The mechanisms here are a bit subtle and the kiwibuild phase
didn't quite get them right. The arg passed to `util.failable`
is supposed to be a boolean, but kiwibuild was passing it the
list of failable arches (which will always evaluate True).
How this is meant to work is that we only make *the Koji task
as a whole* failable (by passing `True` to `util.failable`) if
*all* the arches in it are failable. If *any* arch in the task
is not failable, the task should not be failable.
We allow a subset of arches to fail by passing the Koji task a
list of `optional_arches`, later. If an arch is 'optional', that
arch failing won't cause the Koji task itself to be considered
failed.
This commit fixes the logic (I hope), renames all the variables
and adds a couple of comments to make it clearer what's going on,
and does a bit of making the code simpler.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
Kiwi builds for Azure fixed VHD images are suffixed with "vhdfixed"
instead of plain "vhd". Add that to the list of suffixes.
Signed-off-by: Jeremy Cline <jeremycline@microsoft.com>
According to Neal, Vagrant images produced by Kiwi end in e.g.
`vagrant.libvirt.box` and `vagrant.virtualbox.box` - with a
period between `vagrant` and the image type, not a dash as with
oz. We should accept this slightly different format so we can
correctly derive the productmd `type` and `format` for these.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
The osbuild Koji plugin supports passing customizations for an image
build. This is also supported in the Koji CLI plugin. Some teams want to
pass image customizations for images built as part of Pungi composes.
Extend the osbuild phase to support passing customizations in the Pungi
configuration.
Merges: https://pagure.io/pungi/pull-request/1733
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
This only matters for composes that use the functionality for trimming
addon packages from parent variants.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
If you configure default branch name in new repos to anything else than
master, there will be failures in tests. The test expects the branch to
be called master, but does not ensure it in any way.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
When running with requests<2.18 (i.e. on RHEL 7), streaming responses
are not a context manager and need to be wrapped in contextlib.closing.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
The plugin supports two modes of operation:
1. Mount a shared storage volume into the runroot and have the output
written there.
2. Have the plugin create a tar.gz with the outputs and upload them to
the hub, from where they can be downloaded.
This patch switches from option 1 to option 2.
This requires all input repositories to be passes in as URLs and not
paths. Once the task finishes, Pungi will download the output archives
and unpack them into the expected locations.
JIRA: RHELCMP-13284
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
The config repository is cloned into a path that conflicts with the
regular ostree phase. Let's use a unique name to avoid that problem.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
This patch connects the phase into the main script, and adds other
modifications:
* The archive is now stored in the images/ subdirectory in the compose.
* Documentation is updated to correctly mention that variant repos are
not available.
* Configuration for path and name of the final archive is dropped. There
are reasonable defaults for this and there's no point in having users
configure it.
* The extra message for the archive is no longer sent.
* The pungi-make-ostree utility is no longer required in the buildroot.
The pungi-make-ostree utility doesn't do any significant work. It
modifies configuration files (which can happen on the compose host), and
it starts other processes.
This patch changes the ostree-container phase to no longer need the
script in the buildroot. Instead, the utility is called on the compose
host to do the config manipulation and output the needed commands. Those
are then passed into the runroot task.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
I don't love inferring the type from the filename like this -
it's kinda backwards - but it's an improvement on the current
logic (I don't think 'dvd' is ever currently the correct value
here, I don't think osbuild *can* currently build the type of
image that 'dvd' is meant to indicate). I can't immediately see
any better source of data here (we could use the 'name' or
'package_name' from 'build_info', but those are pretty much
just inputs to the filenames anyway).
Types that are possible in productmd but not covered here are
'cd' (never likely to be used again in Fedora at least, not sure
about RHEL), 'dvd-debuginfo' (again not used in Fedora, may be
used in RHEL), 'ec2', 'kvm' (not sure about those), 'netinst'
(this is a synonym for 'boot', we use 'boot' in practice in
Fedora metadata), 'p2v' and 'rescue' (not sure).
Signed-off-by: Adam Williamson <awilliam@redhat.com>
Generally we want all packages to come from particular event.
There are two exceptions: packages configured via `pkgset_koji_builds`
are pulled in by exact NVR and skip event; and modules in
`pkgset_koji_modules` are pulled in by NSVC and also ignore events.
However, the modular content tag did honor event, and could lead to a
crashed compose if the content tag did not exist at the configured
event.
This patch is a slightly too big hammer. It ignores events for all
modules, not just ones configured by explicit NSVC. It's not a huge deal
as the content tags are created before the corresponding module build is
created, and once all rpm builds are tagged into the content tag, MBS
will never change it again.
JIRA: RHELCMP-12765
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
If the module tag contains the same module build multiple times (because
it's in multiple tags in the inheritance), Pungi will not process that
correctly and try to include the same NSVC in the compose multiple
times. That leads to a crash.
This patch adds another step to the inheritance filter to ensure the
result contains each module only once.
JIRA: RHELCMP-12768
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
```
pungi/checks.py:575:17: F601 [*] Dictionary key literal `"type"` repeated
pungi/phases/pkgset/pkgsets.py:617:12: E721 Do not compare types, use `isinstance()`
tests/test_pkgset_source_koji.py:241:16: E721 Do not compare types, use `isinstance()`
tests/test_pkgset_source_koji.py:244:16: E721 Do not compare types, use `isinstance()`
tests/test_pkgset_source_koji.py:370:16: E721 Do not compare types, use `isinstance()`
tests/test_pkgset_source_koji.py:374:20: E721 Do not compare types, use `isinstance()`
```
Signed-off-by: Timothée Ravier <tim@siosm.fr>
Allow the manifest type used to be specified in the pungi configuration
instead of always selecting the manifest type based on the koji output.
Signed-off-by: Simon de Vlieger <cmdr@supakeen.com>
If the packages are pulled from different repos and a package is already
exists in target directory, pungi raises File exists error and breaks. This
behavior can be suspended and skipped if the package is already available.
Merges: https://pagure.io/pungi/pull-request/1696
Signed-off-by: Ozan Unsal <ounsal@redhat.com>
Rather than trying to use local access when it's accessible, let user
make the decision:
* if koji_cache is configured use it and download stuff
* if not, fall back to local access
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
When a real build is downloaded, Koji can provide a checksum via API.
This commit adds verification of that checksum.
A mismatch will abort the compose. If Koji doesn't provide a checksum
for the particular sigkey, no checking will happen.
Nothing is still checked for scratch builds and images.
This patch requires Koji 1.32. When talking to an older version, there
is no checking done.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
Pungi would by default only ever add files to the cache. That would
eventually result in essentially a mirror of the Koji volume.
This patch adds a helper cleanup script. When called, it goes through
files in the cache and deletes anything that is not hardlinked from
elsewhere and with mtime not updated recently.
Cleaning up files that hardlinked from some compose would not save any
space anyway. The mtime check should account for cases like subpackage
being downloaded but not included in any compose. This would avoid it
from being downloaded over and over again.
When a compose fails or is aborted, there can be a stale lock file left
behind in the cache. This script cleans that up too.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
This patch extends the ability to download files from Koji to image
building phases too.
There is no integrity checking for the downloaded images.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
With this patch, Pungi can be configured with a local directory to be
used as a cache for RPMs, and it will download packages from Koji over
HTTP instead of reading them from filesystem directly.
The files from the cache can then be hardlink as usual.
There is locking in place to avoid different composes running at the
same time to step on each other.
This is now supported for RPMs only, be it real builds or scratch
builds.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
