This reduces the amount of resolving and error checking we have to do.
This exposed a bug in weldr's ComposeEntry type, which will be fixed in
a follow-up commit.
Signed-off-by: Tom Gundersen <teg@jklm.no>
This makes the queue more type safe and allows to get rid of the
`pendingChannel` and `pendingChannels` helpers, which only existed to
create not-yet-existing pending channels.
The store is responsible for two things: user state and the compose queue. This
is problematic, because the rcm API has slightly different semantics from weldr
and only used the queue part of the store. Also, the store is simply too
complex.
This commit splits the queue part out, using the new jobqueue package in both
the weldr and the rcm package. The queue is saved to a new directory `queue/`.
The weldr package now also has access to a worker server to enqueue and list
jobs. Its store continues to track composes, but the `QueueStatus` for each
compose (and image build) is deprecated. The field in `ImageBuild` is kept for
backwards compatibility for composes which finished before this change, but a
lot of code dealing with it in package compose is dropped.
store.PushCompose() is degraded to storing a new compose. It should probably be
renamed in the future. store.PopJob() is removed.
Job ids are now independent of compose ids. Because of that, the local
target gains ComposeId and ImageBuildId fields, because a worker cannot
infer those from a job anymore. This also necessitates a change in the
worker API: the job routes are changed to expect that instead of a
(compose id, image build id) pair. The route that accepts built images
keeps that pair, because it reports the image back to weldr.
worker.Server() now interacts with a job queue instead of the store. It gains
public functions that allow enqueuing an osbuild job and getting its status,
because only it knows about the specific argument and result types in the job
queue (OSBuildJob and OSBuildJobResult). One oddity remains: it needs to report
an uploaded image to weldr. Do this with a function that's passed in for now,
so that the dependency to the store can be dropped completely.
The rcm API drops its dependencies to package blueprint and store, because it
too interacts only with the worker server now.
Fixes#342
This package does not contain an actual queue, but a server and client
implementation for osbuild's worker API. Name it accordingly.
The queue is in package `store` right now, but about to be split off.
This rename makes the `jobqueue` name free for that effort.
weldr needs to know the host architecture. Rather than pinning
a string, pin a real Arch object, and query its name when we
need it.
This verifies the validitiy of the architecture for the given
distro before it is passed to weldr, rather than lazily on
demand.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Require the caller to pass in the required distros explicitly. This
would allow us to easily add distros in osbuild-pipeline and tests
before exposing them in composer itself, for instance.
This means there is no longer a dependency from the distro package
to each of the individual distros, so the distros are now able
to depend on the distro packag for types and interfaces.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Mixing the way to build a distribution with where to get the source
packages from is wrong: it breaks pre-release repos, local mirrors, and
other use cases. To accommodate those, we introduced
`/etc/osbuild-composer/repositories`.
However, that doesn't work for the RCM API, which receives repository
URLs to use from outside requests. This API has been wrongly using the
`additionalRepos` parameter to inject those repos. That's broken,
because the resulting manifests contained both the installed repos and
the repos from the request.
To fix this, stop exposing repositories from the distros, but require
passing them on every call to `Manifest()`. This makes `additionalRepos`
redundant.
Fixes#341
Only the weldr API has the concept of a default distro. Pass that distro
explicitly to `PushCompose()` and fetch the distro from the compose in
all other functions that accessed Store.Distro.
`ComposeRequest` included a `common.Distribution`, which had to be
resolved in PushComposeRequest. Use a real `distro.Distro` object here,
and push resolving it to the rcm package.
Change the `Distribution` on the (lower-case) `composeRequest` to a
string. This struct represents the incoming request. Since we're now
resolving the real distro object from the registry in the same function,
it seems redundant to validate the incoming distro twice.
Use $STATE_DIRECTORY environment variable which is set by systemd
because we use: StateDirectory=osbuild-composer in the service unit.
also change systemd unit to include STATE_DIRECTORY, because
RHEL comes with older systemd version, so we need to set this variable explicitly.
Return errors from all distro's New() functions instead of logging and
returning nil. Also, return errors instead of panicking from
NewRegistry() and NewDefaultRegistry().
WithSingleDistro() doesn't follow go's naming convention for creating
objects (New*). Rename it to NewRegistry() and rename the old
NewRegistry() to NewDefaultRegistry().
The idea is that NewRegistry() can be used to create full Registry
objects from outside the package. NewDefaultRegistry() is a convenience
function that creates a Registry with all known distros.
rpmmd looked at the CACHE_DIRECTORY environment variable to set a path
for the dnf repository cache. Aside from being a smelly thing to do
from a library, this breaks osbuild-pipeline and osbuild-dnf-json-tests,
which don't run as systemd services and thus don't have CACHE_DIRECTORY
set.
Explicitly pass the cache directory to rpmmd. Keep using a path based on
CACHE_DIRECTORY for osbuild-composer. Use the user's `.cache` directory
for osbuild-pipeline and a temporary directory for the tests.
Right now the implementation expects the RCM socket to live in the same
unit file as other osbuild-composer sockets. This would require a
solution where we ship the osbuild-composer.socket in two different
versions: one for regular usage, one for rcm. But that is very
inconvenient and it would probably require some weird scriptlets (and
scriptlets are bad!).
After this change, the RCM API socket lives in a separate file and only
if the socket unit is activated, the API runs. The unit file itself was
introduced in previous commits.
There's a usecase for running workers at a different machine than
the composer. For example when there's need for making images for
architecture different then the composer is running at. Although osbuild has
some kind of support for cross-architecture builds, we still consider it
as experimental, not-yet-production-ready feature.
This commit adds a support to composer and worker to communicate using TCP.
To ensure safe communication through the wild worlds of Internet, TLS is not
only supported but even required when using TCP. Both server and client
TLS authentication are required. This means both sides must have their own
private key/certificate pair and both certificates must be signed using one
certificate authority. Examples how to generate all this fancy crypto stuff
can be found in Makefile.
Changes on the composer side:
When osbuild-remote-worker.socket is started before osbuild-composer.service,
osbuild-composer also serves jobqueue API on this socket. The unix domain
socket is not affected by this changes - it is enabled at all times
independently on the remote one. The osbuild-remote-worker.socket listens
by default on TCP port 8700.
When running the composer with remote worker socket enabled, the following
files are required:
- /etc/osbuild-composer/ca-crt.pem (CA certificate)
- /etc/osbuild-composer/composer-key.pem (composer private key)
- /etc/osbuild-composer/composer-crt.pem (composer certificate)
Changes on the worker side:
osbuild-worker has now --remote argument taking the address to a composer
instance. When present, the worker will try to establish TLS secured TCP
connection with the composer. When not present, the worker will use
the unix domain socket method. The unit template file osbuild-remote-worker
was added to simplify the spawning of workers. For example
systemctl start osbuild-remote-worker@example.com
starts a worker which will attempt to connect to the composer instance
running on the address example.com.
When running the worker with --remote argument, the following files are
required:
- /etc/osbuild-composer/ca-crt.pem (CA certificate)
- /etc/osbuild-composer/worker-key.pem (worker private key)
- /etc/osbuild-composer/worker-crt.pem (worker certificate)
By default osbuild-composer.service will always spawn one local worker.
If you don't want it you need to mask the default worker unit by:
systemctl mask osbuild-worker@1.service
Closing remarks:
Remember that both composer and worker certificate must be signed by
the same CA!
This is needed for unit tests, because it wasn't possible to mock the
rpmmd module before. This also requires that the checksum is moved to
the compose request and evaluated in the endpoint handler instead of
push compose. I think it makes sense to have the checksum in the compose
request directly.
Also a "module platform ID" is required now, but we don't have the
"global" distribution any more, so this patch introduces mapping from a
distribution to the module platform ID.
Osbuild-composer expects two or three listeners and fails if there is
an unexpected number of listening sockets. Checking if there are not two
or if there are not three listeners always returns true even if there are
the desired number of listeners. Therefore, osbuild-composer always
crashes. The check now only crashes if there are a number of listeners
other than 2 or 3.
It contains two basic endpoints:
* POST /v1/compose
* GET /v1/compose/<uuid>
It passes all the tests, but cannot be used for the intended use case
because the store API does not (yet) support distributions and
architectures as a parameters.
During development of a new distro, we need to test composer against
nightly or beta repositories, but we cannot ship composer itself
with the nightly repository information hardcoded in. At the same
time, we want to distinguish between the system repositories of the
host and the repositories we use to generate images (the host may not
use the same distro/version/architecture as the target, and it may
include custom repositories that the target should not).
We therefore ship per distro repository information that can be
overriden (typically in testing) by dropping files in /etc.
For now use the latest nightlies for RHEL-8.2, we may want to
replace these with the official mirrors for GA eventually.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Introduce a DistroRegister object. For now this does not introduce
any functional changes, as the object is always instantited to be
the same. However, in follow-up patches it will get options.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Only panic on compile-time errors (e.g., built for unsupported
architecture). Otherwise, use log.Fatalf(), which is equivalent to
printing and exiting with return code 1. Only ever do this from
main(), in all other cases pass on the error object.
This is mostly relevant when the server disconects, in which case
we'll get EOF, and will now restart cleanly instead of panicing.
Signed-off-by: Tom Gundersen <teg@jklm.no>
The pipeline generation now takes the architecture as an argument.
Currently only x86_64 is supported. The architecture is detected
at start-up, and passed down to each pipeline translation.
For osbuild-pipeline we now requrie the architecture to be passed
in.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Prior to this commit outputs directory used by local target was owned by root.
This made impossible for osbuild-composer to delete images. (osbuild-composer
doesn't run as root).
This commit introduces state directory in which osbuild-composer creates
outputs directory. Because this directory is owned by osbuild-composer, it's
able to delete files inside.
Make osbuild-composer use FromHost() directly. Everywhere else needs to
specify the distro explicitly.
Also don't panic when a distro doesn't exist. Instead, return nil. Make
sure all callers check for that.
Make distros export repository information and use those in the weldr
API. This means that repos are only specified once and that the API
returns the right packages when we allow different distros.
The package list is generated on each request for a package so there is
no longer a need to generate the package list in main or to store these
packages in the API object.
We want to test API methods which calls dnf. Unfortunately, calling dnf
is expensive operation - it requires network access and downloading
a lot of (meta)data. This commit changes the rpmmd implementation
so that it can be mocked.
Make each command accept a `repos` key containing repository
descriptions.
Make weldr API pass the repository like this. Nothing should change,
because the repos were the same (Fedora 30).
This hides the state hanlding in the store package. As before, it
can be disabled by passing `nil` instead of the path to the state
file.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Wrap the channel in Pop and Push methods, so it is not exposed to
the callers. PushCompose replaces the old AddCompose for consistency,
and PopCompose simply reads from the other end of the channel.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Drop the jobUpdates channel, and instead add an UpdateCompose method
to the store, which updates the status of a compose directly.
This allows us to report back errors directly, rather than having to
mirror the staet in the jobqueue API.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Let the weldr API take the store as an argument, rather than create it
itself. This will allow us to share the store with the jobqueue API in
follow-up patches.
Signed-off-by: Tom Gundersen <teg@jklm.no>
osbuild-composer now uses socket activation instead of hardcoded paths
in the code. osbuild-worker is an http client therefore it uses only
service unit. osbuild-worker must be started after the socket is
created. osbuild-composer service requires osbuild-worker to run, because without
it no jobs can be started.
osbuild-composer is executed as a regular user (newly created
_osbuild-composer user) as opposed to the worker which must run as root
in order to execute osbuild itself
This way it can be correctly exposed in the API. We listen on a channel
from the job-queue, where status updates are pushed when the worker is
running/finished (or, in the future, failed).
Signed-off-by: Tom Gundersen <teg@jklm.no>
This is by no means done, and needs more tests, docs and bugfixes,
but push it early so we have a common base to work on.
Based on work by Martin Sehnoutka.
Signed-off-by: Tom Gundersen <teg@jklm.no>
