|Age||Commit message (Collapse)||Author||Files||Lines|
The %clean section has been unnecessary since RHEL6.
This was originally needed for RHEL5, where we continued supporting and
building the client several years after we dropped support in the
Nowadays, we are always building all subpackages everywhere, so we can
simplify the spec a little by removing these conditionals. They will
(hopefully) not be needed again in future.
The server code imports and uses both the yum Python package and the
rpmUtils Python package which is also provided by yum. rpmUtils is used
at build time too due to the Sphinx docs.
In the latest Fedora releases, packages are forbidden from using
/usr/bin/python (as opposed to /usr/bin/python2) and starting from
Fedora 29 it is no longer available in the build root unless you
explicitly depend on it.
Switch to invoking /usr/bin/python2 everywhere instead.
These package renames took place gradually over several Fedora releases,
however from Fedora 29 onwards the old names are no longer provided so
we have a hard cutoff where we need to use the new names.
Fixes this failure with the tests:
ERROR: test suite for <module 'bkr.inttest' from '/usr/lib/python2.7/site-packages/bkr/inttest/__init__.pyc'>
Traceback (most recent call last):
File "/usr/lib/python2.7/site-packages/nose/suite.py", line 209, in run
File "/usr/lib/python2.7/site-packages/nose/suite.py", line 292, in setUp
File "/usr/lib/python2.7/site-packages/nose/suite.py", line 315, in setupContext
File "/usr/lib/python2.7/site-packages/nose/util.py", line 471, in try_run
File "/usr/lib/python2.7/site-packages/bkr/inttest/__init__.py", line 431, in setup_package
File "/usr/lib/python2.7/site-packages/bkr/inttest/__init__.py", line 180, in start
File "/usr/lib/python2.7/site-packages/bkr/inttest/__init__.py", line 190, in _wait_for_listen
File "/usr/lib/python2.7/site-packages/bkr/inttest/__init__.py", line 134, in check_listen
File "/usr/lib64/python2.7/subprocess.py", line 394, in __init__
File "/usr/lib64/python2.7/subprocess.py", line 1047, in _execute_child
OSError: [Errno 2] No such file or directory
When using MySQL, the db could handle being passed a User object and
would truncate it appropriately. There was prexisting code to truncate
the new/old values but it only expected unicode and was skipping User
objects (Which is what is passed to Activity when a system owner is
changed). As MariaDB is not so leanient this causes failures. This fix
will cast all objects passed in to unicode and then truncate them; as
was originally intended.
This was added way back in commit d917d76a because we observed that
MySQL appeared to be truncating string values in the middle of UTF-8
characters, leaving behind invalid bytes.
The *actual* cause of that was because Beaker was connecting to the
database with latin1 encoding. The column was 60 chars so MySQL
truncated the string at 60 *bytes*.
After we switched the database encoding to utf8 in commit f47cf808, this
was no longer an issue -- MySQL would correctly limit the field to 60
Unicode *characters* instead of 60 bytes.
The current distro import code will not be able to import RHEL8 Alpha composes
synced to partner repositories. What complicates this is a different directory
structure between partner repos and internal repos.
This patch adds support for importing RHEL8 alpha internally and for partners
adding code which tries to guess the directory layout.
The `autofetch` method will now wait for a response from beaker server
before sending another ajax request. Requests will also timeout after
10 minutes, then be retried.
To prevent deadlocks with other transactions which could be concurrently
updating the recipe row, such as update_dirty_jobs().
This patch allows to import RHEL8 Alpha composes. The compose paths are hard
coded and any deploys which provide the compose should follow the hard coded
naming scheme of folders in order for the importer to import the compose.
Reasons for this are, that once available the name of the composes should and
will not change.
This is to try and avoid pointless vacation auto-replies going to
Similar to commit 48cda168, beaker-log-delete may run for a long time if
it has a lot of jobs to purge. We want to avoid having an active
transaction for the entire duration of the run, because it puts extra
strain on Innodb and may contribute to the likelihood of bug 1574772
With the switch to MariaDB, the database is less tolerant of different types
than what is declared in the schema. The production instance is running in
strict mode, which enforces this schema. MySQL would simply coerce the string to
an integer depending on what you have passed in, MariaDB raises an error.
In order to keep backwards compatibility, we do in Python what MySQL did.
The regex for kernel oops' was too aggressive, so it has been replaced with a
moderate one, based on data from oops.kernel.org.
Unit testing was added for the PanicDetector.