This can serve as a starting point, but it shows there are a few
problems to solve: we need to verify json that depends on the setup,
in particular, the json the queue contains will contain UUID's that
are generated out of our control.
Moreover, the setup for this test only makes sense for internal test,
so I think we may want to change the logic for whether or not a test
sholud be supported to be run externally to be per test-function,
rather than per call to sendHTTP().
Signed-off-by: Tom Gundersen <teg@jklm.no>
This does not change the behavior, but refactors according to these principles:
1) No two routes are tested in the same function (but it would be ok to split
tests for one route over several funcions)
2) At most one testRoute() call is made per API object, and the state is
completely set up and tore down between tets.
On top of this we should add more test cases to each of the tables, but
I'm leaving this to future PRs.
Signed-off-by: Tom Gundersen <teg@jklm.no>
This simplifies the code a bit, and will be used in follow-up patches
to distinguish between setup calls and explicit tests.
Signed-off-by: Tom Gundersen <teg@jklm.no>
A job is now in "WAITING" state exactly when it is in the channel,
once it is popped it enters "RUNNING" state. It is only possible
to update the state of a job that is in the running state.
This mean that updating to "RUNNING" is entirely optional, but in
the future we may want to use this as a watchdog logic, and require
the worker to update at regular intervals to avoid being restarted.
The job queue API is updated to require a POST followed by one
or several PATCH messages to the returned ID. If a patch is sent
to an ID before the POST it is as if the object does not exist
(regarldess of it being in the queue in WAITING state or not).
Once a job has been POSTed it can be PATCHed to update it zero or
more times with (still) RUNNING before exactly oncee with either
FINISHED or FAILED.
Signed-off-by: Tom Gundersen <teg@jklm.no>
List of currently unsupported ones:
- [[repos.git]]
- [customizations.kernel]
- [[groups]]
- [[packages]] and [[modules]]
Some of customizations have unimplemented behaviour, see TODOs
Not all tests will be compatible with external APIs such as Lorax. When
calling testRoute each test now declares if it can run against an
external API or not. This change allows us to test against Lorax but
skip the cases that will be invalid when not run against
osbuild-composer's API.
In order to maintain parity with lorax the api needs to reply with an
error message equivalent to that used by lorax. Error messages are now
returned inside an error object that contains an id, message, and
optional status code.
For some routes, there are errors when no url parameters are passed. The
httprouter was using named parameters of the form /:param which does not
match for empty parameters. Now, it has been updated to use the
catch-all parameters of the form /*param. This change allows the case of
no parameters. However, parameters will now include a "/" as
their first character. This needs to be removed from the string in the
route handler.
In order to provide the proper error message for
/modules/list/<modules>, searching for the modules needed to be updated.
The requested modules and known packages are iterated over and if there
is a match the module is added to the response. Also, the found module
is dropped from the list of requested modules. If this list is not empty
after searching all of the modules then an error is returned containing
the name of the non-existant module.
Use the autogenerated test framework from VSCode (with minor fixes to
make them compile).
As we do not yet support customizations in blueprint, all the tests are
trivial.
I was not entirely sure about the best way to encode the wanted pipeline
output. I currently represent it as a JSON string and unmarshal it into
the object to compare with.
Signed-off-by: Tom Gundersen <teg@jklm.no>
When osbuild-composer is run as systemd service, we don't want to write
anything into working directory. Currently, we write dnf cache into it.
Instead, let's just use the default dnf cache directory.
log.Fatalf not only writes into a log, but also exits the process.
Instead, we want the caller to handle the error.
Also, the logged message is imho wrong.
Unify into one make file, with two targets:
build:
This unconditionally builds the binaries that wolud be shipped in the package,
this is marked as a phony target, and relies on the go compiler to know what
is up-to-date and what needs to be rebuilt.
install:
This installs binaries and config files into the file system and runs any
scripts to set things up correctly, as if the rpm had been installed.
This drops the old `run` target, which could still be done from the commandline,
but we want to be able to rely on systemd features, so it is not something we
want to test or support.
To run osbuild-composer from a checkout one should now do:
```
$ make build
$ sudo make install
```
This has the same effect as installing the RPM would have, starting and stopping
the sokcets/services is still left to the user.
Signed-off-by: Tom Gundersen <teg@jklm.no>
At the end of each set of tests the blueprint created is deleted.
This tests the delete route as well as removing blueprints
created while testing external apis such as lorax.
This tool outputs the pipeline associated with each output type. In
the future this should grow to take a blueprint as an input, and
also allow the arch/os/version/variant to be configured.
Signed-off-by: Tom Gundersen <teg@jklm.no>
This was only used internally in the store package, so moved the
one type that was still used and remove the rest.
Signed-off-by: Tom Gundersen <teg@jklm.no>
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>
This encapsulates all the state of osbuild, no longer just the state specific
to the weldr API. Make it a first class type, preparing for direct access from
both the weldr and jobqueue APIs.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Let the store in weldr be the only one that keeps state, and push
updates directly there. This fixes a bug where there was an ID mismatch.
Change the API to not let the caller pick the UUID, but provide it
in the response. Use the same UUID as is used to identify composes,
this makes it simpler to trace what is going on.
Signed-off-by: Tom Gundersen <teg@jklm.no>
Each output type will have its own translator from a blueprint to
a pipeline.
In the future, we may also want to distinguish between different
base distributions and architectures, but for now we keep it
simple and hardcode to a given one.
Some placeholder output formats are added, these have not been
tested or reserached, but are included to show how the it will
all fit together.
Signed-off-by: Tom Gundersen <teg@jklm.no>
