- update setup.cfg to advertise supported platforms
- fix tox.ini file use version agnostic envlist
- add openstack-tox-py39 job
- minor fix on test file which wrongly assumed that pop
returns a new dictionary
- added type to release dictionary, which is what would
have prevented previous item from being introduced.
Change-Id: I60b80ebe41a66d39e6f1a51a3d00c0b9292d42fd
- assure we test triple-jobs-gerrit.py file
- fix division by zero bug with above file
- simplified tox.ini file
- prevent accidental regression on CI by checking number of passed tests
- removed test_molecule.py file which was the precursor of what later
become pytest-molecule
Change-Id: Id6ffb9df88a376bcd3c8a0d77b2d378fca8da115
Assures consistent formatting of our python codebase without having
to rely on humans to do it, or to debate during reviews.
Change-Id: I1e62cc755fa60e453dea865f436241ecae330771
Flake8 went broken recently, adopts correct calling method to avoid
regression and fixes broken rules.
Also removes openstack-tox-pep8 from gate which was a guaranteed
failure when triggered as pep8 was removed.
Change-Id: Iba03d3051952f461bb16299a6d32475b2be43041
This was removed from tht https://review.openstack.org/#/c/608311
so lets also remove the job definition
Change-Id: I83c91e22b91fce7232e1b9124b9d969e099b314a
Depends-On: I9c2cda2bf330c1ec93e756708e9d927595dadfd8
Includes removal of file pattern from pep8 job which prevented
it from running on most changes. Linters will use multiple tools
and a pattern would make it fail to spot errors.
This check should run even if no files are changed, so it would
be able to check commit messages.
There are no real load implications because linters check is
just a simple tox execution which has minimal resource requirements.
Includes reactivaction or linting which was not running on
the entire repository. This required few minor fixes.
Partial-Bug: #1786286
Change-Id: I4d4279309af55c2663e98bd0cdec9224f03c1fa0
multinode-upgrades and ovb-containers don't run anymore, let's
remove them from jobs list in tripleo cistatus page
Change-Id: I0748808db54daeeacecf3f1c38a29872287d514a
* Remove fs024 job, not run on master anymore
* Add fs035 (currently named ovb-ha-ipv6)
* Remove ovb-fakeha-caserver (experimental job that always fail)
Change-Id: I389c951df9caec3220c011cd8367cde68af79328
In the zuul v3 switch, a number of things changed that broke the
cistatus page. The reviewer name is now Zuul instead of Jenkins,
the job names are legacy-job-name instead of gate-job-name, and not
all of the review comments include a pipeline name. This change
handles all of those differences and allows us to retrieve job
results again.
Note that it will no longer handle older Jenkins-based jobs. Given
that it will be all zuulv3 in the future I didn't think it was worth
the effort to support both simultaneously.
Change-Id: I494aff33be6141b2ff82595be75bfe20e6a7e382
These are arguably more important than the check or experimental
jobs since failures here require two more ci passes before patches
can get through. Previously we were filtering them out though.
Change-Id: Ic5719965de2831406adaded91cd01eebc5e31809
Previously only check-tripleo and check queue jobs were read. We
have some experimental jobs that were added to the status page that
need to be supported as well.
Change-Id: If38784ceae618557a977ecb8352ca79915b19d5a
In general we don't care about the voting or non-voting status of
a job on the cistatus page. We just want to see whether a job is
passing or not. There are also some jobs running in both voting
and non-voting form depending on the patch, which further complicates
things.
This change makes the job name list use only the voting form of the
names and handles the -nv form automatically if it is found. In
addition to getting us all the relevant results for a job in one
column, this means we won't have to update the cistatus script
every time a job moves from non-voting to voting or vice versa.
Change-Id: I69294bf0e78d5e443e4b1c869991fc324c2d7c23
There are almost no results for these jobs on the status page at
this point, so there's no reason to include them by default.
Change-Id: I63ef153e03bf59133fcffb1b01fb4512ba3bb227
A normal Jenkins message about a job status looks like:
gate-tripleo-ci-centos-7-ovb-ha SUCCESS in 1h 49m 28s
A RETRY_LIMIT message looks like:
gate-tripleo-ci-centos-7-ovb-nonha RETRY_LIMIT
Because we get the duration from this message, the latter causes us
issues parsing the message. This change makes that code handle
the case where there is no time included.
Change-Id: Ieb75e6dc43fb6fb4c8b86f399a23cd26978ba7d2
This is now running in non-voting form on all patches to verify its
stability. Adding it to the status page will make it easier to
keep track of how often it's passing or failing.
Change-Id: Ic3eeeafe567bd3ae2e0c5abb9129b472dbb9bc2d
Include scenario004 to logs. The scenario is not ready yet but will be
shortly. We'll have logs in the dashboard.
Change-Id: Id349ffa5847c2aaa59087affb133340cc08c8051
Roll up the results into three categories: passed, partial failures,
and complete failures. Passed means all reported jobs passed,
partial failure means some jobs passed and some failed, complete
failure means all jobs failed. Note that these results will be
somewhat transient for patches with jobs still in progress.
I think having these numbers easily available will be helpful in
telling us the health of CI. Passed jobs are obviously good,
complete failure jobs are usually legitimate failures (unless there's
a complete CI outage), and partial jobs are usually where you would
look to find intermittent problems (i.e. the hard ones). Keeping
the partial number low should mean TripleO is pretty healthy from a
test perspective.
Change-Id: I5a1dd0f31c26f9d0b41fef9268000fb2f3dc54d2
Because of the number of jobs we're reporting on the ci-status page
now, I see some weird line breaks in the middle of the duration.
This makes it difficult to visually scan job durations. For example,
one job on the page currently looks like this:
397267,1 - 1h
52m log
Since there are already three lines in the first column of the
table, let's just make each job report two lines explicitly so they
don't get broken up in weird ways. This change would result in the
following format for the job above:
397267,1
1h 52m log
Change-Id: Ia067ed16276a1b7ec4e08e636968467d5eacf752
Add the 3 scenarios jobs in TripleO status dashboard on:
tripleo.org/cistatus.html
Also reduce the jobname in each job column by removing
"gate-tripleo-ci-centos-7-".
Change-Id: Ia3a9f9fb09aef37d0eafed4a364f00fa10ccf6b0