Repository navigation
Conversation
Qodo reviews are paused for this user.Troubleshooting steps vary by plan Learn more → On a Teams plan? Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center? |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Build failed (check pipeline). Post ✔️ openstack-k8s-operators-content-provider SUCCESS in 4h 43m 13s |
61b7f11 to
d3899f3
Compare
|
Build succeeded (check pipeline). ✔️ openstack-k8s-operators-content-provider SUCCESS in 3h 21m 42s |
d3899f3 to
b716867
Compare
|
Build succeeded (check pipeline). ✔️ openstack-k8s-operators-content-provider SUCCESS in 8h 06m 36s |
b716867 to
c0a41d1
Compare
|
Build failed (check pipeline). Post ✔️ openstack-k8s-operators-content-provider SUCCESS in 2h 32m 47s |
… skip
RHOSO 19 content is currently published by Konflux only as a bundle
image, not as an opm/FBC index, so it cannot be installed through the
existing CatalogSource(image: ...) path. This adds the pieces needed
to install it out of band instead (e.g. via `operator-sdk run bundle`)
while leaving the rest of the OLM/MetalLB/NMState/cert-manager flow in
install_operators.yml untouched:
- A new pre_operator_install hook point in deploy_architecture.yml,
firing right before the OSP operator install stage, following the
same run_hook pattern already used elsewhere in this role.
- cifmw_kustomize_deploy_skip_openstack_operator_catalog (default
false): when true, strips just the templated
openstack-operator-index CatalogSource and openstack-operator
Subscription from the generated OLM manifest before it's applied,
so an out-of-band install doesn't fight a broken/placeholder one.
- The existing "wait for Subscription"/"wait for InstallPlan" tasks
further down in install_operators.yml also skip when this var is
true: they look for the openstack-operator Subscription inside this
kustomize apply's own result set, which no longer exists there once
it's stripped out, causing an undefined-variable failure ("No first
item, sequence was empty"). The out-of-band installer is responsible
for waiting on its own Subscription/InstallPlan instead.
A downstream hook (ci-framework-jobs' bundle_operator_install.yaml)
needs a QE-only registry credential file that a Zuul pre-run play
writes to /var/tmp/qe-secrets on the original controller node. Once
the reproducer role hands execution off to controller-0 (the CRC node
it provisions), that node has no network path back to the original
controller, so the file becomes permanently unreachable from any later
hook or role. Add cifmw_reproducer_sync_qe_secrets (default false) to
the reproducer role: when true, it syncs a QE secrets directory onto
controller-0 while still running on the original controller, at the
same point push_code.yml already pushes repositories there - the last
point where that's possible.
Bundle installs also have no openstack-operator-index pod, which
env_op_images' log collection reads OPENSTACK_IMG from; its "first"
lookup then fails and, since the role runs inside an include_role,
aborts the rest of log collection. Add
cifmw_env_op_images_allow_missing_operator_index (default false): when
true, that lookup is skipped instead of failing. The value is already
optional where it is consumed (`| default({})`).
Signed-off-by: Arx Cruz <arxcruz@redhat.com>
c0a41d1 to
618b513
Compare
|
Build failed (check pipeline). Post ✔️ openstack-k8s-operators-content-provider SUCCESS in 2h 42m 24s |
This branch has not been deployed
1 participant
RHOSO 19 content is currently published by Konflux only as a bundle image, not as an opm/FBC index, so it cannot be installed through the existing CatalogSource(image: ...) path. This adds the pieces needed to install it out of band instead (e.g. via
operator-sdk run bundle) while leaving the rest of the OLM/MetalLB/NMState/cert-manager flow in install_operators.yml untouched: