fix(substrate): degrade gracefully when ate.dev CRDs are absent by QuentinBisson · Pull Request #2023 · kagent-dev/kagent

QuentinBisson · 2026-06-15T21:44:49Z

What

GET /api/substrate/status returns 500 on clusters that don't have the
agent-substrate CRDs installed (ate.dev/v1alpha1 WorkerPool / ActorTemplate).

listSubstrateCRs calls KubeClient.List on both types unconditionally. When the
CRDs are absent the REST mapper returns *meta.NoKindMatchError, which propagates as
HTTP 500. Since substrate is optional and Enabled is already gated on AteClient != nil, the correct behaviour is to return an empty list.

Fix

Guard both List calls with meta.IsNoMatchError and return nil, nil, nil (empty
result, no error) so the handler responds 200 with Enabled: false and empty slices.

A regression test (TestHandleGetSubstrateStatus_NoCRDs) covers the no-CRD path using
a stub client that returns *meta.NoKindMatchError, asserting HTTP 200 with
Enabled: false and empty resource slices.

Reported in giantswarm/giantswarm#36845.

GET /api/substrate/status returned 500 on clusters without the ate.dev CRDs installed because listSubstrateCRs listed WorkerPool/ActorTemplate unconditionally, propagating the REST mapper NoKindMatchError as a server error. Gate the CRD listing loop on AteClient != nil. When substrate is not configured (the common case), there is nothing to list and no CRD calls are made. When substrate is configured, a missing CRD is a legitimate misconfiguration and the error is surfaced as before. Reported in giantswarm/giantswarm#36845. Signed-off-by: QuentinBisson <quentin@giantswarm.io>

Copilot

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

This PR updates the substrate status endpoint to avoid listing Kubernetes substrate CRs when the substrate isn’t configured (i.e., AteClient is nil), and adds a regression test for that behavior.

Changes:

Guard Kubernetes CR listing behind h.AteClient != nil in HandleGetSubstrateStatus.
Add a new test asserting the “substrate not configured” response shape.
Introduce a minimal client.Client stub intended to simulate missing CRDs.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
go/core/internal/httpserver/handlers/substrate.go	Skips listing substrate CRs unless `AteClient` is configured.
go/core/internal/httpserver/handlers/substrate_test.go	Adds a “not configured” test and a kube client stub for simulating missing CRDs.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

QuentinBisson · 2026-06-15T22:14:47Z

+// noMatchKubeClient is a minimal client.Client stub whose List always returns
+// a *meta.NoKindMatchError, simulating a cluster where the ate.dev CRDs are absent.
+type noMatchKubeClient struct {
+	client.Client
+}
+
+func (noMatchKubeClient) List(_ context.Context, _ client.ObjectList, _ ...client.ListOption) error {
+	return &apimeta.NoKindMatchError{}
+}
+
+// TestHandleGetSubstrateStatus_SubstrateNotConfigured verifies that when AteClient is nil
+// (substrate not configured), the endpoint returns 200 with Enabled:false and empty slices
+// without making any CRD List calls.
+func TestHandleGetSubstrateStatus_SubstrateNotConfigured(t *testing.T) {
+	t.Parallel()
+
+	scheme := runtime.NewScheme()
+	utilruntime.Must(clientgoscheme.AddToScheme(scheme))
+	fakeClient := fake.NewClientBuilder().WithScheme(scheme).Build()
+
+	base := &handlers.Base{KubeClient: fakeClient, Authorizer: &auth.NoopAuthorizer{}}
+	h := handlers.NewSubstrateHandler(base, nil)


QuentinBisson · 2026-06-15T22:14:54Z

+	if h.AteClient != nil {
+		for _, ns := range namespaces {
+			wpEntries, tmplEntries, err := h.listSubstrateCRs(r.Context(), ns)
+			if err != nil {
+				log.Error(err, "list substrate CRs", "namespace", ns)
+				w.RespondWithError(errors.NewInternalServerError("Failed to list substrate resources from Kubernetes", err))
+				return
+			}
+			resp.WorkerPools = append(resp.WorkerPools, wpEntries...)
+			resp.ActorTemplates = append(resp.ActorTemplates, tmplEntries...)
 		}
-		resp.WorkerPools = append(resp.WorkerPools, wpEntries...)
-		resp.ActorTemplates = append(resp.ActorTemplates, tmplEntries...)
 	}

 	if h.AteClient != nil {


QuentinBisson · 2026-06-15T22:14:56Z

+}
+
+
 type stubAteControl struct {


- Merge the two consecutive AteClient != nil blocks into one - Use noMatchKubeClient in the not-configured test so the test would fail if the gate were removed (the error would propagate as 500 rather than the guarded nil path) - Remove extra blank line before stubAteControl Signed-off-by: QuentinBisson <quentin@giantswarm.io>

EItanya · 2026-06-17T21:01:47Z

We can merge this as is, but I think we should probably come up with a better long-term solution here. If a user doesn't have substrate installed, why is this even being called at all?

QuentinBisson · 2026-06-17T21:07:58Z

Sure but then kagent either needs to detect it or have a flag to disable substrate :)

EItanya · 2026-06-17T21:16:53Z

Sure but then kagent either needs to detect it or have a flag to disable substrate :)

Maybe we could do a discovery client check at the beginning for the CRDs and use that as the marker?

QuentinBisson · 2026-06-17T21:24:23Z

You're right. We can use the dynamic RESTMapper instead as that reloads on NoMatch. so if I just guard the List with meta.IsNoMatchError the endpoint picks up substrate installed after
kagent starts with no restart, and I avoid a second call that can disagree with the List.

QuentinBisson · 2026-06-17T21:33:55Z

I updated the PR to:

Guard both KubeClient.List calls with apimeta.IsNoMatchError (return empty + found=false, no error).
Restore listSubstrateCRs to run unconditionally (it was incorrectly gated on AteClient != nil).
Derive Enabled from what actually answered the request: crdsPresent || AteClient != nil. This keeps the two axes (CRDs installed / ate-api gRPC reachable) independent.

Three additional test cases cover the matrix: no CRDs+no client, no CRDs+client set, CRDs present+no client.

… NoMatchError CRD presence (WorkerPool/ActorTemplate) and gRPC client availability (AteClient) are independent axes. Gating listSubstrateCRs on AteClient != nil caused two bugs: - CRDs present but ate-api not deployed: WorkerPools/ActorTemplates silently omitted even though the CRs exist, Enabled reported false. - CRDs absent but AteClient set: List calls reached the API server and 500'd instead of degrading gracefully. Guard each KubeClient.List with meta.IsNoMatchError: on NoMatch return empty + found=false with no error. controller-runtime's dynamic RESTMapper reloads on NoMatch automatically, so substrate installed after kagent boots is picked up on the next request without a restart. Compute Enabled from observed reality: crdsPresent || AteClient != nil. Three new cases covered by tests: - NoCRDs + no gRPC client -> 200, Enabled:false - NoCRDs + gRPC client set -> 200, Enabled:true, actors populated - CRDs present + no gRPC -> 200, Enabled:true, WorkerPools populated Signed-off-by: QuentinBisson <quentin@giantswarm.io>

EItanya · 2026-06-17T22:35:21Z

Sorry, I should've been more clear. I actually liked your first implementation. What I meant was to have a check on startup, I think this is a bit overcomplicated for the situation.

Let's start with the original path and then we can improve later

Revert 149a2d7. The IsNoMatchError guard is overcomplicated for this use case. Keeping the original approach (gate CRD listing on AteClient != nil) and will improve with a startup CRD discovery check separately. Signed-off-by: QuentinBisson <quentin@giantswarm.io>

QuentinBisson · 2026-06-19T16:48:44Z

I reverted the new commit :)

github-actions Bot added the bug Something isn't working label Jun 15, 2026

QuentinBisson mentioned this pull request Jun 15, 2026

fix: vendor ate.dev CRDs so substrate status endpoint degrades gracefully giantswarm/agentic-platform#101

Closed

QuentinBisson force-pushed the fix/substrate-status-no-match-error branch from c88349b to a96374e Compare June 15, 2026 21:58

QuentinBisson force-pushed the fix/substrate-status-no-match-error branch from a96374e to d32223c Compare June 15, 2026 22:04

Merge branch 'main' into fix/substrate-status-no-match-error

d917b18

QuentinBisson marked this pull request as ready for review June 15, 2026 22:06

QuentinBisson requested a review from supreme-gg-gg as a code owner June 15, 2026 22:06

Copilot AI review requested due to automatic review settings June 15, 2026 22:06

QuentinBisson requested review from iplay88keys and jmhbh as code owners June 15, 2026 22:06

Copilot AI reviewed Jun 15, 2026

View reviewed changes

QuentinBisson and others added 3 commits June 15, 2026 23:09

Merge branch 'main' into fix/substrate-status-no-match-error

669c02a

Merge branch 'main' into fix/substrate-status-no-match-error

daaaf9c

EItanya previously approved these changes Jun 17, 2026

View reviewed changes

mesutoezdil mentioned this pull request Jun 17, 2026

fix(substrate): return 200 with empty result when ate.dev CRDs are not installed #2041

Closed

QuentinBisson dismissed EItanya’s stale review via 690b3a4 June 17, 2026 21:33

QuentinBisson force-pushed the fix/substrate-status-no-match-error branch from 690b3a4 to 149a2d7 Compare June 17, 2026 21:38

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(substrate): degrade gracefully when ate.dev CRDs are absent#2023

fix(substrate): degrade gracefully when ate.dev CRDs are absent#2023
QuentinBisson wants to merge 7 commits into
kagent-dev:mainfrom
QuentinBisson:fix/substrate-status-no-match-error

QuentinBisson commented Jun 15, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

QuentinBisson Jun 15, 2026

Uh oh!

QuentinBisson Jun 15, 2026

Uh oh!

QuentinBisson Jun 15, 2026

Uh oh!

EItanya commented Jun 17, 2026

Uh oh!

QuentinBisson commented Jun 17, 2026

Uh oh!

EItanya commented Jun 17, 2026

Uh oh!

QuentinBisson commented Jun 17, 2026 •

edited

Loading

Uh oh!

QuentinBisson commented Jun 17, 2026 •

edited

Loading

Uh oh!

EItanya commented Jun 17, 2026 •

edited

Loading

Uh oh!

QuentinBisson commented Jun 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

QuentinBisson commented Jun 15, 2026

What

Fix

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

QuentinBisson Jun 15, 2026

Choose a reason for hiding this comment

Uh oh!

QuentinBisson Jun 15, 2026

Choose a reason for hiding this comment

Uh oh!

QuentinBisson Jun 15, 2026

Choose a reason for hiding this comment

Uh oh!

EItanya commented Jun 17, 2026

Uh oh!

QuentinBisson commented Jun 17, 2026

Uh oh!

EItanya commented Jun 17, 2026

Uh oh!

QuentinBisson commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

QuentinBisson commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

EItanya commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

QuentinBisson commented Jun 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

QuentinBisson commented Jun 17, 2026 •

edited

Loading

QuentinBisson commented Jun 17, 2026 •

edited

Loading

EItanya commented Jun 17, 2026 •

edited

Loading