Systemd provides an option for allocating DynamicUsers
which we want to use in NixOS to harden service configuration.
However, we discovered that the user wasn't allocated properly
for services. After some digging this turned out to be, of course,
a cache inconsistency problem.
When a DynamicUser creation is performed, Systemd check beforehand
whether the requested user already exists statically. If it does,
it bails out. If it doesn't, systemd continues with allocating the
user.
However, by checking whether the user exists, nscd will store
the fact that the user does not exist in it's negative cache.
When the service tries to lookup what user is associated to its
uid (By calling whoami, for example), it will try to consult
libnss_systemd.so However this will read from the cache and tell
report that the user doesn't exist, and thus will return that
there is no user associated with the uid. It will continue
to do so for the cache duration time. If the service
doesn't immediately looks up its username, this bug is not
triggered, as the cache will be invalidated around this time.
However, if the service is quick enough, it might end up
in a situation where it's incorrectly reported that the
user doesn't exist.
Preferably, we would not be using nscd at all. But we need to
use it because glibc reads nss modules from /etc/nsswitch.conf
by looking relative to the global LD_LIBRARY_PATH. Because LD_LIBRARY_PATH
is not set globally (as that would lead to impurities and ABI issues),
glibc will fail to find any nss modules.
Instead, as a hack, we start up nscd with LD_LIBRARY_PATH set
for only that service. Glibc will forward all nss syscalls to
nscd, which will then respect the LD_LIBRARY_PATH and only
read from locations specified in the NixOS config.
we can load nss modules in a pure fashion.
However, I think by accident, we just copied over the default
settings of nscd, which actually caches user and group lookups.
We already disable this when sssd is enabled, as this interferes
with the correct working of libnss_sss.so as it already
does its own caching of LDAP requests.
(See https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/deployment_guide/usingnscd-sssd)
Because nscd caching is now also interferring with libnss_systemd.so
and probably also with other nsss modules, lets just pre-emptively
disable caching for now for all options related to users and groups,
but keep it for caching hosts ans services lookups.
Note that we can not just put in /etc/nscd.conf:
enable-cache passwd no
As this will actually cause glibc to _not_ forward the call to nscd
at all, and thus never reach the nss modules. Instead we set
the negative and positive cache ttls to 0 seconds as a workaround.
This way, Glibc will always forward requests to nscd, but results
will never be cached.
Fixes#50273
Instead of setting User/Group only when DynamicUser is disabled, the
previous version of the code set it only when it was enabled. This
caused services with DynamicUser enabled to actually run as nobody, and
services without DynamicUser enabled to run as root.
Regression from fbb7e0c82f.
This cleans up the CockroachDB expression, with a few suggestions from
@aszlig.
However, it brought up the note of using systemd's StateDirectory=
directive, which is a nice feature for managing long-term data files,
especially for UID/GID assigned services. However, it can only manage
directories under /var/lib (for global services), so it has to introduce
a special path to make use of it at all in the case someone wants a path
at a different root.
While the dataDir directive at the NixOS level is _occasionally_ useful,
I've gone ahead and removed it for now, as this expression is so new,
and it makes the expression cleaner, while other kinks can be worked out
and people can test drive it.
CockroachDB's dataDir directive, instead, has been replaced with
systemd's StateDirectory management to place the data under
/var/lib/cockroachdb for all uses.
There's an included RequiresMountsFor= clause like usual though, so if
people want dependencies for any kind of mounted device at boot
time/before database startup, it's easy to specify using their own
mount/filesystems clause.
This can also be reverted if necessary, but, we can see if anyone ever
actually wants that later on before doing it -- it's a backwards
compatible change, anyway.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Currently there are two calls to curl in the reloadScript, neither which
check for errors. If something is misconfigured (like wrong authToken),
the only trace that something wrong happened is this log message:
Asking Jenkins to reload config
<h1>Bad Message 400</h1><pre>reason: Illegal character VCHAR='<'</pre>
The service isn't marked as failed, so it's easy to miss.
Fix it by passing --fail to curl.
While at it:
* Add $curl_opts and $jenkins_url variables to keep the curl command
lines DRY.
* Add --show-error to curl to show short error message explanation when
things go wrong (like HTTP 401 error).
* Lower-case the $CRUMB variable as upper case is for exported environment
variables.
The new behaviour, when having wrong accessToken:
Asking Jenkins to reload config
curl: (22) The requested URL returned error: 401
And the service is clearly marked as failed in `systemctl --failed`.
This also includes a full end-to-end CockroachDB clustering test to
ensure everything basically works. However, this test is not currently
enabled by default, though it can be run manually. See the included
comments in the test for more information.
Closes#51306. Closes#38665.
Co-authored-by: Austin Seipp <aseipp@pobox.com>
Signed-off-by: Austin Seipp <aseipp@pobox.com>
As the comment notes, restarts/exits of dhcpcd generally require
restarting the NTP service since, if name resolution fails for a pool of
servers, the service might break itself. To be on the safe side, try
restarting Chrony in these instances, too.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Setting the server list to be empty is useful e.g. for hardware-only
or virtualized reference clocks that are passed through to the system
directly. In this case, initstepslew has no effect, so don't emit it.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
