Add octavia service to SKMO central and leaf regions by vakwetu · Pull Request #759 · openstack-k8s-operators/architecture

vakwetu · 2026-05-14T23:23:37Z

Add OVN Octavia networking for SKMO multi-region deployment
Compose SKMO overlays on multi-namespace bases with small components for
Octavia NNCP, EDPM octbr bridge mappings, and network-values patches.
NNCP stages reuse base values.yaml templates; octavia-network-values
component keeps kustomize load-restrictor-safe paths for CI kustomize_deploy.

Signed-off-by: Ade Lee alee@redhat.com
Assisted-by: Claude Sonnet 4.6 noreply@anthropic.com

openshift-ci · 2026-05-14T23:23:44Z

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: vakwetu
Once this PR has been reviewed and has the lgtm label, please assign abays for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

vakwetu · 2026-05-14T23:27:46Z

hey @beagles and @gthiemonge . I'm adding octavia to the skmo ci job. Still testing - but can you confirm I've done everything I need to do. Thanks.

centosinfra-prod-github-app · 2026-05-26T18:35:15Z

This change depends on a change that failed to merge.

Change openstack-k8s-operators/ci-framework#3965 is needed.

fultonj · 2026-06-01T13:35:46Z

I'd like to hear from the octavia experts on this patch first.

fultonj · 2026-06-08T09:22:24Z

Waiting for octavia experts to review

gthiemonge · 2026-06-08T09:53:53Z

/lgtm

abays · 2026-06-08T10:35:41Z

recheck

Logs for old failure no longer exist. Let's try again and review logs if necessary.

abays · 2026-06-08T10:43:44Z

If @vakwetu is satisfied with testing, LGTM

Add EDPM recreate hook and extend network-values templates for Octavia NAD generation. Leaf-region Tempest config lives in ci-framework-jobs 05-tests.yaml; reproducer uses post_tests: [] like the uni job. Depends-On: openstack-k8s-operators/architecture#759 Signed-off-by: Ade Lee <alee@redhat.com> Assisted-by: Claude Opus 4.6

centosinfra-prod-github-app · 2026-06-08T22:19:13Z

Build failed (check pipeline). Post recheck (without leading slash)
to rerun all jobs. Make sure the failure cause has been resolved before
you rerun jobs.

https://gateway-cloud-softwarefactory.apps.ocp.cloud.ci.centos.org/zuul/t/rdoproject.org/buildset/7c321be332614987b0860e7e422d6d66

✔️ noop SUCCESS in 0s
❌ rhoso-architecture-validate-multi-namespace-skmo FAILURE in 5m 00s

Deydra71 · 2026-06-22T13:42:19Z

@abays it did, but not successfully yet. Tho it's failing on federation part after successfully configuring Octavia. There's currently a federation bug that's being fixed --> openstack-k8s-operators/ci-framework#4006 so if we want full test project run we need to wait a bit

abays · 2026-06-22T13:54:26Z

+kind: Component
+
+components:
+  - ../../../examples/va/multi-namespace-skmo/components/octavia-network-values


This is an anti-pattern in regards to the architecture repo layout. Rather than this kustomizaton.yaml depending on the values from the examples dir, the ../../../examples/va/multi-namespace-skmo/components/octavia-network-values/kustomization.yaml file should depend on this dir (va/multi-namespace-skmo/networking-octavia) and should include its own octavia-network-values.yaml in its list of resources.

abays · 2026-06-22T14:11:00Z

The Octavia replacement blocks in this file and examples/va/multi-namespace-skmo/control-plane2/kustomization.yaml are identical. While not a blocking issue, it might be cleaner if that config could be extracted to a higher-level common kustomization.yaml that both sub-dirs could then include as a resource/component. In fact, it could be placed in va/multi-namespace-skmo and moved out of examples entirely, which would make the inclusion cleaner too.

Deydra71 · 2026-06-23T14:23:26Z

Had to move octavia-ca-passphrase Secret from secretGenerator in va/octavia-controlplane to a direct resource file in each control-plane directory so the NamespaceTransformer correctly sets the namespace

Deydra71 · 2026-06-24T08:57:55Z

Had to disable edpm nodes FQDN validation

abays · 2026-06-29T11:14:03Z

When this is built (via [1]), there is no associated NamespaceTransformer. As a result, it has no metadata.namespace, which could be a problem.

[1] https://github.com/openstack-k8s-operators/architecture/pull/759/changes#diff-75710097274171eb9c7de73d522ead572beeb47ab2ee818a8d458a14359ddd4f

Understood, will add the namespace directly

abays · 2026-06-29T11:14:45Z

When this is built (via [1]), there is no associated NamespaceTransformer. As a result, it has no metadata.namespace, which could be a problem.

[1] https://github.com/openstack-k8s-operators/architecture/pull/759/changes#diff-7a39d8689d0ab40481d5adb8501faa7ba4317ce05bb6b3754f1b371f886db5d8

abays · 2026-06-29T11:23:59Z

*values.yaml should ideally exist in the examples dir and not in va, dt or lib dirs. The *values.yaml files are exposed in the examples dir as the interfaces for user customization.

abays · 2026-06-29T11:24:45Z

*values.yaml should ideally exist in the examples dir and not in va, dt or lib dirs. The *values.yaml files are exposed in the examples dir as the interfaces for user customization. Perhaps you had a specific reason for doing this though, as I see it was done a few times with different *values.yaml files. Could you explain the rationale behind it?

No rationale, I moved them to va/ during the first refactoring to build the octavia-controlplane component chain, but didn't realize it belongs examples dir by convention, will move.

abays · 2026-06-29T11:28:44Z

+  octavia:
+    dnsDomain: octavia.example.com
+    mtu: 1500
+    vlan: 23


This vlan: 23 is defined here and consumed in [1]. But if this value was changed to something else, I see that it would still be hard-coded to 23 here [2]. Not a blocking issue, but just something to be aware of.

[1] https://github.com/openstack-k8s-operators/architecture/pull/759/changes#diff-f56986a4376344555d922a11476626568461efa2c013fcf4ecd8720cbe518209R141-R150
[2] https://github.com/openstack-k8s-operators/architecture/pull/759/changes#diff-78a76258db67b22a42639a642b302d2e51593283370201c3da20951ebbaad255R56

abays · 2026-06-29T11:30:30Z

Is this missing a resource or component to add the Octavia interface to the OCP nodes via NNCPs?

I think this is by design, Octavia bridge is configured by central region NNCP only, since all OCP nodes share the same network, so only one is needed (?)

That could be. I defer to you and Ade, so do whatever you think is right.

abays · 2026-06-29T11:32:42Z

+  ovn:
+    ovnController:
+      nicMappings:
+        datacentre: enp9s0


I don't see this being used anywhere.

It's not, forgot to remove after refactor

abays · 2026-06-29T11:34:23Z

+  ovn:
+    ovnController:
+      nicMappings:
+        datacentre: enp9s0


Where is this used?

abays · 2026-06-29T11:41:44Z


        edpm_nodes_validation_validate_controllers_icmp: false
        edpm_nodes_validation_validate_gateway_icmp: false
+        edpm_nodes_validation_check_for_fqdn: false


Was it intentional to change the non-SKMO multi-namespace VA here as well?

I see why it's wrong now, since it's shared it's diabled for the other VA as well... I will add the edpm_nodes_validation_check_for_fqdn: false override to multi-namespace-skmo/components/octavia-edpm-bridge/edpm-octavia-ansible-vars.yaml

Actually that probably won't work either, it doesn't appear in the rendered NodeSet output.

Can we add a replacement in each SKMO edpm kustomization that sources from ConfigMap e.g. skmo-edpm-overrides.yaml and injects edpm_nodes_validation_check_for_fqdn: false directly into the OpenStackDataPlaneNodeSet ?

abays · 2026-06-29T11:41:55Z


        edpm_nodes_validation_validate_controllers_icmp: false
        edpm_nodes_validation_validate_gateway_icmp: false
+        edpm_nodes_validation_check_for_fqdn: false


Was it intentional to change the non-SKMO multi-namespace VA here as well?

abays · 2026-06-29T12:06:18Z

All the config here is identical across all three nodes. Do we need to use a label selector? Could we just drop the labelSelector and apply to all nodes? Or are there OCP nodes we don't want to target, and hence the specific selectors?

Looking at it, the selectors really seem redundant. However I'm not sure about @vakwetu originial idea/purpose (maybe for future testing expansion), so I'd keep it if it's not a blocker.

No worries. We can always improve it later in a follow-up if @vakwetu agrees.

@vakwetu confirmed that labelSelector has no purpose, so I removed it

Deydra71 · 2026-07-01T14:02:17Z

@abays I addressed the comments above and also created newcomponents dir under examples/va/multi-namespace-skmo/ because octavia values are consumed by multiple independent kustomize build targets, so each build stage can include the ConfigMap into its own scope withou t duplication. Let me please know if it's okay with you.

Add reusable va/ components for Octavia: - networking-octavia: NetConfig skeleton and NAD replacements - octavia-controlplane: shared OSCP replacements (service config, images, network attachments, OVN nicMappings, CA passphrase secret) Add user-facing values under examples/va/multi-namespace-skmo/components/: - octavia-network-values: dedicated ConfigMap with network parameters - octavia-service-values: dedicated ConfigMap with service parameters Wire Octavia into examples/va/multi-namespace-skmo: control-plane networking (NAD, NetConfig, NNCP), EDPM bridge mapping. Scope edpm_nodes_validation_check_for_fqdn override to SKMO only via kustomize replacement (SKMO EDPM nodes cannot resolve FQDN during validate-network due to MetalLB L2 timing with Octavia networking). Signed-off-by: Ade Lee <alee@redhat.com> Assisted-by: Claude Opus 4.6

abays

/lgtm

@fultonj Do you want to review, or should I approve?

openshift-ci Bot requested review from abays and leifmadsen May 14, 2026 23:23

vakwetu requested review from fultonj and removed request for abays and leifmadsen May 14, 2026 23:23

vakwetu added the do-not-merge/work-in-progress label May 14, 2026

vakwetu requested review from beagles and gthiemonge May 14, 2026 23:26

vakwetu force-pushed the add-octavia-to-skmo branch 2 times, most recently from 424a81b to e9471c4 Compare May 26, 2026 18:31

openshift-ci Bot removed the do-not-merge/work-in-progress label May 26, 2026

vakwetu added the do-not-merge/work-in-progress label May 26, 2026

openshift-ci Bot removed the do-not-merge/work-in-progress label May 26, 2026

openshift-ci Bot assigned gthiemonge Jun 8, 2026

openshift-ci Bot added the lgtm label Jun 8, 2026

vakwetu force-pushed the add-octavia-to-skmo branch from e9471c4 to a20043c Compare June 8, 2026 22:12

openshift-ci Bot removed the lgtm label Jun 8, 2026

vakwetu added the do-not-merge/work-in-progress label Jun 8, 2026

openshift-ci Bot removed the do-not-merge/work-in-progress label Jun 9, 2026

vakwetu force-pushed the add-octavia-to-skmo branch from a20043c to 90d5ca6 Compare June 9, 2026 01:41

abays reviewed Jun 22, 2026

View reviewed changes

Deydra71 force-pushed the add-octavia-to-skmo branch 3 times, most recently from a9c9641 to 6da0165 Compare June 23, 2026 14:22

Deydra71 force-pushed the add-octavia-to-skmo branch from 6da0165 to 40c306d Compare June 24, 2026 08:57

Deydra71 force-pushed the add-octavia-to-skmo branch 2 times, most recently from fa4b45c to 1a98f8b Compare June 25, 2026 09:14

abays reviewed Jun 29, 2026

View reviewed changes

Deydra71 force-pushed the add-octavia-to-skmo branch 2 times, most recently from 63fe6b0 to ba0724c Compare July 1, 2026 13:47

Deydra71 force-pushed the add-octavia-to-skmo branch from ba0724c to 5f07fb1 Compare July 2, 2026 13:49

abays reviewed Jul 2, 2026

View reviewed changes

openshift-ci Bot assigned abays Jul 2, 2026

openshift-ci Bot added the lgtm label Jul 2, 2026

Uh oh!

Conversation

vakwetu commented May 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

openshift-ci Bot commented May 14, 2026

Uh oh!

vakwetu commented May 14, 2026

Uh oh!

centosinfra-prod-github-app Bot commented May 26, 2026

Uh oh!

fultonj commented Jun 1, 2026

Uh oh!

fultonj commented Jun 8, 2026

Uh oh!

gthiemonge commented Jun 8, 2026

Uh oh!

abays commented Jun 8, 2026

Uh oh!

abays commented Jun 8, 2026

Uh oh!

centosinfra-prod-github-app Bot commented Jun 8, 2026

Uh oh!

Deydra71 commented Jun 22, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Deydra71 commented Jun 23, 2026

Uh oh!

Deydra71 commented Jun 24, 2026

Uh oh!

abays Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

abays Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Deydra71 Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

vakwetu commented May 14, 2026 •

edited

Loading

abays Jun 29, 2026 •

edited

Loading

abays Jun 29, 2026 •

edited

Loading

Deydra71 Jun 29, 2026 •

edited

Loading