is needed in order to properly lock your screen using the C-a C-x
(lockscreen) command _and_ being back to re-login, because the "other" PAM
service/fallback is to deny authentication.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Thus
networking.interfaces = [ { name = "eth0"; ipAddress = "192.168.15.1"; } ];
can now be written as
networking.interfaces.eth0.ipAddress = "192.168.15.1";
The old notation still works though.
Running agetty on ttyS0 interferes with the backdoor, which uses ttyS0
as its standard error. After agetty starts, writes to the stderr file
descriptor will return EIO (though doing "exec 2>/proc/self/fd/2" will
miracuously fix this).
http://hydra.nixos.org/build/3252782
Cgroups are handled by systemd now. Systemd's cgroup support does not
do all the things that cgrulesengd does, but they're likely to
interact poorly with each other.
‘systemd-vconsole-setup’ by default operates on /dev/tty0, the
currently active tty. Since it puts /dev/tty0 in Unicode or ASCII
mode, if the X server is currently active when it runs, keys such as
Alt-F4 won't reach the X server anymore. So use /dev/tty1 instead.
We now just have a simple attribute called "version24" which replaces all those
pesky versionOlder that were spreading throughout the file and makes things way
more readable.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
By default the path is determined related to ServerRoot. Unfortunately
ServerRoot is pointing to the Nix store and the web server can't write to it.
We now create a directory called "runtime" withen the stateDir and point
DefaultRuntimeDir to it.
For more information on the DefaultRuntimeDir directive, please see:
http://httpd.apache.org/docs/2.4/mod/core.html#defaultruntimedir
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
NameVirtualHost no longer has any effect on version 2.4 and just emits ugly
warnings, so let's not use it if we use 2.4.
More information: http://httpd.apache.org/docs/2.4/upgrading.html#misc
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The Order/Deny directives are deprecated in version 2.4, so we're going to
define two wrappers for allDenied and allGranted in order to properly generate
configurations for both version 2.2 and 2.4.
For more information an access control changes, see:
http://httpd.apache.org/docs/2.4/upgrading.html#access
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Now, MPMs can be loaded at runtime and it's no longer required to compile in one
of the MPM modules statically. So, if version is >= 2.4, load the MPM module
corresponding to the multiProcessingModule value of the service module.
For details, please see: http://httpd.apache.org/docs/2.4/mpm.html
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Beginning with version 2.3, the authn were refactored. As a result, authn_alias
is now part of the new module authn_core, so let's use authn_core instead of
authn_alias.
For details please see: http://httpd.apache.org/docs/2.4/upgrading.html#misc
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
The new configuration.nix option 'environment.enableBashCompletion'
determines whether bash completion is automatically enabled system-wide
for all interactive shells or not. The default setting is 'off'.
The new configuration.nix option 'environment.enableBashCompletion'
determines whether bash completion is automatically enabled system-wide
for all interactive shells or not. The default setting is 'off'.
My main reason for adding this is the ability to turn off helpers
altogether. If you are not using any of the special protocols, keeping
them turned off is safest, and in case you do want to use them, it's
best to configure them through the new CT target for your network
topology. Perhaps some sane defaults for nixos can be examined in the
future.
This change has no impact if you don't touch the added options, so no
need to adapt.
This is meant to replace /proc/sys/net/ipv4/conf/*/rp_filter, which
only works for ipv4. Furthermore, it's nicer to handle this kind of
filtering in the firewall.
There are some more subtle differences, please see:
https://home.regit.org/netfilter-en/secure-use-of-helpers/
I chose to enable this by default (when the firewall is enabled) as
it's a good idea in general. Only people with advanced routing needs
might not want this, but I guess they don't use the nixos firewall
anyway and use a custom solution. Furthermore, the option only becomes
available in kernel 3.3+, so conservative nixos users that just stick
to the default kernel will not need to act now just yet.
The triggers are just arbitrary strings that are included in the unit
under X-Restart-Triggers. The idea is that if they change between
reconfigurations, switch-to-configuration will restart the unit
because its store path changed. This is mostly useful for services
that implicitly depend on generated files in /etc. Thus you can say
restartTriggers = [ confFile ];
where ‘confFile’ is the derivation that generated the /etc file in
question.
Flushing is bad if the Nix store is on a remote filesystem accessed
over that interface.
http://hydra.nixos.org/build/3184162
Also added a interface option ‘prefixLength’ as a better alternative
to ‘subnetMask’.
Instead it's enough to depend on
sys-subsystem-net-devices-<interface>.device, which in turn has a
"wants" dependency on the service (if any) that creates the interface.
For each statically configured interface, we now create a unit
‘<interface>-cfg.service’ which gets started as soon as the network
device comes up. Similarly, each bridge defined in
‘networking.bridges’ and virtual interface in ‘networking.interfaces’
is created by a service ‘<interface>.service’.
So if we have
networking.bridges.br0.interfaces = [ "eth0" "eth1" ];
networking.interfaces =
[ { name = "br0";
ipAddress = "192.168.1.1";
}
];
then there will be a unit ‘br0.service’ that depends on
‘sys-subsystem-net-devices-eth0.device’ and
‘sys-subsystem-net-devices-eth1.device’, and a unit ‘br0-cfg.service’
that depends on ‘sys-subsystem-net-devices-br0.device’.
The special handling for cronjobs should probably move to the cron
module (logcheckIgnore = bool option) in the future, as it's more
natural to just declare a cronjob, and mark it as "log-ignored",
instead of adding cronjobs through logcheck.
But as systemCronjobs is not an attrset yet (just simple strings),
this would require adding an attrset for cronjobs or parsing strings
in the nix language to get hold of the cron-user and command.
So for now, I keep the interface within logcheck's module.
However SLIM is still broken and you have to create a
/usr/lib/dri/fglrx_dri.so symlink pointing to
/run/opengl-driver/lib/fglrx_dri.so
At least fgl_glxgears shows 10 times more frames per second now
This is broken because it requires restarting applications to see new
NSS modules. The proper way to handle NSS modules is through nscd.
See commit 554ae9908b.
The upower daemon needs the gdbus command (which is weird given that
upower links against dbus_glib, but ah well...). This fixes suspend
in KDE with systemd.
Alsa-utils provides a udev rule to restore volume settings, so use
that instead of restoring them from a systemd service. The
"alsa-store" service saves the settings on shutdown.
It's not enough to say "after udev-settle.service" since
udev-settle.service is not wanted/required anywhere - we need to say
"wants udev-settle.service" as well.
This should fix problems with ALSA and X11 initialisation that people
have been seeing.
So instead of:
boot.systemd.services."foo".serviceConfig =
''
StartLimitInterval=10
CPUShare=500
'';
you can say:
boot.systemd.services."foo".serviceConfig.StartLimitInterval = 10;
boot.systemd.services."foo".serviceConfig.CPUShare = 500;
This way all unit options are available and users can set/override
options in configuration.nix.
This makes it easier for systemd to track it and avoids race conditions such as
this one:
systemd[1]: PID file /run/sshd.pid not readable (yet?) after start.
systemd[1]: Failed to start SSH Daemon.
systemd[1]: Unit sshd.service entered failed state.
systemd[1]: sshd.service holdoff time over, scheduling restart.
systemd[1]: Stopping SSH Daemon...
systemd[1]: Starting SSH Daemon...
sshd[2315]: Server listening on 0.0.0.0 port 22.
sshd[2315]: Server listening on :: port 22.
sshd[2335]: error: Bind to port 22 on 0.0.0.0 failed: Address already in use.
sshd[2335]: error: Bind to port 22 on :: failed: Address already in use.
sshd[2335]: fatal: Cannot bind any address.
systemd[1]: Started SSH Daemon.
When spamd isn't running as 'root', it cannot access the usual ~/.spamassassin
path where user-specific files normally reside. Instead, we use the path
/var/lib/spamassassin-<user> to store those home directories.
