Filip Valder
6b189c4b4a
Koji tag validation: Catch situation when tag arg has no value + refactoring + add comments.
2017-03-02 15:01:52 +01:00
Ralph Bean
ca61d6bb29
Allow passing in multiple tag names to the validation decorator.
...
We have problems when we try to wrap one decorator around another.
There are ways to do that with the
[decorator](https://pypi.python.org/pypi/decorator ) module nicely,
however.. they are ugly.
Take a look at this:
https://github.com/micheles/decorator/blob/master/src/decorator.py#L217-L218
And this:
https://github.com/micheles/decorator/blob/master/src/decorator.py#L185-L188
The approach in this PR is.. simpler.
2017-02-28 17:15:55 -05:00
Filip Valder
6dd223fd12
support for dicts + tests
2017-02-28 20:54:03 +01:00
Matt Prahl
7dfb647e08
Merge #347 Possibility to specify optional parameters when submitting build
2017-02-28 14:26:20 +00:00
Filip Valder
a4366d9be7
Koji tag validation
...
- Decorator supporting str/list args
- ProgrammingError exception which may uncover typos in the names of args to validate
- Configurable whitelist of koji tag prefixes
- Add basic validation to test_build.TestModuleBuilder
2017-02-28 09:08:32 +01:00
Jakub Kadlčík
cef690f260
Not accept copr parameters when different builder is used
2017-02-27 15:53:39 +01:00
Jakub Kadlčík
34ea4a8efe
Move validate_optional_params function to utils.py
2017-02-26 22:23:50 +01:00
Matt Prahl
56d6d90ece
Enforce that the mmd can't define a different module name, stream, or version
2017-02-24 10:37:40 -05:00
Jakub Kadlčík
38bdc92248
Have immutable default value for optional_params
2017-02-23 20:51:31 +01:00
Jakub Kadlčík
92dce63091
Allow submitting optional parameters such as copr_owner and copr_project
2017-02-23 20:51:31 +01:00
Matt Prahl
385045d778
Reuse components from previous module builds
2017-02-21 14:22:30 -05:00
Matt Prahl
a5cc4eb280
Add a module's commit hash, scmurl, and the buildrequires' commit hashes, stream, and version in the modulemd
2017-02-21 09:53:05 -05:00
Matt Prahl
26c08f5e31
Separate record_component_builds into two functions
2017-02-21 09:32:09 -05:00
Jan Kaluža
18f9089643
Merge #323 Possibility to submit yaml file
2017-02-21 14:21:45 +00:00
Jan Kaluza
1d8bc3e33a
Use concurrent.futures instead of our own ThreadPool implementation
2017-02-21 08:02:50 +01:00
Jan Kaluza
aa6db9fcce
Use StopIteration in QueueBasedThreadPool to stop the threads.
2017-02-21 08:02:50 +01:00
Jan Kaluza
e7acde9aa5
Use six.moves to import Queue
2017-02-21 08:02:50 +01:00
Jan Kaluza
64d96966e2
Use Queue based ThreadPool instead of the default Python's one. Improve locking and debugging in the Mock backend.
2017-02-21 08:02:50 +01:00
Jakub Kadlčík
43acbdbdba
Implement possibility to submit yaml files (See #310 )
2017-02-19 22:35:55 +01:00
Jakub Kadlčík
1591b9c5e7
Create method for loading modulemd object from yaml
2017-02-19 22:35:55 +01:00
Ralph Bean
9bb7b8a716
Force MMD's int version to a str when interacting with our db.
2017-02-15 12:09:49 -05:00
Filip Valder
818a3feb6f
check Koji active tasks for component relicts (same NVR) from previous (terminated) builds
2017-02-08 15:56:41 +01:00
Jan Kaluza
0b307550c4
Mock: WIP threading support
2017-02-01 13:20:17 +01:00
Jan Kaluža
59d070df35
Merge #278 Fix #276 - Use PDC to resolve buildroot components instead of MBS database
2017-01-12 14:34:13 +00:00
Jan Kaluza
008f2c9d29
Make SCM check faster by consolidating two SCM checks to single check executed in ThreadPool. Use 'git checkout' instead of HTTP HEAD request, because it is much faster in Fedora cgit case.
2017-01-11 15:09:49 +01:00
Jan Kaluza
0f797c2917
Remove fake base-runtime module from DB and old resolve_profiles method because they are not used anymore. Fix the tests to work with remove fake base-runtime.
2017-01-10 15:58:26 +01:00
Matt Prahl
261549c8c2
Migrate scheduler to be run by fedmsg-hub
2016-12-15 16:27:15 -05:00
Jan Kaluza
3788bd2d8d
Inherit the SCM repository from the parent module to included modules.
2016-12-13 13:24:25 +01:00
Jan Kaluza
e08b6d1a46
Allow resubmit of the same module also for 'init' status.
2016-12-13 10:59:10 +01:00
Jan Kaluza
d06c13d973
Fix the bug when module build freezes when all the components in batch have been already built by builder.
2016-12-12 21:03:44 +01:00
Matt Prahl
9d734d38ce
Create fake baseruntime module directly to avoid DetachedInstanceError
2016-12-12 09:00:44 -05:00
Filip Valder
7fcfe05f85
Fix issue #251 :
...
- support for mupltiple SCM URLs (logic was broken when there were more
than one SCM URL in SCMURLS conf. var)
- unsplit-brain handling of SCM URLs between frontend and backend
2016-12-12 13:05:27 +01:00
Filip Valder
79715ecfbe
missing: import logging
2016-12-05 15:18:26 +01:00
Ralph Bean
9303be7a4e
Merge #228 Drop alembic fake base-runtime code.
2016-12-01 15:56:10 +00:00
Ralph Bean
a2122f84a4
Move insert_fake_baseruntime to utils.py.
2016-11-30 13:11:14 -05:00
Jan Kaluza
f7350a395f
Allow file:// URLs when building modules locally
2016-11-30 10:33:20 +01:00
Jan Kaluza
4f6d683645
Fix #173 - Handle includedmodules by building their components recursively.
2016-11-23 06:38:16 +01:00
Jan Kaluza
528b6ab20a
Do not submit failed components for a build
2016-11-21 13:07:18 +01:00
Jan Kaluza
9aca8057a3
Handle the exception in builder.build(...) code and do not mark whole module as failed - mark just the particular component.
2016-11-18 19:09:03 +01:00
Filip Valder
8c6a5f4919
Merge #196 Fix #94 - handle IOError
2016-11-18 14:11:20 +00:00
Filip Valder
d8942ba44a
Load defaults from config
2016-11-18 10:24:34 +01:00
Filip Valder
225d116d45
lint
2016-11-18 10:22:35 +01:00
Ralph Bean
411da3def5
Resolve component refs on build submission.
...
Fixes #203 .
The modulemd files we had used to all pin the commit hashes of the
components they were pulling in. We changed this with modulemd-1.0 and
now modules can define their components in terms of their *stream*
(a.k.a. their dist-git *branch*). This is awesome, and is a big
stepping stone towards greater things.
The problem is that we used to take that git ref literally from the
modulemd and pass it on to koji. Now that instead we are passing on a
branch name to koji, it falls down. Koji requires us to send it an
explicit commit hash.
This change takes that branch name and converts it to an explicit commit
hash when the module is first submitted.
Verified that things are working again:
- http://koji.stg.fedoraproject.org/koji/taskinfo?taskID=90084980
2016-11-17 14:49:48 -05:00
Matt Prahl
244b77f54c
Add component build throttling
2016-11-16 15:24:17 -05:00
Matt Prahl
67d07023c5
Rename function to start_build_batch for further flexibility in the future
2016-11-16 15:24:16 -05:00
Matt Prahl
fb043cd2c7
Add TODO comment
2016-11-16 15:24:16 -05:00
Jan Kaluza
2b7ebdfc6f
Fix #172 - Set ComponentBuild.batch according to mmd.buildorder attribute
2016-11-16 12:41:06 +01:00
Jan Kaluza
966a73251a
Add support for modulemd-1.0 backward-incompatible changes like commit being replaced by ref
2016-11-14 15:32:51 +01:00
Jan Kaluza
7b9e4f7234
Initial version of mock backend
2016-11-10 09:57:00 +01:00
Ralph Bean
99b0993fe0
Pass around more info on pre-built components so we can resume module builds.
2016-10-29 16:12:24 -04:00
