In the mixed rhel component job the rhel8 container update
prep fails due to yum update with the rhel9 undercloud repos
causing related-bug.
This disables yum update for rhel8 content in this job
Related-Bug: https://bugzilla.redhat.com/show_bug.cgi?id=2154179
Change-Id: I3a9ac11e7a3c3e5d3e32ec47b06dfca76e31a727
Ping test is no longer being used in the last 5 years and was replaced
by tempest. However there are still pieces of code being executed, and
replacing some files, for example the testrepository.subunit file that
tempest generate and it is being used by the ci logging scrapper to
collect data about tempest.
Change-Id: Ia2b18f88652f144d5b593c5d209a9acf2d0374c8
Currently artg_repos_dir path points to
/home/zuul/src/opendev.org/openstack.
On EL9, we have /home/zuul and /home/zuul-worker dir
exists and default user is zuul-worker.
artg_repos_dir is looking for clonned repos in /home/zuul path but repos
are clonned in /home/zuul-worker, leading to failure.
More info about issue is here:
https://review.rdoproject.org/r/c/testproject/+/35229/23#message-265a04d5718b866a2ecf508af68df2e4159af457
Signed-off-by: Chandan Kumar (raukadah) <chkumar@redhat.com>
Change-Id: Iea958e037ab63896def82608bc2ce2ba0d30948c
Some variables like tempestmail_log_server and tempest_track_resources
are no longer required because it belongs to validate-tempest role, that
is deprecated now. Other variables like tempest_test_image_path and
tempest_tempest_conf_overrides should be used only when required. In the
case of these testenv, this is overriding what is in tempest.yml task in
tripleo-quickstart-extras, making it hard to change cirros image.
Removing here will have no impact, and all common tempest variables
should indeed be centralized in tempest.yml task.
Change-Id: I66da92fcabf204ab9ed3653132ee5221c220c5a0
Currently we're getting rate limited and until recently we weren't
actually leveraging the cache. In order to work around this, let's stop
funneling all our docker.io requests through a single IP.
Change-Id: Ia39e562bbb76b5c13e559b4e4d499123a6531ac8
Related-Bug: #1889122
Docker host config is configured in release files and depends on
release, not on environment. It's configured in dependency patch
instead.
Depends-On: https://review.opendev.org/737234
Closes-Bug: #1884371
Change-Id: I3e2b57c87e681fa5e1bb3502e3313cafffb20809
- Collect /home/{{ undercloud_user }}/container-builds/ which will be
useful when a job has built the container images instead of pulling
them from a remote registry.
- Configure the docker_registry_host variable in the testenvs.
If a job is configured with build_container_images set to True
(default is False), the conntainer images will be built prior to
deploying TripleO.
Change-Id: I49895bc8632656a78d107bf9d418f865161bc9f9
The url download.cirros-cloud.net is blocked
from rdo nodes. Create a backup of the image
on images.rdoproject.org
Related-Bug: 1867807
Change-Id: Ibe5ff806539479fb94d4a61aff9ac06a883a6895
leave gzipping on non-OVB jobs in 3party as it allows to
browse gzipped files.
Closes-Bug: #1861378
Change-Id: I4f6aaba81c91bc015d67c9c03de44f09ed0a66b1
Recently infra disabled the generation in zuul of ara html for
performance issues in the logging strategy.
This change reenables the generation from collect logs role
Change-Id: I4a4a6599aacaa32a6901b4e2e5497ea7b50e39cd
These changes were missed by Sysadmin's proposed
Migration Patch, this patch takes care of these
missings.
Also updated kolla review pull task to use
review.opendev.org instead of opendev.org as
remote url because opendev.org currently don't
have review refs/changes synched due to gitea bug.
Related-Bug: #1825732
Change-Id: I30646b0c5b95f391e9ab4bd767b468280ccb3842
Use the new registry provided by Infra which is Docker Registry v2 and
not v1 anymore. I'll probably be more stable for us when doing HTTP calls
against the registry.
Change-Id: I8294d734984fee721c3d856c40fc4795288bd060
Related-Bug: #1819632
atop is 25mb and our logs now have
over 100mb in containers-multinode.
Reducing the overall log size in upstream
Related-Bug: #1817600
Change-Id: I245a5fa5be85fe6bc8b87144d5ddb531a41d4993
This using just 127.0.0.1 seems to work just fine
in upstream and rdo jobs w/ unbound. When users need
to replicate the job in their own infra they may run
into issues.
The standalone jobs update the dns server as part
of the standalone deployment. Add a redundant dns
server is the only way to ensure both upstream and private
workflows will work consistently.
Related-Review that introduced unbound and localhost
https://review.openstack.org/#/c/494545/
In the case of local reproducers using libvirt the
image does NOT have unbound cache.
Change-Id: I58917eb84217958afd5cba1fa77ce0032aac4255
Multiple versions of ansible are causing jobs
to fail in the upstream gate. Let's ensure it's
working correctly before enabling upstream
Related-Bug: #1795009
Depends-On: Ie9f5d3b6380caa6824ca940ca48ed0fcf6308608
Change-Id: I367cd9afac400fc277d5c6b7b455fafe1725169a
- OVB and multinode/singlemode jobs are passed different settings
files based on the location where the jobs run. This review
changes toci-gate-test to pass both the general and rdocloud
multinode setting files as is the case with OVB.
- Removes the period job registry definitions.
These should be passed in release files only.
- Sets update_containers to run with
containerized_overcloud or containerized_undercloud
Closes-Bug: #1775874
Change-Id: I8f498073f1df9de70372f09b8c43e0d4adb5b42a
We'll be switching this to true by default, so we need to set the
undercloud clean option to False for CI.
Change-Id: I70ac9bbe57d7078a1aa2eb8b07503efb8ba5bbfe
Needed-By: I027a76d2f37a5af5bd0a599783287632298acf3f
Related-Bug: #1761810
container updates have been turned on for the
3rd party tripleo jobs in software factory [1] since
Jan 24 2018. I believe this should be reliable
enough for upstream at this point.
[1] https://review.openstack.org/#/c/529399/
Change-Id: I8868bf6974fdf268b9deed84c69ff0479df3f080
This review moved MTU settings to the testenv files
and adds settings to adjust the Neutron settings
accordingly on the undercloud and overcloud.
Depends-On: I3106137cdb216ba593e4788f2a7f319669c8081c
Change-Id: I10a2a8df25e3110b6584b47dca693c19abeb4930
zuul-cloner is now checking out changes to
"/home/zuul/git.openstack.org/openstack"
Closes-Bug: #1737042
Change-Id: I9739dbcac551f234b595060cf257be91247a5d65
This reverts Ifcefa1cbfb05667d124d5cfcdcfc8ff6ccd4e631 now that the
migration to zuulv3 is complete and update the path to match the new
location.
Change-Id: I7b9f512199cbab10413ae35096a04697b80622a5
It's a loss of time and useless since we have no access to the machines
to debug if needed.
Change-Id: Ia4291c18c8b4fe331b0b8b09eb605abc0776f7e0
Co-Authored-By: Alex Schultz <aschultz@redhat.com>
Related-Bug: #1659065
In Zuul v2 and v3 cached images are in different paths, so it's
impossible to configure job to be compatible with both versions.
Use temporary the image from URL until one version will be active.
Change-Id: Ifcefa1cbfb05667d124d5cfcdcfc8ff6ccd4e631
This commit reverts the workaround that was the second run of repo-setup
on overcloud noeds. This workaround was a short-term fix to allow stable
releases to use CI, however a better fix has been created involving the
refactoring of the vxlan networking code that is dependent on this
patch.
Change-Id: Ibcb7d4e0ef603b94a9b8f59b8f81450919c1a371
Use local cached cirros image for tempest testing, don't download
it every job from cirros cloud.
Change-Id: Ifc6c10bbb809a13182712322435ae47330cf4bb5
Depends-On: Idc979d60e7ed5d36646d2a9c3fd41e5f85ee15ca
Use current variables for set up multinode settings, release
extra args for using it in featureset files.
Depends-On: Ia409617cab5a31f19875844c676dad87518977fc
Change-Id: I341995fa21ce759104a0b3431f6d434f93a5fdcc
The current method of including the multinode common config does
not allow for overriding anything set there. This is because
quickstart.sh will always put "-e" args after the the other config
args.
This patch instead includes it at the top of the multinode playbook.
Anything passed on the command line will always override variables
included this way.
This also moves some variables out of the multinode testenv config
that are not specific to tripleo-ci. This patch depends on a
quickstart patch that puts these variables in the common multinode
config.
Change-Id: Ic4fccd4e71224149e20558af50a8b4f84f0ed67a
Depends-On: I7fd723b5547896693e57c76415ab98b98abfdcfc
Depends-On: I55f705a390ae5b927a9fb0f8d8ec153ea61cc3c9
All the upstream multinode and ovb jobs should be
configured to use the appropriate docker regsitry
settings.
Closes-Bug: #1710533
Change-Id: I9b3ac36033c0b9e994a13188f750e9e1d484bd2d
According to bug info we have unbound dns server with cache
installed on the image, so we should use it instead of Google DNS.
Change-Id: Ic7624aaf352ec2082417486d42df9cee18093da5
Closes-Bug: #1711262
ssh_user was not explicitally set to undercloud_user for oooq multinode
jobs. It was not possible to reproduce oooq multinode jobs without using
the hardcoded stack user.
This makes the variable default to undercloud_user, which defaults to
$USER.
Change-Id: Ia751d3a5a03b971485376aed3a0400e401af6fa8
In case we need net iso with multinode jobs. Follows what's been
introduced in change If3989f24f077738845d2edbee405bd9198e7b7db.
Change-Id: I9011f14452b71b91f1280433000f2ef8a97585af