feat: add Gateway API support for v1 controllers by kimwnasptd · Pull Request #865 · kubeflow/notebooks

kimwnasptd · 2026-01-29T15:44:50Z

This PR is the clone of #628, with some extra fixes.

The PR only extends the PVCViewer, Notebook and Tensorboard Controllers to have 3 new env vars, for controlling if the Gateway API should be used instead of Istio's resources.

Extra Changes

Specifically, this PR introduces the following changes on top of the original PR:

adds the Gateway API to the controller's scheme
adds permissions for HTTPRoute objects to ClusterRoles of PVCViewers and Tensorboard Controllers
extends the Tensorboard Controller to also reconcile changes on HTTPRoutes or VirtualServices

NOTE: I've made minimal updates to go.mod, as to not interfere with the current efforts of fixing CVEs.

Review Notes

PVCViewers is using. v1 version of gateway-api module, while the other two the v1beta1. This was what was initially done in the PR, which I guess was in order to avoid updating the controller-runtime module (which needs code changes)
Without adding the Gateway API to the schema of the controllers I was getting the error mentioned in feat: add ambient mode support to kubeflow controllers #628 (comment)

cc @deusebio
/cc @thesuperzapper
/cc @andyatmiami

google-oss-prow · 2026-01-29T15:44:57Z

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please ask for approval from kimwnasptd. For more information see the Kubernetes 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

andyatmiami

@kimwnasptd - in effort to give you some feedback as you have patiently been waiting - I reviewed all the files EXCEPT the 3 *_controller.go files..

I will look to get the _controller.go files reviewed ASAP - but realized this weekend is actually pretty busy for me - so unlikely i'll have the focus time to really reason through code... but I at least went through everything else to drop various thoughts/comments.

I wholly realize we are looking to kinda wind down notebooks-v1 - so I wouldn't necessarily force any of the comments left thus far - so you are welcome to push back/disagree and we can leave code as-is... at minimum - please provide justification so we can be transparent on why we did/didn't act on any given comment.

(i also haven't tested code - yet - again, simply due to my time/availability today.. but will certainly be planning a 2nd review + actual testing)

andyatmiami · 2026-02-07T19:31:11Z

components/notebook-controller/go.mod

+	github.com/onsi/gomega v1.18.1
+	github.com/prometheus/client_golang v1.12.1
+	k8s.io/api v0.24.1
+	k8s.io/apimachinery v0.24.1
+	k8s.io/client-go v0.24.1
+	sigs.k8s.io/controller-runtime v0.12.1
+	sigs.k8s.io/gateway-api v0.5.1


Question moreso for @thesuperzapper... do we want to try to get a little more aggressive on these API versions if we are already going to be upgrading them?

Here is a list of latest releases for changed packages:

github.com/onsi/gomega v1.39.1

github.com/prometheus/client_golang v1.23.2

k8s.io/api v0.35.0

k8s.io/apimachinery v0.35.0

k8s.io/client-go v0.35.0

sigs.k8s.io/controller-runtime v0.23.1

sigs.k8s.io/gateway-api v1.4.1

( this comment will likely apply to all *controller lib's go.mod files )

The only "problematic" one here is the sigs.k8s.io/controller-runtime one. I tried to bump it into a newer version but it ultimately requires more code changes on the core of the controller, as the newer controller-runtime has slightly different function signatures in some cases.

In the end I ended up reverting it, and this is the main reason that some components are also using the v1beta1 (v0.5.1) package for Gateway API

Let me give it a go to bump the packages to those exact versions and test out the changes.

If they look good, then it will be a nice addition for future proofing ourselves for CVEs also.

For me the main goal would be to bump the sigs.k8s.io/gateway-api v1.4.1 so that the Gateway API code is as much aligned as possible across the controllers.

Tough luck.

I tried to do the above combination, updating everything aside sigs.k8s.io/controller-runtime but because I was bumping sigs.k8s.io/gateway-api it was also then bumping sigs.k8s.io/controller-runtime to v0.22.1.

So I then tried to update everything else aside those two packages

require ( github.com/go-logr/logr v1.4.3 github.com/onsi/ginkgo v1.16.5 github.com/onsi/gomega v1.39.1 github.com/prometheus/client_golang v1.23.1 k8s.io/api v0.35.0 k8s.io/apimachinery v0.35.0 k8s.io/client-go v0.35.0 sigs.k8s.io/controller-runtime v0.12.1 sigs.k8s.io/gateway-api v0.5.1 )

But I end up getting build errors. Most probably because there's some mismatch between the sigs.k8s.io/controller-runtime v0.12.1 version and the k8s.io/client-go v0.35.0 packages:

------ > [builder 7/7] RUN CGO_ENABLED=0 GOOS=linux go build -a -o manager main.go: 9.317 # sigs.k8s.io/controller-runtime/pkg/cache 9.317 /go/pkg/mod/sigs.k8s.io/controller-runtime@v0.12.1/pkg/cache/informer_cache.go:144:9: cannot use i.Informer (variable of interface type "k8s.io/client-go/tools/cache".SharedIndexInformer) as Informer value in return statement: "k8s.io/client-go/tools/cache".SharedIndexInformer does not implement Informer (wrong type for method AddEventHandler) 9.317 have AddEventHandler("k8s.io/client-go/tools/cache".ResourceEventHandler) ("k8s.io/client-go/tools/cache".ResourceEventHandlerRegistration, error) 9.317 want AddEventHandler("k8s.io/client-go/tools/cache".ResourceEventHandler) 9.317 /go/pkg/mod/sigs.k8s.io/controller-runtime@v0.12.1/pkg/cache/informer_cache.go:158:9: cannot use i.Informer (variable of interface type "k8s.io/client-go/tools/cache".SharedIndexInformer) as Informer value in return statement: "k8s.io/client-go/tools/cache".SharedIndexInformer does not implement Informer (wrong type for method AddEventHandler) 9.317 have AddEventHandler("k8s.io/client-go/tools/cache".ResourceEventHandler) ("k8s.io/client-go/tools/cache".ResourceEventHandlerRegistration, error) 9.317 want AddEventHandler("k8s.io/client-go/tools/cache".ResourceEventHandler) ------

andyatmiami · 2026-02-07T19:37:32Z

components/notebook-controller/config/manager/params.env

Kind worried how we are evolving this file.. as it opens up the door to combinations of environment variables that are obviously non-sensical and/or hard to reason about...

What would it mean if someone configured the following:

USE_ISTIO=true USE_GATEWAY_API=true ISTIO_GATEWAY=kubeflow/kubeflow-gateway K8S_GATEWAY_NAME=kubeflow-gateway K8S_GATEWAY_NAMESPACE=kubeflow

I would probably something like this (but I also wouldn't die on this hill)

GATEWAY_VENDOR=<istio/k8s> GATEWAY_FQN=... USE_ISTIO=true ISTIO_GATEWAY=kubeflow/kubeflow-gateway

then for backwards compatibility:

if USE_ISTIO: true - then GATEWAY_VENDOR: istio

if USE_ISTIO: true - then GATEWAY_FQN is set to ISTIO_GATEWAY (if its defined)

I think this gives us sufficient backwards compatibility while also avoiding potentially undefined config combinations...

Additionally - it seems the gateway name/namspace for GatewayAPI cares about the individual pieces (not the FQN) - but given a namspace cannot contain / in its name - splitting apart this singular env var seems straightforward enough..

WDYT?

But can't we still end up with non-sensical and/or hard to reason about scenarios with the proposed names?

For example, what would it mean if someone configured the following:

USE_ISTIO=true GATEWAY_VENDOR=k8s ISTIO_GATEWAY=kubeflow/kubeflow-gateway GATEWAY_FQN=kubeflow/istio-ingressgateway

So IIUC, as long as the older env vars keep the same name (which they should, to preserve backwards compatibility) then we can end up with conflicting scenarios with the env vars.

My suggestion is the following:

We acknowledge that there can be such conflicts

We agree on what the schema for the new env vars will be

By default, if the new env vars are not set, the controller code should behave the old way (creating VirtualServices).

We add some login in the main.go of the controller to error out immediately with a helpful log message, if such a conflicted configuration is found (USE_ISTIO true and Use_GATEWAY_API to true)

Add comments about that also in the manifests

WDYT?

Fair point @kimwnasptd!

I guess my other "concern" - albeit its a minor one - is in adding new environment variables that are also "vendor-specific*

USE_ISTIO=true USE_GATEWAY_API=true ISTIO_GATEWAY=kubeflow/kubeflow-gateway K8S_GATEWAY_NAME=kubeflow-gateway K8S_GATEWAY_NAMESPACE=kubeflow

If for some reason (and this is likely wildly unlikely) - a 3rd ( ❗ ) alternative rose to prominence - we'd then need to add 2-3 new environment variables again.. as we are scoping env vars to the "vendor"

Whereas in this mode - we break that pattern - using "generic" names + a "discriminator" (i.e. enum in GATEWAY_VENDOR) to capture the data..

USE_ISTIO=true GATEWAY_VENDOR=k8s ISTIO_GATEWAY=kubeflow/kubeflow-gateway GATEWAY_FQN=kubeflow/istio-ingressgateway

While for notebooks-v1 exclusively I will absolutely admit this does not matter - setting a pattern here may/could influence what we'd want to do for notebooks-v2

But, I've already probably typed too much on the topic.. I agree with your suggestions in prior comment.. and i'd (personally) lean towards the env vars I am suggesting - but its not a hill I would die on..

I am comfortable that you have internalized my feedback - and am happy to then defer to whatever implementation you feel best.

Yes, I see your point.

The only reason I preferred the USE_GATEWAY_API is because (theoretically) the goal of the Gateway API was to be the standard way on describing Gateways / Routing in K8s.

Let's get some feedback from @thesuperzapper

andyatmiami · 2026-02-07T19:40:01Z

components/notebook-controller/config/manager/manager.yaml

+  selector:
+    matchLabels:
+      app: notebook-controller
+      kustomize.component: notebook-controller


apologies if I am being dense - but what is the underling motivation behind adding this?

is it required to be defined for GatewayAPI uses cases?

Ah, good catch. This should be redundant as we are already setting those in

notebooks/components/notebook-controller/config/default/kustomization.yaml

Lines 12 to 14 in e8c7d97

commonLabels:

app: notebook-controller

kustomize.component: notebook-controller

Removing the changes.

2243bf5

andyatmiami · 2026-02-07T19:41:00Z

components/pvcviewer-controller/config/manager/params.env

@@ -0,0 +1,6 @@
+USE_ISTIO=true


same comment as here:

https://github.com/kubeflow/notebooks/pull/865/changes#r2777978332

andyatmiami · 2026-02-07T19:41:43Z

components/notebook-controller/config/manager/manager.yaml

+        envFrom:
+        - configMapRef:
+            name: config


this was beautiful simplification ( long time coming 🤣 )

andyatmiami · 2026-02-07T19:43:47Z

components/pvcviewer-controller/go.mod

I feel like maybe we should take this effort to align various package dependency versions that exist across different go.mod files for the various controllers with this repo.

This then also factors into my discussion on if/how aggressively we wanted to bump up dependencies

https://github.com/kubeflow/notebooks/pull/865/changes#r2777971257

But, I can also wholly appreciate this is scope creep - so feel free to push back.

andyatmiami · 2026-02-07T19:44:46Z

components/tensorboard-controller/config/manager/params.env

same comment as here:

https://github.com/kubeflow/notebooks/pull/865/changes#r2777978332

andyatmiami · 2026-02-07T19:45:27Z

components/tensorboard-controller/go.mod

I feel like maybe we should take this effort to align various package dependency versions that exist across different go.mod files for the various controllers with this repo.

This then also factors into my discussion on if/how aggressively we wanted to bump up dependencies

https://github.com/kubeflow/notebooks/pull/865/changes#r2777971257

But, I can also wholly appreciate this is scope creep - so feel free to push back.

kimwnasptd · 2026-02-10T10:21:03Z

Thank you so much for your time reviewing this one @andyatmiami!

andyatmiami

@kimwnasptd

Sorry for delay - finally had time/focus to step through the controller changes... appreciate your time in crafting this new feature.

Happy to have conversations on any of these points if you feel it necessary - and also feel empowered to 'overrule' me should you feel any suggestion unnecessary.

I'll try to make sure @thesuperzapper looks over this too to capture his feedback so you don't get further slow-dripped on requests for change.

ℹ️ I realize the VirtualService "framework" (and the controllers as a whole) do NOT presently have unit tests... so it seems (possibly?) unfair to ask tests be contributed in the context of this PR. But just in the sake of transparency - its very unfortunate w.r.t the immaturity of tests here - as this new code contribution (and its significant impact in controller capabilities) could/would REALLY benefit now and moving forward with having tests give us some peace of mind.

That being said - I know we are looking to get notebooks-v1 in a more strict "maintenance mode" ASAP (if/as notebooks-v2 is GA'd).. so might be a moot point and/or not burn us in the future.

andyatmiami · 2026-02-15T17:05:17Z

components/notebook-controller/controllers/notebook_controller.go

+	// Reconcile VirtualService if we use ISTIO.
 	if os.Getenv("USE_ISTIO") == "true" {
 		err = r.reconcileVirtualService(instance)
 		if err != nil {
 			return ctrl.Result{}, err
 		}
 	}

+	// Reconcile HTTPRoute if we use Gateway API.
+	if os.Getenv("USE_GATEWAY_API") == "true" {
+		err = r.reconcileHTTPRoute(instance)
+		if err != nil {
+			return ctrl.Result{}, err
+		}
+	}
+


Other controllers seem to have this logic wrapping in an if/else such that only ONE of these env vars would ever take effect.

But here in notebook-controller we are seemingly supporting reconciling BOTH "modes" at one? Is this intentional?

Feels like this should also be an if/else...

Not sure why they implemented it to be different in this case, but I can't think of any reason to not use the same structure as the other places.

f6d6d67

andyatmiami · 2026-02-15T17:06:58Z

components/notebook-controller/controllers/notebook_controller.go

 	// watch Istio virtual service
 	if os.Getenv("USE_ISTIO") == "true" {
 		virtualService := &unstructured.Unstructured{}
 		virtualService.SetAPIVersion("networking.istio.io/v1alpha3")
 		virtualService.SetKind("VirtualService")
 		builder.Owns(virtualService)
 	}

+	// watch HTTPRoute
+	if os.Getenv("USE_GATEWAY_API") == "true" {
+		builder.Owns(&gwapiv1beta1.HTTPRoute{})
+	}
+


Similar observation with mutually exclusive if statements as raised here:

https://github.com/kubeflow/notebooks/pull/865/changes#r2809711972

andyatmiami · 2026-02-15T17:16:48Z

components/notebook-controller/controllers/notebook_controller.go

+func generateHTTPRoute(instance *v1beta1.Notebook) (*gwapiv1beta1.HTTPRoute, error) {
+	name := instance.Name
+	namespace := instance.Namespace
+	pathPrefix := fmt.Sprintf("/notebook/%s/%s", namespace, name)


I wonder if we should factor the generation of this path out into its own function. Particularly because of the similar-but-different structure in VirtualService:

https://github.com/kubeflow/notebooks/pull/865/changes#diff-a436986c5e93996f9035eee6798d064ab8659957e8443eb44b3a4d3db289bde4R493

As I understand it... VirtualService uses simple string prefix matching - whereas HttpRoute is segment-aware. As such - we WANT a trailing slash for VirtualService but NOT for HttpRoute

A simple helper function driven off our env variables (which specify what "mode" we are in) will help keep this aligned and also help future maintainers understand this relationship.

Nah, the backslash at the end should also be used in the HTTPRoute as well. While indeed the path prefix of httproutes is different, still all routes that will be accessed are going to be /notebook/test-ns/test-nm/*

I pushed a commit in which I introduced the common function
fa51c8a

andyatmiami · 2026-02-15T17:18:21Z

components/pvcviewer-controller/controllers/pvcviewer_controller.go


+
+func (r *PVCViewerReconciler) reconcileHTTPRoute(ctx context.Context, log logr.Logger, viewer *kubefloworgv1alpha1.PVCViewer, commonLabels map[string]string) error {
+	prefix := fmt.Sprintf("%s/%s/%s", viewer.Spec.Networking.BasePrefix, viewer.Namespace, viewer.Name)


Similar comment as in:

https://github.com/kubeflow/notebooks/pull/865/changes#r2809723594

andyatmiami · 2026-02-15T17:18:37Z

components/tensorboard-controller/controllers/tensorboard_controller.go

 }

+func generateHTTPRoute(tb *tensorboardv1alpha1.Tensorboard) (*gwapiv1beta1.HTTPRoute, error) {
+	prefix := fmt.Sprintf("/tensorboard/%s/%s", tb.Namespace, tb.Name)


similar comment as in:

https://github.com/kubeflow/notebooks/pull/865/changes#r2809723594

andyatmiami · 2026-02-15T17:43:07Z

components/notebook-controller/controllers/notebook_controller.go

If a deployment switches from USE_GATEWAY_API=true to USE_ISTIO=true (or vice versa), the previously created routing resource (HTTPRoute or VirtualService) is orphaned. The owner references will eventually clean it up when the parent resource is deleted, but as long as the parent exists, the old routing resource remains and could still be actively routing traffic -- potentially conflicting with the new one.

I'm not sure how we want to handle this (and/or the existing behavior of just ignoring this paradigm during a reconcile loop is sufficient)

But its at least probably worth discussing and maybe adding an inline comment (or something) in the reconcile function calling this out..

⚠️ Please note this also applies to pvcviewer and tensorboard - but for the sake of brevity I will just leave this comment here.

This is a very good point to sync on.

The current behavior follows the KServe pattern, in which it's not officially supported to switch between the two. Which in practical terms mean that VirtualServices will be leftover.

Fully agree to document this. I can definitely add this to the reconcile function, and also potentially to the READMEs of the components. Do you have other thoughts on this @andyatmiami

andyatmiami · 2026-02-15T17:46:01Z

components/pvcviewer-controller/controllers/pvcviewer_controller.go

 			"kind":       "VirtualService",
 		},
 	}
+	httpRouteTemplate = &gwapiv1.HTTPRoute{}


strictly speaking - we don't need to define this as a var and could just reference it inline.. but I can appreciate since the virtualservice counterpart is defined here (and needs to be a var) - co-locating has its benefits.

yeh, I'm also OK with both. Happy to pick one if you have a preference

andyatmiami · 2026-02-15T17:47:32Z

components/tensorboard-controller/config/manager/params.env

could we add a trailing newline to make the computers happy ?

oh interesing! what do you mean by that? 😅

andyatmiami · 2026-02-15T17:47:41Z

components/pvcviewer-controller/config/manager/params.env

could we add a trailing newline to make the computers happy ?

andyatmiami · 2026-02-15T17:49:53Z

components/notebook-controller/controllers/notebook_controller.go

+
+	httpRoute := &gwapiv1beta1.HTTPRoute{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      virtualServiceName(name, namespace),


totally minor nit you can ignore if you choose..

but having an httproute create its name through a function named virtualServiceName just feels a little off ..

routeResourceName - or something similarly vendor-agnostic - might be nice 🤷

now it's too late to unsee, so I'm making the change 😄

8f41973

kimwnasptd · 2026-02-22T13:00:03Z

/ok-to-test

Signed-off-by: madmecodes <ayushguptadev1@gmail.com> Signed-off-by: kimwnasptd <kimonas.sotirchos@canonical.com>

Signed-off-by: Kimonas Sotirchos <kimonas.sotirchos@canonical.com>

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd · 2026-02-22T18:51:40Z

Thank you so much for the detailed review @andyatmiami!

I fully agree with your sentiment on the unit tests. I'll see if I can find some time to add some basic unit tests regarding the HTTPRoute and VirtualSerices. The goal will be though to not need to introduce any new packages, just to ensure we generate the expected resources.

Although I can't promise I'll make it until next week, as my schedule is a bit hectic.

The rest should be ready for another pass!

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd · 2026-02-22T19:19:43Z

Also a note, I added the EXPERIMENTAL_ prefix to all env vars that we have for controlling the Gateway API functionality, to align with the discussions we had also with @thesuperzapper on the Notebooks WG calls.

5d8066a

github-project-automation bot added this to Kubeflow Notebooks Jan 29, 2026

github-project-automation bot moved this to Needs Triage in Kubeflow Notebooks Jan 29, 2026

google-oss-prow bot requested review from andyatmiami and thesuperzapper January 29, 2026 15:44

google-oss-prow bot added the area/controller area - related to controller components label Jan 29, 2026

google-oss-prow bot added area/v1 area - version - kubeflow notebooks v1 size/XL labels Jan 29, 2026

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch 2 times, most recently from 2e68509 to 76db9b8 Compare January 29, 2026 15:52

andyatmiami reviewed Feb 7, 2026

View reviewed changes

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch 2 times, most recently from 5351b9c to 2243bf5 Compare February 10, 2026 10:25

andyatmiami suggested changes Feb 15, 2026

View reviewed changes

google-oss-prow bot assigned andyatmiami Feb 15, 2026

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch 2 times, most recently from 81a0f25 to 41c45e0 Compare February 22, 2026 12:58

google-oss-prow bot added the ok-to-test label Feb 22, 2026

madmecodes and others added 9 commits February 22, 2026 14:18

feat: add ambient mode support to kubeflow controllers

a32cdad

Signed-off-by: madmecodes <ayushguptadev1@gmail.com> Signed-off-by: kimwnasptd <kimonas.sotirchos@canonical.com>

update: rename and using the gateway api package

5871eb4

Signed-off-by: madmecodes <ayushguptadev1@gmail.com> Signed-off-by: kimwnasptd <kimonas.sotirchos@canonical.com>

update: notebook controller envFrom that was missing

4754c65

Signed-off-by: madmecodes <ayushguptadev1@gmail.com> Signed-off-by: kimwnasptd <kimonas.sotirchos@canonical.com>

update: env vars fix

824dd0c

Signed-off-by: madmecodes <ayushguptadev1@gmail.com> Signed-off-by: kimwnasptd <kimonas.sotirchos@canonical.com>

fix(kimonas): add Gateway API to controllers' scheme

e127c0b

Signed-off-by: Kimonas Sotirchos <kimonas.sotirchos@canonical.com>

fix(kimonas): add clusterrole permissions for httproutes

9870c33

Signed-off-by: Kimonas Sotirchos <kimonas.sotirchos@canonical.com>

fix(kimonas): reconcile httproutes for tensorboard-controller

35f6c99

Signed-off-by: Kimonas Sotirchos <kimonas.sotirchos@canonical.com>

review: remove unnecessary labels

850f071

Signed-off-by: Kimonas Sotirchos <kimonas.sotirchos@canonical.com>

review: if/else for gateway/istio handlnig

f6d6d67

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch from 41c45e0 to f6d6d67 Compare February 22, 2026 13:19

fix: update envtest for gsc pull errors in unit tests

cb9e0e2

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

google-oss-prow bot added size/XXL and removed size/XL labels Feb 22, 2026

review: conditionally include gateway api in builder

177bb8b

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch from d67c2ef to f608aa1 Compare February 22, 2026 17:41

review: helper for path prefix

fa51c8a

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch from f608aa1 to fa51c8a Compare February 22, 2026 17:50

kimwnasptd added 2 commits February 22, 2026 20:25

review: don't return err on generateHttpRoute

08045e4

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

review: run 'make manifests' for pvcviewers

a24da99

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch from eba3925 to a24da99 Compare February 22, 2026 18:25

review: don't use * for rbac in notebooks-controller

c102e46

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd force-pushed the feat-kimwnasptd-pr-628-clone branch from 515db6d to c102e46 Compare February 22, 2026 18:31

review: use generic function for notebook vs name

8f41973

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

feat: add EXPERIMENTAL_ prefix to env vars

5d8066a

Signed-off-by: Kimonas Sotirchos <kimwnasptd@gmail.com>

kimwnasptd mentioned this pull request Feb 28, 2026

feat: Add ambient mode support to profile-controller (clone) kubeflow/dashboard#185

Open

juliusvonkohout mentioned this pull request Mar 12, 2026

feat: add ambient mode support to kubeflow controllers #628

Closed

	commonLabels:
	app: notebook-controller
	kustomize.component: notebook-controller



		func (r PVCViewerReconciler) reconcileHTTPRoute(ctx context.Context, log logr.Logger, viewer kubefloworgv1alpha1.PVCViewer, commonLabels map[string]string) error {
		prefix := fmt.Sprintf("%s/%s/%s", viewer.Spec.Networking.BasePrefix, viewer.Namespace, viewer.Name)

Conversation

kimwnasptd commented Jan 29, 2026

Extra Changes

Review Notes

Uh oh!

google-oss-prow bot commented Jan 29, 2026

Uh oh!

andyatmiami left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kimwnasptd Feb 9, 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!

kimwnasptd Feb 10, 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!

kimwnasptd commented Feb 10, 2026

Uh oh!

andyatmiami left a comment

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!

kimwnasptd Feb 22, 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!

kimwnasptd Feb 22, 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!

kimwnasptd Feb 22, 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

kimwnasptd Feb 9, 2026 •

edited

Loading

kimwnasptd Feb 10, 2026 •

edited

Loading

kimwnasptd Feb 22, 2026 •

edited

Loading

kimwnasptd Feb 22, 2026 •

edited

Loading

kimwnasptd Feb 22, 2026 •

edited

Loading