feat(machine-a-tron): k8s controller for mock BMC services#3955
Open
akorobkov-nvda wants to merge 17 commits into
Open
feat(machine-a-tron): k8s controller for mock BMC services#3955akorobkov-nvda wants to merge 17 commits into
akorobkov-nvda wants to merge 17 commits into
Conversation
Contributor
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
WalkthroughThis PR adds a Kubernetes controller that polls machine-a-tron status, discovers mock BMC Services, and reconciles host and DPU Services. It includes an HTTP client, container image, Helm deployment resources, parent-chart integration, tests, and CI image building. ChangesMachine-a-tron Kubernetes controller
Estimated code review effort: 4 (Complex) | ~60 minutes Sequence Diagram(s)sequenceDiagram
participant Controller
participant MATServices
participant MachineATron
participant KubernetesAPI
Controller->>MATServices: Discover labeled Services
MATServices-->>Controller: Service URLs and pod names
Controller->>MachineATron: GET /machines/status
MachineATron-->>Controller: Machine and BMC status
Controller->>KubernetesAPI: List managed Services
Controller->>KubernetesAPI: Create, update, recreate, or delete Services
Suggested labels: 🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
akorobkov-nvda
changed the title
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 6
🧹 Nitpick comments (4)
dev/k8s/machine-a-tron-controller/pkg/controller/discovery.go (1)
57-57: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winConsider making the cluster DNS domain configurable.
svc.cluster.localis hardcoded; clusters with a custom--cluster-domainwould break discovery URLs. As per path instructions fordev/**, development tooling should use "safe defaults" while remaining consistent with deployment manifests — exposing this as a constructor parameter (defaulting tocluster.local) would make the tool portable.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@dev/k8s/machine-a-tron-controller/pkg/controller/discovery.go` at line 57, Make the cluster DNS domain configurable in the discovery component by adding a constructor parameter with a safe default of "cluster.local", then use that value instead of the hardcoded suffix in the URL built by the discovery logic. Update the relevant constructor call sites while preserving existing behavior when no custom domain is provided.Source: Path instructions
dev/k8s/machine-a-tron-controller/pkg/controller/controller.go (2)
156-160: 🩺 Stability & Availability | 🔵 TrivialReserve the BMC IP range from dynamic Service allocation.
Setting
ClusterIPdirectly from the BMC's static IP works only if that address is guaranteed free. Unless the BMC IP subrange is excluded from Kubernetes' dynamic Service IP allocator (e.g. via--service-cluster-ip-rangecarve-out or a reserved sub-range), an unrelated auto-assigned Service could already hold that IP, causingCreate/Updatehere to fail with an allocation conflict.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller.go` around lines 156 - 160, Reserve the BMC IP range used by the ClusterIP assignment in the Kubernetes Service CIDR configuration, using a dedicated non-overlapping sub-range excluded from dynamic Service allocation. Update the deployment/configuration referenced by the controller’s Service creation path so addresses assigned from machine.BMC.IP cannot be auto-allocated to unrelated Services.
361-367: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick winReuse
matclient.Clientinstances across reconcile passes.A new
matclient.Clientis constructed per instance on everyReconcile()call. WhenWithInsecureSkipVerifyis part ofclientOpts(likely, given self-signed BMC certs), each call clones a brand-new*http.Transport, discarding the previous pass's connection pool and forcing a fresh TCP+TLS handshake every cycle instead of reusing keep-alives.♻️ Proposed fix: cache clients per instance URL
type Reconciler struct { discovery *MatPodDiscovery serviceBuilder *ServiceBuilder k8sClient K8sServiceClient clientOpts []matclient.Option logger zerolog.Logger + clients map[string]*matclient.Client } @@ for _, instance := range instances { - client, err := matclient.NewClient(instance.URL, r.clientOpts...) - if err != nil { - result.Errors = append(result.Errors, fmt.Errorf("creating client for %s: %w", instance.URL, err)) - fetchFailed = true - continue + client, ok := r.clients[instance.URL] + if !ok { + var err error + client, err = matclient.NewClient(instance.URL, r.clientOpts...) + if err != nil { + result.Errors = append(result.Errors, fmt.Errorf("creating client for %s: %w", instance.URL, err)) + fetchFailed = true + continue + } + if r.clients == nil { + r.clients = make(map[string]*matclient.Client) + } + r.clients[instance.URL] = client }🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller.go` around lines 361 - 367, Cache matclient.Client instances by instance.URL across Reconcile passes instead of constructing them on every iteration in Reconcile. Add or reuse controller-level client storage, return the cached client when available, and only call matclient.NewClient for unseen URLs; preserve the existing error aggregation and fetchFailed behavior for client-creation failures.dev/k8s/machine-a-tron-controller/pkg/controller/controller_test.go (1)
280-389: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winAdd a recreate-on-ClusterIP-change test case.
The
TestComputeServiceDifftable covers create/update/delete but omits theRecreatepath (immutable ClusterIP change), which is an explicitly called-out behavior for this controller.✅ Proposed additional test case
+ { + name: "recreate on ClusterIP change", + desired: []*corev1.Service{ + func() *corev1.Service { + svc := makeTestService("svc-1", "mat-id-1") + svc.Spec.ClusterIP = "10.0.0.5" + return svc + }(), + }, + existing: []*corev1.Service{ + func() *corev1.Service { + svc := makeTestService("svc-1", "mat-id-1") + svc.Spec.ClusterIP = "10.0.0.6" + return svc + }(), + }, + wantRecreateCount: 1, + },(Requires adding a
wantRecreateCountfield to the test struct and assertingdiff.Recreate.)🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller_test.go` around lines 280 - 389, Add a wantRecreateCount field to the TestComputeServiceDiff table, add a case with matching service identity but different ClusterIP values expecting one recreate, and assert diff.Recreate alongside Create, Update, and Delete counts. Keep the existing cases’ expected recreate count at zero.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.github/workflows/ci.yaml:
- Line 1863: Update the core-ci-pass job’s needs list to include
build-mat-k8s-controller, matching its inclusion in build-summary and
notify-build-status. Preserve the existing dependency entries and ensure a
failed controller build prevents core-ci-pass from passing.
- Around line 819-843: Add build-mat-k8s-controller to the core-ci-pass
aggregator’s required job dependencies so its failure makes the merge-gating
check fail. Preserve the existing job configuration and ensure the dependency
references this exact job identifier.
In `@dev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.go`:
- Line 123: Validate the sync interval before the time.NewTicker call in the
controller startup flow, rejecting zero or negative values from either
SYNC_INTERVAL or --sync-interval. Fail fast with a clear error message, and only
create the ticker after the interval passes validation.
In `@dev/k8s/machine-a-tron-controller/Makefile`:
- Around line 4-10: Add an `all` target to the Makefile and include it in the
`.PHONY` declaration, wiring it to the appropriate default build workflow so
`make all` succeeds and satisfies checkmake.
In `@dev/k8s/machine-a-tron-controller/pkg/matclient/client.go`:
- Around line 109-123: Update applyInsecureTLS to set the cloned transport’s
TLSClientConfig.MinVersion to the required minimum TLS protocol, while
preserving InsecureSkipVerify for dev/test certificates and the existing
transport cloning behavior.
In `@dev/k8s/machine-a-tron-controller/README.md`:
- Around line 68-72: Update the README command example for make run to use the
$HOME-based kubeconfig path instead of ~/.kube/config, ensuring Make and the
shell pass the expanded absolute path to client-go.
---
Nitpick comments:
In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller_test.go`:
- Around line 280-389: Add a wantRecreateCount field to the
TestComputeServiceDiff table, add a case with matching service identity but
different ClusterIP values expecting one recreate, and assert diff.Recreate
alongside Create, Update, and Delete counts. Keep the existing cases’ expected
recreate count at zero.
In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller.go`:
- Around line 156-160: Reserve the BMC IP range used by the ClusterIP assignment
in the Kubernetes Service CIDR configuration, using a dedicated non-overlapping
sub-range excluded from dynamic Service allocation. Update the
deployment/configuration referenced by the controller’s Service creation path so
addresses assigned from machine.BMC.IP cannot be auto-allocated to unrelated
Services.
- Around line 361-367: Cache matclient.Client instances by instance.URL across
Reconcile passes instead of constructing them on every iteration in Reconcile.
Add or reuse controller-level client storage, return the cached client when
available, and only call matclient.NewClient for unseen URLs; preserve the
existing error aggregation and fetchFailed behavior for client-creation
failures.
In `@dev/k8s/machine-a-tron-controller/pkg/controller/discovery.go`:
- Line 57: Make the cluster DNS domain configurable in the discovery component
by adding a constructor parameter with a safe default of "cluster.local", then
use that value instead of the hardcoded suffix in the URL built by the discovery
logic. Update the relevant constructor call sites while preserving existing
behavior when no custom domain is provided.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 3dd9240a-418e-49b6-916d-35ce5e8b9a28
⛔ Files ignored due to path filters (1)
dev/k8s/machine-a-tron-controller/go.sumis excluded by!**/*.sum
📒 Files selected for processing (26)
.github/workflows/ci.yamlcrates/machine-a-tron/README.mddev/k8s/machine-a-tron-controller/Dockerfiledev/k8s/machine-a-tron-controller/Makefiledev/k8s/machine-a-tron-controller/README.mddev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.godev/k8s/machine-a-tron-controller/go.moddev/k8s/machine-a-tron-controller/pkg/controller/controller.godev/k8s/machine-a-tron-controller/pkg/controller/controller_test.godev/k8s/machine-a-tron-controller/pkg/controller/discovery.godev/k8s/machine-a-tron-controller/pkg/controller/k8s_client.godev/k8s/machine-a-tron-controller/pkg/matclient/client.godev/k8s/machine-a-tron-controller/pkg/matclient/client_test.godev/k8s/machine-a-tron-controller/pkg/matclient/types.gohelm/charts/nico-machine-a-tron/Chart.yamlhelm/charts/nico-machine-a-tron/README.mdhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/Chart.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/_helpers.tplhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/deployment.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/rbac.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/serviceaccount.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/values.yamlhelm/charts/nico-machine-a-tron/templates/bmc-services.yamlhelm/charts/nico-machine-a-tron/templates/deployment.yamlhelm/charts/nico-machine-a-tron/templates/service.yamlhelm/charts/nico-machine-a-tron/values.yaml
Contributor
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
.github/workflows/ci.yaml (1)
819-844: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winTrigger the controller build when controller files change.
This job is gated by
source_files_changed, but thepreparepath filter at Lines [168]-[173] does not includedev/k8s/machine-a-tron-controller/**. Consequently, changes to the controller’s Go sources or Dockerfile can skip this job and leave the deployed image stale.Add the controller path to the source filter or introduce a dedicated change flag.
As per path instructions, GitHub Actions trigger correctness and CI coverage gaps must be reviewed.
Proposed fix
source_files: - 'crates/**' + - 'dev/k8s/machine-a-tron-controller/**' - 'Cargo.toml'🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.github/workflows/ci.yaml around lines 819 - 844, Update the prepare path filter used to produce source_files_changed so it includes dev/k8s/machine-a-tron-controller/**. Preserve the existing build-mat-k8s-controller condition and ensure changes to the controller sources or Dockerfile set source_files_changed to true, triggering the controller image build.Source: Path instructions
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In @.github/workflows/ci.yaml:
- Around line 819-844: Update the prepare path filter used to produce
source_files_changed so it includes dev/k8s/machine-a-tron-controller/**.
Preserve the existing build-mat-k8s-controller condition and ensure changes to
the controller sources or Dockerfile set source_files_changed to true,
triggering the controller image build.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: bc6891e1-6c2f-4160-b296-af703f5b0bba
📒 Files selected for processing (5)
.github/workflows/ci.yamldev/k8s/machine-a-tron-controller/Makefiledev/k8s/machine-a-tron-controller/README.mddev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.godev/k8s/machine-a-tron-controller/pkg/matclient/client.go
🚧 Files skipped from review as they are similar to previous changes (3)
- dev/k8s/machine-a-tron-controller/README.md
- dev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.go
- dev/k8s/machine-a-tron-controller/pkg/matclient/client.go
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (1)
.github/workflows/ci.yaml (1)
174-174: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick winUse a controller-specific change gate.
source_files_changedalso matches broad paths such ascrates/**,Cargo.toml, andCargo.lock. As a result, unrelated Rust changes trigger this two-architecture controller image build even though its context is onlydev/k8s/machine-a-tron-controller. Add a dedicated controller-path output and gate this job on that output, while retainingsource_files_changedforbuild-machine-a-tron.As per path instructions, this workflow should minimize unnecessary CI work and preserve accurate coverage.
Also applies to: 820-827
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.github/workflows/ci.yaml at line 174, Update the path-filter outputs and the controller image job’s condition in the CI workflow to use a dedicated output matching only dev/k8s/machine-a-tron-controller/**. Keep source_files_changed as the gate for build-machine-a-tron, and ensure the controller-specific output is used consistently for the two-architecture controller build.Source: Path instructions
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In @.github/workflows/ci.yaml:
- Line 174: Update the path-filter outputs and the controller image job’s
condition in the CI workflow to use a dedicated output matching only
dev/k8s/machine-a-tron-controller/**. Keep source_files_changed as the gate for
build-machine-a-tron, and ensure the controller-specific output is used
consistently for the two-architecture controller build.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 47106ce5-ef3e-4a3a-a298-73a5951d6bfc
📒 Files selected for processing (1)
.github/workflows/ci.yaml
akorobkov-nvda
force-pushed
the
feat/machine-a-tron-k8s-ctrl
branch
from
July 23, 2026 17:33
9bf6d81 to
5068f1b
Compare
|
🌿 Preview your docs: https://nvidia-preview-pull-request-3955.docs.buildwithfern.com/infra-controller |
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
dev/k8s/machine-a-tron-controller/Dockerfile (1)
5-13: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick winCross-compile instead of emulating the arm64 builder.
The CI job builds
linux/amd64andlinux/arm64on an amd64 runner. Without an explicitBUILDPLATFORMbuilder andTARGETOS/TARGETARCHarguments, the arm64 build may run the Go toolchain under emulation, slowing CI and adding an unnecessary QEMU dependency.As per path instructions, Dockerfiles should explicitly support architecture-aware builds.
Proposed fix
# Build stage -FROM golang:1.26.4-alpine@sha256:3ad57304ad93bbec8548a0437ad9e06a455660655d9af011d58b993f6f615648 AS builder +FROM --platform=$BUILDPLATFORM golang:1.26.4-alpine@sha256:3ad57304ad93bbec8548a0437ad9e06a455660655d9af011d58b993f6f615648 AS builder +ARG TARGETOS +ARG TARGETARCH WORKDIR /app ... -RUN CGO_ENABLED=0 GOOS=linux go build -ldflags="-s -w" -o mat-k8s-controller ./cmd/mat-k8s-controller +RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags="-s -w" -o mat-k8s-controller ./cmd/mat-k8s-controller🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@dev/k8s/machine-a-tron-controller/Dockerfile` around lines 5 - 13, Update the Dockerfile’s builder stage to use the native build platform and declare the target platform arguments, then pass TARGETOS and TARGETARCH to the Go build command so linux/arm64 is cross-compiled rather than emulated. Preserve the existing dependency download, static build settings, output name, and pinned base image.Source: Path instructions
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@dev/k8s/machine-a-tron-controller/pkg/controller/controller.go`:
- Around line 72-78: Update BuildServiceName to use the complete matID when
constructing the Service name; remove the shortID truncation logic and preserve
the existing “mat-bmc-<machineType>-<matID>” format.
In `@helm/charts/nico-machine-a-tron/README.md`:
- Line 101: Update the dpuPerHostCount example comment to use the literal ->
arrow instead of the escaped -> entity, preserving the surrounding YAML
example unchanged.
---
Nitpick comments:
In `@dev/k8s/machine-a-tron-controller/Dockerfile`:
- Around line 5-13: Update the Dockerfile’s builder stage to use the native
build platform and declare the target platform arguments, then pass TARGETOS and
TARGETARCH to the Go build command so linux/arm64 is cross-compiled rather than
emulated. Preserve the existing dependency download, static build settings,
output name, and pinned base image.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 380282ec-0288-4e49-b256-d65c5d549e18
⛔ Files ignored due to path filters (1)
dev/k8s/machine-a-tron-controller/go.sumis excluded by!**/*.sum
📒 Files selected for processing (26)
.github/workflows/ci.yamlcrates/machine-a-tron/README.mddev/k8s/machine-a-tron-controller/Dockerfiledev/k8s/machine-a-tron-controller/Makefiledev/k8s/machine-a-tron-controller/README.mddev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.godev/k8s/machine-a-tron-controller/go.moddev/k8s/machine-a-tron-controller/pkg/controller/controller.godev/k8s/machine-a-tron-controller/pkg/controller/controller_test.godev/k8s/machine-a-tron-controller/pkg/controller/discovery.godev/k8s/machine-a-tron-controller/pkg/controller/k8s_client.godev/k8s/machine-a-tron-controller/pkg/matclient/client.godev/k8s/machine-a-tron-controller/pkg/matclient/client_test.godev/k8s/machine-a-tron-controller/pkg/matclient/types.gohelm/charts/nico-machine-a-tron/Chart.yamlhelm/charts/nico-machine-a-tron/README.mdhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/Chart.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/_helpers.tplhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/deployment.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/rbac.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/serviceaccount.yamlhelm/charts/nico-machine-a-tron/charts/mat-k8s-controller/values.yamlhelm/charts/nico-machine-a-tron/templates/bmc-services.yamlhelm/charts/nico-machine-a-tron/templates/deployment.yamlhelm/charts/nico-machine-a-tron/templates/service.yamlhelm/charts/nico-machine-a-tron/values.yaml
🚧 Files skipped from review as they are similar to previous changes (17)
- helm/charts/nico-machine-a-tron/templates/service.yaml
- helm/charts/nico-machine-a-tron/charts/mat-k8s-controller/Chart.yaml
- helm/charts/nico-machine-a-tron/Chart.yaml
- helm/charts/nico-machine-a-tron/charts/mat-k8s-controller/values.yaml
- dev/k8s/machine-a-tron-controller/README.md
- dev/k8s/machine-a-tron-controller/Makefile
- dev/k8s/machine-a-tron-controller/pkg/matclient/client_test.go
- dev/k8s/machine-a-tron-controller/go.mod
- crates/machine-a-tron/README.md
- dev/k8s/machine-a-tron-controller/pkg/matclient/types.go
- helm/charts/nico-machine-a-tron/charts/mat-k8s-controller/templates/_helpers.tpl
- helm/charts/nico-machine-a-tron/values.yaml
- helm/charts/nico-machine-a-tron/templates/deployment.yaml
- dev/k8s/machine-a-tron-controller/cmd/mat-k8s-controller/main.go
- dev/k8s/machine-a-tron-controller/pkg/controller/k8s_client.go
- dev/k8s/machine-a-tron-controller/pkg/controller/discovery.go
- dev/k8s/machine-a-tron-controller/pkg/controller/controller_test.go
poroh
approved these changes
Jul 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Implements a Kubernetes controller that dynamically creates Services for machine-a-tron mock BMC endpoints, enabling multi-pod deployments where each BMC needs stable network exposure.
Why this change?
When machine-a-tron runs in Kubernetes with multiple pods, mock BMCs need stable network exposure so NICo components can reach Redfish endpoints. This controller consumes machine-a-tron's
/machines/statusAPI and reconciles one Service per mock BMC.Machine-a-tron itself stays Kubernetes-agnostic - all K8s logic lives in this separate controller.
What's included:
Controller (
dev/k8s/machine-a-tron-controller/):nvidia-infra-controller/mat-service=truelabel/machines/statusfrom each discovered instanceHelm subchart (
helm/charts/nico-machine-a-tron/charts/mat-k8s-controller/):mat-k8s-controller.enabled=trueCI/CD:
build-mat-k8s-controllerjob to main CI workflowRelated issues
Type of Change
Testing
Manual testing verified:
Additional Notes
ServiceCIDR requirement: BMC IPs must fall within Kubernetes ServiceCIDR range