by adding targets and curl wait loops to services to ensure services
are not started before their depended services are reachable.
Extra targets cfssl-online.target and kube-apiserver-online.target
syncronize starts across machines and node-online.target ensures
docker is restarted and ready to deploy containers on after flannel
has discussed the network cidr with apiserver.
Since flannel needs to be started before addon-manager to configure
the docker interface, it has to have its own rbac bootstrap service.
The curl wait loops within the other services exists to ensure that when
starting the service it is able to do its work immediately without
clobbering the log about failing conditions.
By ensuring kubernetes.target is only reached after starting the
cluster it can be used in the tests as a wait condition.
In kube-certmgr-bootstrap mkdir is needed for it to not fail to start.
The following is the relevant part of systemctl list-dependencies
default.target
● ├─certmgr.service
● ├─cfssl.service
● ├─docker.service
● ├─etcd.service
● ├─flannel.service
● ├─kubernetes.target
● │ ├─kube-addon-manager.service
● │ ├─kube-proxy.service
● │ ├─kube-apiserver-online.target
● │ │ ├─flannel-rbac-bootstrap.service
● │ │ ├─kube-apiserver-online.service
● │ │ ├─kube-apiserver.service
● │ │ ├─kube-controller-manager.service
● │ │ └─kube-scheduler.service
● │ └─node-online.target
● │ ├─node-online.service
● │ ├─flannel.target
● │ │ ├─flannel.service
● │ │ └─mk-docker-opts.service
● │ └─kubelet.target
● │ └─kubelet.service
● ├─network-online.target
● │ └─cfssl-online.target
● │ ├─certmgr.service
● │ ├─cfssl-online.service
● │ └─kube-certmgr-bootstrap.service
to protect services from crashing and clobbering the logs when
certificates are not in place yet and make sure services are activated
when certificates are ready.
To prevent errors similar to "kube-controller-manager.path: Failed to
enter waiting state: Too many open files"
fs.inotify.max_user_instances has to be increased.
+ isolate etcd on the master node by letting it listen only on loopback
+ enabling kubelet on master and taint master with NoSchedule
The reason for the latter is that flannel requires all nodes to be "registered"
in the cluster in order to setup the cluster network. This means that the
kubelet is needed even at nodes on which we don't plan to schedule anything.
- All kubernetes components have been seperated into different files
- All TLS-enabled ports have been deprecated and disabled by default
- EasyCert option added to support automatic cluster PKI-bootstrap
- RBAC has been enforced for all cluster components by default
- NixOS kubernetes test cases make use of easyCerts to setup PKI
The `| tee` invocation always masked the return value of the
switch-to-configuration test.
```
~ $ false | tee && echo "oh no"
oh no
```
The added wrapper script will still output everything to stderr, while
passing failures to the test harness.
trace: warning: config.services.gitea.database.password will be stored as plaintext
in the Nix store. Use database.passwordFile instead.
(Arguably, this shouldn't be a warning at all. But making it happy is
easier than having a debate on the value of this warning.)
trace: warning: The options services.ndppd.interface and services.ndppd.network will probably be removed soon,
please use services.ndppd.proxies.<interface>.rules.<network> instead.
trace: warning: The option `services.rspamd.bindUISocket' defined in `<unknown-file>' has been renamed to `services.rspamd.workers.controller.bindSockets'.
trace: warning: The option `services.rspamd.bindSocket' defined in `<unknown-file>' has been renamed to `services.rspamd.workers.normal.bindSockets'.
trace: warning: The option `services.rspamd.workers.”rspamd_proxy".type` defined in `<unknown-file>' has enum value `proxy` which has been renamed to `rspamd_proxy`
With this option it's possible to specify a custom expression for
`roundcube`, i.e. a roundcube environment with third-party plugins as
shown in the testcase.
* pr-55320:
nixos/release-notes: mention breaking changes with matrix-synapse update
nixos/matrix-synapse: reload service with SIGHUP
nixos/tests/matrix-synapse: generate ca and certificates
nixos/matrix-synapse: use python to launch synapse
pythonPackages.pymacaroons-pynacl: remove unmaintained fork
matrix-synapse: 0.34.1.1 -> 0.99.0
pythonPackages.pymacaroons: init at 0.13.0
Hydra should support multiple Nix versions (and currently contains fixes
to work with Nix 2.0 and higher).
Further Nix versions can be added to the `hydraPkgs` expression in the
test case which lists all supported Nix versions for Hydra.
* redmine: 3.4.8 -> 4.0.1
* nixos/redmine: update nixos test to run against both redmine 3.x and 4.x series
* nixos/redmine: default new installs from 19.03 onward to redmine 4.x series, while keeping existing installs on redmine 3.x series
* nixos/redmine: add comment about default redmine package to 19.03 release notes
* redmine: add aandersea as a maintainer
This is just a set of globs to remove from the active plugins directory
after autoconfiguration is complete.
I also removed the hard-coded disabling of "diskstats", since it seems
to work just fine now.
This allows the VM to provide a `configuration.nix` file to the VM.
The test doesn't work in sandbox because it needs Internet (however it
works interactively).
The Openstack metadata service exposes the EC2 API. We use the
existing `ec2.nix` module to configure the hostname and ssh keys of an
Openstack Instance.
A test checks the ssh server is well configured.
This is mainly to reduce the size of the image (700MB). Also,
declarative features provided by cloud-init are not really useful
since we would prefer to use our `configuration.nix` file instead.
Don't add the testing "webcam" device,
which is unexpected to see when querying
what devices fwupd believes exist :).
Won't change behavior for anyone defining
the blacklistPlugin option already,
but doesn't seem worth making more complicated.
postgis: cleanup
Another part of https://github.com/NixOS/nixpkgs/pull/38698, though I did cleanup even more.
Moving docs to separate output should save another 30MB.
I did pin poppler to 0.61 just to be sure GDAL doesn't break again next
time poppler changes internal APIs.
* postgresql: reorganize package and it's extensions
Extracts some useful parts of https://github.com/NixOS/nixpkgs/pull/38698,
in particular, it's vision that postgresql plugins should be namespaced.
He prefers to contribute to his own nixpkgs fork triton.
Since he is still marked as maintainer in many packages
this leaves the wrong impression he still maintains those.
For large setups it is useful to list all databases explicit
(for example if temporary databases are also present) and store them in extra
files.
For smaller setups it is more convenient to just backup all databases at once,
because it is easy to forget to update configuration when adding/renaming
databases. pg_dumpall also has the advantage that it backups users/passwords.
As a result the module becomes easier to use because it is sufficient
in the default case to just set one option (services.postgresqlBackup.enable).
Although this can be added to `extraOptions` I figured that it makes
sense to add an option to explicitly promote this feature in our
documentation since most of the self-hosted gitea instances won't be
intended for common use I guess.
Also added a notice that this should be added after the initial deploy
as you have to register yourself using that feature unless the install
wizard is used.
Nexus increased their default minimum disk space requirement to 4GB:
```
com.orientechnologies.orient.core.exception.OLowDiskSpaceException: Error occurred while executing a
write operation to database 'OSystem' due to limited free space on the disk (1823 MB). The database
is now working in read-only mode. Please close the database (or stop OrientDB), make room on your hard
drive and then reopen the database. The minimal required space is 4096 MB. Required space is now
set to 4096MB (you can change it by setting parameter storage.diskCache.diskFreeSpaceLimit) .
server# [ 72.560866] zqnav3mg7m6ixvdcacgj7p5ibijpibx5-unit-script-nexus-start[627]: DB name="OSystem"
```
Including the rest on the VM 8GB should be the most suitable solution.
As the installer test also takes 8GB of disk size this should still be
in an acceptable range.
According to systemd-nspawn(1), --network-bridge implies --network-veth,
and --port option is supported only when private networking is enabled.
Fixes#52417.
Introduces the option security.protectKernelImage that is intended to control
various mitigations to protect the integrity of the running kernel
image (i.e., prevent replacing it without rebooting).
This makes sense as a dedicated module as it is otherwise somewhat difficult
to override for hardened profile users who want e.g., hibernation to work.
Although the package itself builds fine, the module fails because it
tries to log into a non-existant file in `/var/log` which breaks the
service. Patching to default config to log to stdout by default fixes
the issue. Additionally this is the better solution as NixOS heavily
relies on systemd (and thus journald) for logging.
Also, the runtime relies on `/etc/localtime` to start, as it's not
required by the module system we set UTC as sensitive default when using
the module.
To ensure that the service's basic functionality is available, a simple
NixOS test has been added.
pkgs.owncloud still pointed to owncloud 7.0.15 (from May 13 2016)
Last owncloud server update in nixpkgs was in Jun 2016.
At the same time Nextcloud forked away from it, indicating users
switched over to that.
cc @matej (original maintainer)
The intention of the previous change was to move krb5-config to .dev (it
gives the locations of headers), but it grabbed all of the user-facing
binaries too. This puts them back.
Allow switching out kerberos server implementation.
Sharing config is probably sensible, but implementation is different enough to
be worth splitting into two files. Not sure this is the correct way to split an
implementation, but it works for now.
Uses the switch from config.krb5 to select implementation.
They consistently fail since openjdk bump with some out-of-space errors.
That's not a problem by itself, but each test instance ties a build slot
for many hours and consequently they also delay channels as those wait
for all builds to finish.
Feel free to re-enable when fixed, of course.
The test now runs wayland, which means we can no longer use X11 style testing.
Instead we get gnome shell to execute javascript through its dbus interface.
Since 83b27f60ce, the tests were moved
into all-tests.nix and some of the tooling has changed so that
subattributes of test expressions are now recursively evaluated until a
derivation with a .test attribute has been found.
Unfortunately this isn't the case for all of the tests and the
runInMachine doesn't use the makeTest function other tests are using but
instead uses runInMachine, which doesn't generate a .test attribute.
Whener a .test attribute wasn't found by the new handleTest function, it
recurses down again until there is no value left that is an attribute
set and subsequently returns its unchanged value. This however has the
drawback that instead of getting different attributes for each
architecture we only get the last architecture in the supportedSystems
list.
In the case of the release.nix, the last architecture in
supportedSystems is "aarch64-linux", so the runInMachine test is always
built on that architecture.
In order to work around this, I changed runInMachine to emit a .test
attribute so that it looks to handleTest like it was a test created via
makeTest.
Signed-off-by: aszlig <aszlig@nix.build>
Docker images used to be, essentially, a linked list of layers. Each
layer would have a tarball and a json document pointing to its parent,
and the image pointed to the top layer:
imageA ----> layerA
|
v
layerB
|
v
layerC
The current image spec changed this format to where the Image defined
the order and set of layers:
imageA ---> layerA
|--> layerB
`--> layerC
For backwards compatibility, docker produces images which follow both
specs: layers point to parents, and images also point to the entire
list:
imageA ---> layerA
| |
| v
|--> layerB
| |
| v
`--> layerC
This is nice for tooling which supported the older version and never
updated to support the newer format.
Our `buildImage` code only supported the old version, so in order for
`buildImage` to properly generate an image based on another image
with `fromImage`, the parent image's layers must fully support the old
mechanism.
This is not a problem in general, but is a problem with
`buildLayeredImage`.
`buildLayeredImage` creates images with newer image spec, because
individual store paths don't have a guaranteed parent layer. Including
a specific parent ID in the layer's json makes the output less likely
to cache hit when published or pulled.
This means until now, `buildLayeredImage` could not be the input to
`buildImage`.
The changes in this PR change `buildImage` to only use the layer's
manifest when locating parent IDs. This does break buildImage on
extremely old Docker images, though I do wonder how many of these
exist.
This work has been sponsored by Target.
GitLab 11.5.1 dropped the dependency to posix_spawn, which is broken on
32bit. (See https://gitlab.com/gitlab-org/gitlab-ce/issues/53525)
The only part missing is decreasing virtualisation.memorySize to
something that a 32 bit qemu still executes.
The maximum seems to be 2047, and tests passed with that value for me.
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>
When reworking the rspamd workers I disallowed `proxy` as a type and
instead used `rspamd_proxy` which is the correct name for that worker
type. That change breaks peoples existing config and so I have made this
commit which allows `proxy` as a worker type again but makes it behave
as `rspamd_proxy` and prints a warning if you use it.
This commit adds an assertion that checks that either `configFile` or
`configuration` is configured for alertmanager. The alertmanager config
can not be an empty attributeset. The check executed with `amtool` fails
before the service even has the chance to start. We should probably not
allow a broken alertmanager configuration anyway.
This also introduces a test for alertmanager configuration that piggy
backs on the existing prometheus tests.
The nixos test is a bit misleading, as the given nginx configuration
would always cause gitlab to redirect to localhost, which is clearly not
what you want in a production setup.
Instead we now enable services.nginx.recommendedProxySettings,
curl against http://gitlab, and assure we get redirected to that same
hostname, too.
previously one of the tests would fail with
boot-after-install: starting vm
Path length (109) is longer than maximum supported length (108) and will be truncated at /nix/store/0ingn8cwwnl84i374hcl6nafsm2c5m2p-perl-5.28.0/lib/perl5/5.28.0/x86_64-linux-thread-multi/Socket.pm line 872, <__ANONIO__> line 268.
boot-after-install# qemu-system-x86_64: -monitor unix:./monitor: Failed to connect socket ./monitor: No such file or directory
error: QEMU died prematurely
QEMU died prematurely
vde_switch: EOF on stdin, cleaning up and exiting
cleaning up
builder for '/nix/store/zbpxwwxwy7f6z5z3kg4nf5mjqsywzjvx-vm-test-run-installer-filesystemEncryptedWithKeyfile.drv' failed with exit code 4
Shortening the test name fixes the issue.
This will make the list much easier to re-use, eg. for `nixosTests`
The drawback is that this approaches makes the
```
nix-build release.nix -A tests.opensmtpd.x86_64-linux
```
command about twice as slow (3s to 6s): it now has to evaluate `nixpkgs`
once for each architecture, instead of just having the hardcoded list of
tests that allowed to say “ok just evaluate for x86_64-linux”.
On the other hand, complete evaluation of `release.nix` should be much
faster because we no longer import `nixpkgs` for each test: testing with
the following command went from 30s to 18s, and that's just for a few
tests.
```
time nix-instantiate --eval --strict nixos/release.nix -A tests.nat
```
I initially wanted to test on the whole `release.nix`, but there are too
many broken tests and it takes too long to eval them all, especially
compared to the fact that the current implementation breaks some setup.
Given developers can just `nix-build nixos/tests/my-test.nix`, it sounds
like an overall win.
The `rmilter` module has options for configuring `postfix` to use it but
since that module is deprecated because rspamd now has a builtin worker
that supports the milter protocol this commit adds similar `postfix`
integration options directly to the `rspamd` module.
The lines stored in `extraConfig` and `worker.<name?>.extraConfig`
should take precedent over values from included files but in order to do
this in rspamd UCL they need to be stored in a file that then gets
included with a high priority. This commit uses the overrides option to
store the value of the two `extraConfig` options in `extra-config.inc`
and `worker-<name?>.inc` respectively.
When the workers option for rspamd was originally implemented it was
based on a flawed understanding of how workers are configured in rspamd.
This meant that while rspamd supports configuring multiple workers of
the same type, so that different controller workers could have different
passwords, the NixOS module did not support this because it would write
an invalid configuration file if you tried.
Specifically a configuration like the one below:
```
workers.controller = {};
workers.controller2 = {
type = "controller";
};
```
Would result in a rspamd configuration of:
```
worker {
type = "controller";
count = 1;
.include "$CONFDIR/worker-controller.inc"
}
worker "controller2" {
type = "controller";
count = 1;
}
```
While to get multiple controller workers it should instead be:
```
worker "controller" {
type = "controller";
count = 1;
.include "$CONFDIR/worker-controller.inc"
}
worker "controller" {
type = "controller";
count = 1;
}
```
When implementing #49620 I included an enable option for both the
locals and overrides options but the code writing the files didn't
actually look at enable and so would write the file regardless of its
value. I also set the type to loaOf which should have been attrsOf
since the code was not written to handle the options being lists.
This fixes both of those issues.
This module permits to preload Docker image in a VM in order to reduce
OIs on file copies. This module has to be only used in testing
environments, when the test requires several Docker images such as in
Kubernetes tests. In this case,
`virtualisation.dockerPreloader.images` can replace the
`services.kubernetes.kubelet.seedDockerImages` options.
The idea is to populate the /var/lib/docker directory by mounting qcow
files (we uses qcow file to avoid permission issues) that contain images.
For each image specified in
config.virtualisation.dockerPreloader.images:
1. The image is loaded by Docker in a VM
2. The resulting /var/lib/docker is written to a QCOW file
This set of QCOW files can then be used to populate the
/var/lib/docker:
1. Each QCOW is mounted in the VM
2. Symlink are created from these mount points to /var/lib/docker
3. A /var/lib/docker/image/overlay2/repositories.json file is generated
4. The docker daemon is started.
As reported by @andir, the regular expressions that match the sandbox
output are no longer matching in the recent Chromium bump as of
bb03fbc2c8.
Instead of a boolean field that determines whether namespace sandboxes
are on, the namespace sandbox is now an enum within "Layer 1 Sandbox".
I've modified the regular expressions accordingly and also ran the test
for the stable branch, which now succeeds.
Signed-off-by: aszlig <aszlig@nix.build>
Issue: https://github.com/NixOS/nixpkgs/issues/49442
Cc: @bendlas, @andir
By default rspamd will look for multiple files in /etc/rspamd/local.d
and /etc/rspamd/override.d to be included in subsections of the merged
final config for rspamd. Most of the config snippets in the official
rspamd documentation are made to these files and so it makes sense for
NixOS to support them and this is what this commit does.
As part of rspamd 1.8.1 support was added for having custom Lua
rules stored in $LOCAL_CONFDIR/rspamd.local.lua which means that it is
now possible for NixOS to support such rules and so this commit also
adds support for this to the rspamd module.
Rationale
---------
Currently, tests are hard to discover. For instance, someone updating
`dovecot` might not notice that the interaction of `dovecot` with
`opensmtpd` is handled in the `opensmtpd.nix` test.
And even for someone updating `opensmtpd`, it requires manual work to go
check in `nixos/tests` whether there is actually a test, especially
given not so many packages in `nixpkgs` have tests and this is thus most
of the time useless.
Finally, for the reviewer, it is much easier to check that the “Tested
via one or more NixOS test(s)” has been checked if the file modified
already includes the list of relevant tests.
Implementation
--------------
Currently, this commit only adds the metadata in the package. Each
element of the `meta.tests` attribute is a derivation that, when it
builds successfully, means the test has passed (ie. following the same
convention as NixOS tests).
Future Work
-----------
In the future, the tools could be made aware of this `meta.tests`
attribute, and for instance a `--with-tests` could be added to
`nix-build` so that it also builds all the tests. Or a `--without-tests`
to build without all the tests. @Profpatsch described in his NixCon talk
such systems.
Another thing that would help in the future would be the possibility to
reasonably easily have cross-derivation nix tests without the whole
NixOS VM stack. @7c6f434c already proposed such a system.
This RFC currently handles none of these concerns. Only the addition of
`meta.tests` as metadata to be used by maintainers to remember to run
relevant tests.
* journald: forward message to syslog by default if a syslog implementation is installed
* added a test to ensure rsyslog is receiving messages when expected
* added rsyslogd tests to release.nix
- added package option to specify which version of redmine
- added themes option back in to allow specifying redmine themes
- added plugins option back in to allow specifying redmine plugins
- added database.socket option to allow mysql unix socket authentication
- added port option to allow specifying the port rails runs on
- cleaned up Gemfile so it is much less hacky
- switched to ruby version 2.4 by default as suggested by documentation http://www.redmine.org/projects/redmine/wiki/redmineinstall#Installing-Redmine
- fixed an annoyance (bug) in the service causing recursive symlinks
- fixed ownership bug on log files generated by redmine
- updates reflecting renames in nixos options
- added a nixos test
From commit b63f65aea0dea11c20e9299210af1d2ee4299b58:
I used tmpfiles.d instead of activation snippets to create the logs.
It's good enough for upstream and other distros; it's probably good
enough for us.
The "reboot-wtmp" subtest fails because it it assumes that there is a
reboot record even on the initial boot. This is only the case if wtmp is
created within the activation script, but the implementation now uses
tmpfiles.d, so the creation of the file is done at a much later stage.
Apart from that, if you think about the state after the installation as
"first boot", using the term "reboot" wouldn't probably make sense
either.
So in our subtest, we now reboot the machine and check the wtmp record
afterwards as we did before.
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @edolstra, @jameysharp, @Mic92
The socket activation I added to the rspamd module doesn't actually work
and can't be made to work without changes to rspamd.
See: #47421
See: rspamd/rspamd#2035
The test failed on Hydra in one instance because a request to the
server was sent before indexing was finished.
Retry the request until it succeeds (or times out).
The test failed in one run on Hydra, logs look like
dhcpcd changed ipv6 routing at just the wrong time.
Disable dhcpcd. It's not needed, the test uses static IPs anyway.
Test didn't run because it tried to create a VM with 4096M RAM
but qemu-system-i386 has a hard 2047M memory limit.
- reduce memory to 2047M on i686.
- increase timeout 300s -> 1800s because the tests are much slower
on i686 and timed out.
The `pkgs.yabar` package is relatively old (2016-04) and contains
several issues fixed on master. `yabar-unstable` containsa recent master
build with several fixes and a lot of new features (I use
`yabar-unstable` for some time now and had no issues with it).
In the upstream bugtracker some bugs could be fixed on ArchLinux by
simply installing `yabar-git` (an AUR package which builds a recent
master).
To stabilize the module, the option `programs.yabar.package` now
defaults to `pkgs.yabar-unstable` and yields a warning with several
linked issues that are known on `pkgs.yabar`.
The test has been refactored as well to ensure that `yabar` actually
starts (and avoid non-deterministic random success) and takes a
screenshot of a very minimalistic configuration on IceWM.
Fixes#46899
This attempts to improve stability of the test by using existing
services for miniupnpd and transmission.
It also uses explicit addresses for the network interfaces so that the
external IP addresses are valid internet addresses (thus fixing
validation problems from upnpc).
Also disable eth0 from being used to transfer torrents over without that
being the intention.
A sporadic failure occured on Hydra because a request was sent
to smtpd after the systemd unit was started, but before the daemon
was actually listening. Fix by checking for open ports first.
A sporadic failure occured on Hydra because a request was sent
to the daemon after the systemd unit was started, but before the
daemon was actually listening. Fix by checking for open port first.
