Failure rates have been creeping up and the integrated gate
jobs are churning on this issue. Let's switch off voting
for now while we try to diagnose the root cause and fix it.
Related-Bug: #1536622
Change-Id: I88293243b26991e047d3dd345915896235e22345
Rename pep8 to linters job for documentation tool repositories.
Also removes unneeded '' around requirements job.
Change-Id: I7cf4766642dc3e7d7652e6d26ab2f14f9fcd7861
Keystone doesn't support python 3 yet. The job is only there
because some parts are working and we don't want them to break. As
such, there's no need to run the py34 env for stable/liberty. We
can save some test resources by not running this job.
Change-Id: I2d9285b7dcdfdbbfa1589a1450d4ea20aae11c34
For all the python jobs, always make the constraints available. This
allows projects to have py27 target use constraints without having a
parallel set of jobs.
Change-Id: Ia31bfcda7d7a7756e4fcc51773f39f20e69313df
Includes fairy-slipper and openstack-doc-tools. I don't
see clouddocs-maven-plugin but that'll be obsolete soon.
While fairy-slipper will create output for the api-site,
it also is a doc tool.
Change-Id: If26885c321dbdb60e09279fa4f43abffa5a2b55b
This change is to remove the puppet unit test jobs from fuel library for
puppet versions 3.6 and 3.7. We are keeping testing on versions 3.3,
3.4, 3.8, and 4.
Change-Id: I66daa31ed6b5909c5356d0bf35a2c3bbbbd5ed82
Use branch-git-prep instead of zuul-git-branch-prep to prepare the local
source repository, since it appears the latter does not work properly in
post jobs, yet.
Change-Id: I8cbcd49ff4bf7234e149985535f394bb686c58c3
Signed-off-by: Doug Hellmann <doug@doughellmann.com>
The large-ops overall jenkins job timeout is 60 minutes. However,
since the move to upper-constraints, and the lack of cached wheels,
we're spending 13 minutes installing python packages. This is very
often tripping use past the 60 minute timeout very barely. Bump this
up for more headroom so that we only fail if tempest actually fails,
not because pip is slow.
Change-Id: I49365b972e7095e3a65d9483181814c42de5df8b
fuel-library completly cleaned for properly support Puppet4
so let's make the job a voting job again in the check queue
only for the master branch.
Change-Id: I1f27c8c29eb92fe1a70c69929fc8ee71f68e6c44
We should post the events to #openstack-oslo channel
as well now that osprofiler is under Oslo umbrella
Change-Id: Ia00d5fb8e24a2f15d4064fa8baa7fa12c35d770d
Run the shell script from the project directory otherwise Jenkins will
fail:
/tmp/hudson7882092834424069370.sh: line 2: ./run_unit_tests.sh: No such
file or directory
Change-Id: I38d127e6b6c4d15abd41a023cefbc67b2e455655
This reverts commit 9e13d17efd2f8dea3a3bb93dc8b99145879f19bd.
The prior patch successfully removed all permissions
for registered users from sandbox. Not what I was
after with this patch.
Change-Id: I3e8bbb74b119d98a97b1fe653721f324663bbd70
Currently the gitweb access information lists sandbox's acl
as giving +2 +A to all registered users. Looking at a patch to
sandbox reveals that +2 +A is not in fact a permission all
registered users can access.
With the belief that since the all projects acl setting gives
+1 to all registered users and therefore that the project
specific sandbox acl is not overridding the all projects acl,
the patch adds exclusiveGroupPermissions to the sandbox acl
to override the all projects acl and grant +2 +A to all
registered users for the sandbox repo.
Change-Id: I4f69d9ec3fdcdc524760f153fa61b0d9108cd8c6
ceilometer tries to connect to gnocchi when started. a retry mechanism
is in place if gnocchi is unavailable. on a slow gate, the retry
limit is hit before gnocchi is started so ceilometer abandons its
attempt to connect and resulting checks fail as no data is recorded.
this patch changes it so gnocchi is started first to ensure it's ready
when ceilometer attempts to connect to it.
Change-Id: I77bc50235222422e5a0f7b6797ae0ed8640a3495
The rax-iad region fails quite often in making a reliable subnode for
a multinode job. It's failed in the gate 20 times in the last week, 12
times in the last 24 hours, and is current our top elastic recheck
bug.
The root cause is unknown, however it is heavily contributing to the
gate backlog.
Change-Id: I6b67dc236307efabed3039bc7dd695d3942585bf
Related-Bug: #1531187
CPU flags are too different from other providers; updates to
devstack-gate will fix this and we can add it back.
Change-Id: I09281c7920db0048ba190c375c1cfbfbfd552064
This job has not been passing in a long time (if ever).
It's in the experimental pipeline for tempest, but the check pipe for
ironic. This patch moves it to ironic's experimental pipeline. We'll
move it back when it has a chance of passing.
Change-Id: I1f36381c7101dfdfaca1eee9aa9ca92c9c4769b8
Puppet OpenStack group has a library [1] that is used by our modules and
contain some common ruby code.
This library really help us to:
* be consistent in how we test our modules
* avoid duplicated code
So far, we've put Rakefile and beaker helper in the gem, and we're in
the process to use it in our modules.
It's dropping lot of code! And so far we like it.
The problem is it's very easy to break the library because it has no
testing yet. If you break the library, you break all Puppet CI.
This patch aims to initiate it:
* create a job: gate-puppet-openstack-spec-helper-unit-dsvm-centos7
* run this job only on puppet-openstack_spec_helper project
* set the job as non-voting
* the job will just run ./run_unit_tests.sh so we will handle the rest
in openstack/puppet-openstack_spec_helper repository.
* this patch only focus on unit tests. Further patches will come for
beaker and eventually more bits we want to test.
[1] https://git.openstack.org/cgit/openstack/puppet-openstack_spec_helper/
Change-Id: I9c7d9efae0907e4cc4c14d76dcf44ef32f78a785
