The uploading of artifacts back to the worker server for the on-premise
(Weldr) use case was signaled to the worker by setting the `ImageName`
in the `OSBuildJob` definition. The code also relies on the osbuild
exports being specified in the `OSBuildJob`, instead of in the target
(this is not implemented yet).
Prepare the ground for moving osbuild export definition from
`OSBuildJob` to `Target` by introducing an explicit `Worker Server"
upload target. This target will signal to the worker that it should
upload the image back to the worker server. The new target is not yet
used by any API implementation.
Extend the worker osbuild job implementation to handle the new upload
target.
The filename of the image as produced by osbuild for a given export is
currently set in each target options type in the `Filename` struct
member. However, the value is not really specific to any target type,
but to the specific export used for the target. For this reason move the
value form target type options to the `Target` struct inside a new
struct `OsbuildArtifact` under the name`ExportFilename`.
The backward compatibility with older implementations of the composer
and workers is kept on the JSON (Un)mashaling level, where the JSON
object is always a super-set of the old and new way of providing the
export filename in the Target.
Some variable names used in the `OsBuildJob` `Run()` method were not
very self-explanatory, which made the code harder to understand and
navigate. These were `args`, `options`, `t`. Rename them to be more
self-explanatory of their purpose.
A backward compatibility code handling the conversion of VMDK image to
stream-optimized sub-format has been kept in the implementation since
PR#2529 [1] merged on May 4th 2022. Since this change, no API
implementation is submitting jobs, which would hit this conversion code,
because VMDK images are already being produced in the desired
sub-format.
On-premise deployments are expected to use the same composer and worker
versions. There are no composer / worker instances in production, which
are not running the modified code.
Delete the backward compatibility code.
[1] https://github.com/osbuild/osbuild-composer/pull/2529
Modify the `OsBuildJob` implementation to handle multiple upload targets
in a cycle. However, there is still no API implementation, which would be
adding `OsBuildJobs` with multiple targets to the job queue.
The limitations are that only a single osbuild export is supported, and
the same artifact will be used for each target.
At the end of the job, errors from all targets are gathered. In case
there are none, the job succeeds. In case at least one target failed,
the job fails as well. In such a case, a slice of errors from all failed
targets is added to the job error as details.
Add a new worker client error type `ErrorTargetError` representing that
at least one of job targets failed. The actual target errors are added
to the job detail.
Add a new `OSBuildJobResult.TargetErrors()` method for gathering a slice
of target errors contained within an `OSBuildJobResult` instance. Cover
the method with unit test.
Ensure that a target result with a proper error is added to the Job
result, in case the there was any error encountered. This error is not
used at all for now. Keep setting the `JobError` to the same error set
in the target result for now.
This is a step towards job results containing multiple target results
with each or them having potentially an error set as well.
Do not pass the `worker.OSBuildJobResult` to `uploadToS3()`, but instead
return target errors from the function. This will make the error
handling of all upload targets consistent and easier to modify to
support multiple targets.
Add a new generic container registry client via a new `container`
package. Use this to create a command line utility as well as a
new upload target for container registries.
The code uses the github.com/containers/* project and packages to
interact with container registires that is also used by skopeo,
podman et al. One if the dependencies is `proglottis/gpgme` that
is using cgo to bind libgpgme, so we have to add the corresponding
devel package to the BuildRequires as well as installing it on CI.
Checks will follow later via an integration test.
The only functional change is that
base_path = ""
will be now parsed as:
config.BasePath == ""
which wasn't possible before.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
The struct is factored out 1:1. The only functional change in this commit is
worker now logging in case of a missing config (which means just loading the
defaults).
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
This commit moves the field to the koji struct where it actually belongs.
Also, it renames it to relax_timeout_factor for the sake of consistency.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
The TOML library translates the field names 1:1, so now you have to use:
[Composer]
proxy: "abcd"
This is not idiomatic though so let's add the toml tag to make it [composer].
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
When the Koji target support was added to the osbuild job, based on the
osbuild-koji job, the meaning of target option values got messed up.
The side effect of the issue is that when Koji composes are
submitted via Cloud API the resulting image is currently always uploaded
back to the worker server.
`OsBuildKoji` job
-----------------
- `OSBuildKojiJob.ImageName` is set to the filename of the image as
exported by osbuild.
- `OSBuildKojiJob.KojiFilename` is set to the desired filename which
should be used when uploading the image to Koji.
`OsBuild` job + `KojiTargetOptions` before
------------------------------------------
- `OSBuildJob.ImageName` is set to the filename of the image as exported
by osbuild. This is done only by the Cloud API code for Koji composes.
Cloud API does not set this for regular composes and any other target.
The variable is set in common case only by Weldr API code with the
same meaning and it is used by the `OsBuild` job implementation as an
indication that the image should be uploaded back to the worker server.
- `Target.ImageName` is not set at all. Other targets use it for the
desired filename which should be used when uploading the image to the
target environment.
- `KojiTargetOptions.Filename` is set to the desired filename which
should be used when uploading the image to Koji. All other target
types use `Filename` variable in their options for the filename of the
image as exported by osbuild.
`OsBuild` job + `KojiTargetOptions` after
-----------------------------------------
- `OSBuildJob.ImageName` is still set to the filename of the image as
exported by osbuild. This is kept for a backward compatibility of new
composer with older workers.
- `Target.ImageName` is set to the desired filename which should be used
when uploading the image to Koji.
- `KojiTargetOptions.Filename` is set to the filename of the image as
exported by osbuild.
This change is backward incompatible, meaning that old worker won't be
able to handle Koji compose requests submitted via Cloud API using a new
composer and also a new worker won't be able to handle Koji compose
requests submitted by a new composer. This is intentional, because after
discussion with Ondrej Budai, the Cloud API Koji integration is
currently not used anywhere in production.
The return statement was forgotten when the Koji target support was
added. As a result, a Job with a failed Koji upload would be reported
as successful, while at the same time having a `JobError` set.
It always felt wrong that the method uploaded the blob under a different name
than the one specified in the blob metadata.
This commit moves the responsibility of specifying the right extension to
the callers. azure.EnsureVHDExtension helper was added to simplify this.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
Adjust the `koji-finalize` job implementation to be able to handle
results from both the `osbuild` and `osbuild-koji` jobs.
In case of `osbuild` job, the result is of type
`worker.OSBuildJobResult` and the important values are stored in the
Koji upload target options. For now assume that there may be only a
single upload target results.
In case of `osbuild-koji` job, the result is of type
`worker.OSBuildKojiJobResult` and the important values are already part
of the structure. Add "Old" suffix to all functions handling this case.
Ensure that none of the job dependencies failed. This covers the case
when there are more than one job dependencies, which will be the case
for Koji composes.
Previously, the `OSBuild` job assumed that it can have only a single
job dependency, which could be only the `ManifestJobByID`. This won't
work well for the Koji use case, because the Koji OSBuild job has also
dependency on the Koji-init job.
Extend the `worker.OSBuildJob` structure with a new field, which holds
the `ManifestJobByIDResult` index in the job's dynamic arguments slice.
This value is considered in case when there is more than one dependency
of the `OSBuild` job.
Define supported job type names as constants and use them in all places,
instead of string literals.
There are multiple benefits of this approach. Using constants removed
the room for typos in the string literals. One can use autocompletion in
IDE for job types. Using constant makes it easier to find all references
where it is used and thus all places that are handling a specific job
type.
It is generally useful to have this information in the
`OSBuildJobResult`. This information is currently part of the
`OSBuildKojiJobResult`. Instead of moving it to the new
`KojiTargetResultOptions`, lets move it to the `OSBuildJobResult`
structure and set it for all jobs.
The `distro` package is now used for distro definitions supported by
osbuild-composer, not for introspecting the Host system. Move
`GetRedHatRelease()` and `GetHostDistroName()` functions to the `common`
package.
This error is failing to parse correctly on the workers as a
dnfjson.Error. The old rpmmd.DNFError was returned by pointer, however
the internal/dnfjson package returns the Error by value.
Use case
--------
If Endpoint is not set and Region is - upload to AWS S3
If both the Endpoint and Region are set - upload the Generic S3 via Weldr API
If neither the Endpoint and Region are set - upload the Generic S3 via Composer API (use configuration)
jobimpl-osbuild
---------------
Add configuration fields for Generic S3 upload
Support S3 upload requests coming from Weldr or Composer API to either AWS or Generic S3
Weldr API for Generic S3 requires that all connection parameters but the credentials be passed in the API call
Composer API for Generic S3 requires that all conneciton parameters are taken from the configuration
Adjust to the consolidation in Target and UploadOptions
Target and UploadOptions
------------------------
Add the fields that were specific to the Generic S3 structures to the AWS S3 one
Remove the structures for Generic S3 and always use the AWS S3 ones
Worker Main
-----------
Add Endpoint, Region, Bucket, CABundle and SkipSSLVerification to the configuration structure
Pass the values to the Server
Weldr API
---------
Keep the generic.s3 provider name to maintain the API, but unmarshel into awsS3UploadSettings
tests - api.sh
--------------
Allow the caller to specifiy either AWS or Generic S3 upload targets for specific image types
Implement the pieces required for testing upload to a Generic S3 service
In some cases generalize the AWS S3 functions for reuse
GitLab CI
---------
Add test case for api.sh tests with edge-commit and generic S3
Added CleanCache() method to the solver that deletes all the caches if
the total size grows above a certain (configurable) limit
(default: 500 MiB).
The function is called externally to handle errors (usually log or
ignore completely) and to avoid calling multiple times for multiple
depsolves of a single request.
The cleanup is extremely simple and is meant as a placeholder for more
sophisticated cache management. The goal is to simply avoid ballooning
cache sizes that might cause issues for users or our own services.
The repository checksums in the response from dnf-json aren't used
anywhere. Since we're making changes to dnf-json and depsolving, now is
a good opportunity to drop them completely.
Move package set chain collation to the distro package and add
repositories to the package sets while returning the package sets from
their source, i.e., the ImageType.PackageSets() method.
This also removes the concept of "base repositories". There are no
longer repositories that are added implicitly to all package sets but
instead each package set needs to specify *all* the repositories it will
be depsolved against.
This paves the way for the requirement we have for building RHEL 7
images with a RHEL 8 build root. The build root package set has to be
depsolved against RHEL 8 repositories without any "base repos" included.
This is now possible since package sets and repositories are explicitly
associated from the start and there is no implicit global repository
set.
The change requires adding a list of PackageSet names to the core
rpmmd.RepoConfig. In the cloud API, repositories that are limited to
specific package sets already contain the correct package set names and
these are now copied to the internal RepoConfig when converting types in
genRepoConfig().
The user-specified repositories are only associated with the payload
package sets like before.
Attach the repository configurations that are specific to a package set
directly on the PackageSet object. This simplifies the Depsolve()
signature and avoids requiring a `nil` when no additional repositories
are required. More importantly, it makes associating repositories to
package sets explicit, no longer relying on matching array indices or
map keys.
Remove the single Depsolve function from the dnfjson package and the
depsolve command from the dnf-json tool. The new ChainDepsolve
functions and chain-depsolve command can handle single depsolves in the
same way so there's no need to keep (and have to maintain) two versions
of very similar code.
The ChainDepsolve function (in Go) and chain-depsolve command (in
Python) have been renamed to plain Depsolve and depsolve respectively,
since they are now general purpose depsolve functions.
All calls to rpmmd.Depsolve() are now replaced with the equivalent call
to solver.Depsolve() (or dnfjson.Depsolve() for one-off calls).
Attached an unconfigured dnfjson.BaseSolver to all APIs and server
configurations where rpmmd.RPMMD used to be. This BaseSolver instance
loads the repository credentials from the system and carries the cache
directory, much like the RPMMD field used to do. The BaseSolver is used
to create an initialised (configured) solver with the platform variables
(module platform ID, release ver, and arch) before running a Depsolve()
or FetchMetadata() using the NewWithConfig() method.
The FillDependencies() call in the modulesInfoHandler() of the weldr API
has been replaced by a direct call to the Depsolve() function. This
rpmmd function was only used here. Replacing the rpmmd.Depsolve() call
in rpmmd.FillDependencies() with dnfjson.Depsolve() would have created
an import cycle. The FillDependencies() function could have been moved
to dnfjson, but since it's only used in one place, moving the one-line
function body into the caller is ok.
For testing:
The mock-dnf-json is compiled to a temporary directory during test
initialisation and used for each Depsolve() or FetchMetadata() call.
The weldr API tests now use the mock dnfjson. Each rpmmd_mock.Fixture
now also has a dnfjson_mock.ResponseGenerator.
All API calls in the tests use the proper functions from dnfjson and
only the dnf-json script is mocked. Because of this, some of the
expected results in responses_test had to be changed to match correct
behaviour:
- The "builds" array of each package in the result of a module or
project list is now sorted by version number (ascending) because we
sort the package list in the result of dnfjson by NVR.
- 'check_gpg: true' is added to the expected response of the depsolve
test. The repository configs in the test weldr API specify 'CheckGPG:
True', but the mock responses returned it as false, so the expected
result didn't need to include it. Since now we're using the actual
dnfjson code to convert the mock response to the internal structure,
the repository settings are correctly used to set flag to true for
each package associated with that repository.
- The word "occurred" was mistyped as "occured" in rpmmd and is now
fixed in dnfjson.
API
---
Allow the user to pass the CA public certification or skip the verification
AWSCloud
--------
Restore the old version of newAwsFromCreds for access to AWS
Create a new method newAwsFromCredsWithEndpoint for Generic S3 which sets the endpoint and optionally overrides the CA Bundle or skips the SSL certificate verification
jobimpl-osbuild
---------------
Update with the new parameters
osbuild-upload-generic-s3
-------------------------
Add ca-bunlde and skip-ssl-verification flags
tests
-----
Split the tests into http, https with certificate and https skip certificate check
Create a new base test for S3 over HTTPS for secure and insecure
Move the generic S3 test to tools to reuse for secure and insecure connections
All S3 tests now use the aws cli tool
Update the libvirt test to be able to download over HTTPS
Update the RPM spec
Kill container with sudo
Before, the autoscaling group discovery is failing with error:
Error getting the Autoscaling instances: MissingRegion MissingRegion:
could not find region configuration
Before the instance was vulnerable to an OTA update while processing a
request. Because there is no way of retriggering a job in Composer, it
is better to avoid this situation.
The way we are doing it is by setting the `protected` flag onto the
instance when a job is being processed. This way the AWS scheduler
does hopefully not shutdown the machine at the wrong time.
Main caveats of this solution:
* Starvation: If a worker keeps accepting new jobs, then it might not be
updated.
* Inconsistency: There exist a window between the job acceptation and the
protection where the worker can be shutdown without having the time to
protect itself.
The VMDK image is already produced as stream-optimized. Therefore stop
setting the `StreamOptimized` option in `OSBuildJob` structure by both,
Weldr and Cloud APIs.
Keep the handling of the option in worker for backward compatibility,
in case an older instance of Composer server is used, which does not
produce VMDK manifests as stream-optimized. In such case, the worker
needs to convert the image.
In the internal deployment, we want to talk with composer over a http/https
proxy. This proxy adds new composer.proxy field to the worker config that
causes the worker to connect to composer and the oauth server using
a specified proxy.
NB: The proxy is not supported when connection to composer via unix sockets.
For testing this, I added a small HTTP proxy implementation, pls don't
use this in production, it's just good enough for tests.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
Address is always required so not passing one is a clear error, let's return
exit code 2 which go itself returns when bad arguments are passed in.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
Add support for importing the GCE image into GCP using Weldr API. The
credentials to be used can be specified in the upload settings and will
be then used by the worker to authenticate with GCP.
The GCP target credentials are passed to Weldr API as base64 encoded
content of the GCP credentials JSON file. The reason is that the JSON
file contains many values and its format could change in the future.
This way, the Weldr API does not rely on the credentials file content
format in any way.
Add a new test case for the GCP upload via Weldr and run it in CI.
Signed-off-by: Tomas Hozza <thozza@redhat.com>
Refactor the handling of GCP credentials in the worker to be equivalent
to what is done for AWS. The main idea is that the code decides which
credentials to use when processing each job. This change will allow
preferring credentials passed via upload `TargetOptions` with the job,
over the credentials configured in worker's configuration or the default
way of authenticating implemented by the Google library.
Move loading of GCP credentials to the internal `gcp` library into
`NewFromFile()` function accepting path to the file with credentials.
Signed-off-by: Tomas Hozza <thozza@redhat.com>
