【WIP】Support model express bootstrap acceleration by hzxuzhonghu · Pull Request #1862 · volcano-sh/kthena

hzxuzhonghu · 2026-10-09T06:55:30Z

What type of PR is this?

/kind enhancement

What this PR does / why we need it:

Depend on #1849

Which issue(s) this PR fixes:
Fixes #

Bug evidence (required for bug-related PRs):

Special notes for your reviewer:

Does this PR introduce a user-facing change?:

Copilot

🟡 Changes recommended

Bootstrap hash tracking, observation ordering, schema consistency, and required coverage must be corrected before approval.

5 open findings

Recompute observed hashes when Pods change · New
Allow empty readyURL in CRD validation · New
Record Pod hash before readiness handling · New
Add required release note for API addition · New
Add controller-manager E2E coverage · New

What changed in this PR

Adds ModelExpress-backed, source-aware bootstrap batching to ModelServing.

Changes:

Adds bootstrap strategy APIs, validation, generated clients, CRDs, and documentation.
Implements configuration hashing, environment injection, and batch admission.
Integrates bootstrap state into reconciliation and datastore tracking.

File	Description
`pkg/model-serving-controller/webhook/validator.go`	Validates bootstrap configuration.
`pkg/model-serving-controller/webhook/validator_test.go`	Tests validation behavior.
`pkg/model-serving-controller/utils/revision_data.go`	Exposes plugin normalization.
`pkg/model-serving-controller/utils/revision_data_test.go`	Verifies revision neutrality.
`pkg/model-serving-controller/plugins/modelexpress_plugin.go`	Adds managed ModelExpress plugin.
`pkg/model-serving-controller/plugins/modelexpress_plugin_test.go`	Tests plugin injection and scope.
`pkg/model-serving-controller/datastore/store.go`	Tracks role bootstrap hashes.
`pkg/model-serving-controller/datastore/store_test.go`	Tests hash tracking.
`pkg/model-serving-controller/datastore/dump.go`	Exports bootstrap hashes.
`pkg/model-serving-controller/controller/model_serving_controller.go`	Integrates batch admission.
`pkg/model-serving-controller/controller/model_serving_controller_test.go`	Updates controller tests.
`pkg/model-serving-controller/controller/model_serving_controller_bootstrap_test.go`	Tests bootstrap reconciliation.
`pkg/model-serving-controller/bootstrap/config.go`	Calculates pool configuration hashes.
`pkg/model-serving-controller/bootstrap/admitter.go`	Implements admission budgets.
`pkg/model-serving-controller/bootstrap/admitter_test.go`	Tests admission rules.
`pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go`	Adds generated deep copies.
`pkg/apis/workload/v1alpha1/model_serving_types.go`	Defines bootstrap API fields.
`pkg/apis/workload/v1alpha1/labels.go`	Defines the hash annotation.
`examples/model-serving/bootstrap-acceleration.yaml`	Adds an example workload.
`docs/proposal/modelserving-modelexpress-startup-acceleration.md`	Documents the design.
`docs/kthena/sidebars.ts`	Links the user guide.
`docs/kthena/docs/user-guide/bootstrap-acceleration.md`	Adds usage documentation.
`docs/kthena/docs/reference/crd/workload.serving.volcano.sh.md`	Updates CRD reference.
`client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go`	Adds apply configuration field.
`client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go`	Adds generated ModelExpress builder.
`client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go`	Adds generated strategy builder.
`client-go/applyconfiguration/utils.go`	Registers apply configurations.
`charts/kthena/charts/workload/crds/workload.serving.volcano.sh_modelservings.yaml`	Updates the packaged CRD.

Files not reviewed (5)

client-go/applyconfiguration/utils.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go: Generated file
pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go: Generated file

🧠 Review effort: Balanced

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

+	// BootstrapAccelerateStrategy creates role replicas in source-aware batches
+	// for peer-to-peer model weight loading. If unset, creation is unchanged.
+	// +optional
+	BootstrapAccelerateStrategy *BootstrapAccelerateStrategy `json:"bootstrapAccelerateStrategy,omitempty"`


+	admitter, err := c.newBootstrapAdmitter(ms)
+	if err != nil {
+		return fmt.Errorf("failed to build bootstrap admitter: %v", err)
+	}
+
 	// 1. Sync the number of ServingGroups to match the expected replicas defined in spec.
-	if err := c.syncServingGroupReplicas(ctx, ms, revision); err != nil {
+	if err := c.syncServingGroupReplicas(ctx, ms, revision, admitter); err != nil {


Copilot

🟡 Changes recommended

Source readiness and template identity can be misclassified, and the planned cluster-level E2E coverage is missing.

3 open findings

Pool key ignores per-Pod template hash mismatches · New
Add controller-manager E2E coverage
Add required release note for API addition

3 resolved since last review

Recompute observed hashes when Pods change
Record Pod hash before readiness handling
Allow empty readyURL in CRD validation

Files not reviewed (5)

client-go/applyconfiguration/utils.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go: Generated file
pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go: Generated file

Previously missed (2)

In code that hasn't changed since last review

Running role remains active when Pod becomes NotReady

pkg/model-serving-controller/controller/model_serving_controller.go:640

RoleRunning is sticky when a Running Pod merely becomes NotReady: updatePod's default branch does not demote the role, so this can count a currently unready replica as a bootstrap source and release another batch. Either derive source readiness from the current Pods here or update the Pod handler to transition the role out of Running on Ready→NotReady changes.

Missing E2E coverage for batch wake-ups and rolling updates

docs/proposal/modelserving-modelexpress-startup-acceleration.md:406

The promised controller-manager E2E coverage is absent: no test under test/e2e exercises batch wake-ups, real PodGroup gang cores, or rolling-update pacing. The unit tests use fake clients/managers and cannot validate these cluster-level interactions, so add the E2E described here before merging.

🧠 Review effort: Balanced

+				replicas = append(replicas, bootstrap.Replica{
+					Pool: bootstrap.PoolKey{
+						Role:             roleName,
+						RoleTemplateHash: role.RoleTemplateHash,
+						ConfigHash:       role.BootstrapConfigHash,
+					},


FAUST-BENCHOU · 2026-10-14T02:59:26Z

+}
+
+// ModelExpressConfig holds values injected into the Pods of in-scope roles.
+type ModelExpressConfig struct {


why not put these model express specific config into plugin

FAUST-BENCHOU · 2026-10-14T03:08:32Z

+    roles: ["prefill"]
+    seedReplicas: 1
+    sourceFanOut: 2
+    modelExpress:


a little wierd for me

FAUST-BENCHOU · 2026-10-14T03:19:45Z

+// ModelServingSpec defines the specification of the ModelServing resource.
+type ModelServingSpec struct {
+	// ...existing fields...
+
+	// BootstrapAccelerateStrategy creates role replicas in batches so that most of them
+	// load model weights peer-to-peer from ready replicas instead of from storage.
+	// +optional
+	BootstrapAccelerateStrategy *BootstrapAccelerateStrategy `json:"bootstrapAccelerateStrategy,omitempty"`
+}
+
+// BootstrapAccelerateStrategyProvider identifies the peer-to-peer weight transfer system.
+// +kubebuilder:validation:Enum={ModelExpress}
+type BootstrapAccelerateStrategyProvider string
+
+const (
+	BootstrapAccelerateStrategyProviderModelExpress BootstrapAccelerateStrategyProvider = "ModelExpress"
+)
+
+// BootstrapAccelerateStrategy configures source-aware creation of role replicas.
+type BootstrapAccelerateStrategy struct {
+	// Provider is the weight transfer system used by the inference engines.
+	// +kubebuilder:default=ModelExpress
+	// +optional
+	Provider BootstrapAccelerateStrategyProvider `json:"provider,omitempty"`
+
+	// Roles limits bootstrap acceleration to the listed roles. Empty means all roles.
+	// +listType=set
+	// +optional
+	Roles []string `json:"roles,omitempty"`
+
+	// SeedReplicas is the number of role replicas per role that may be created while no
+	// ready source exists. A percentage is taken of the desired role replicas
+	// (spec.replicas * role.replicas) and rounded up, e.g. 10% of 3 is 1.
+	// With gang scheduling, the effective budget is at least the role's minRoleReplicas
+	// (default: all role replicas in one ServingGroup). Must resolve to at least 1
+	// when the desired count is nonzero; a desired count of zero creates nothing.
+	// +kubebuilder:validation:XIntOrString
+	// +kubebuilder:default=1
+	// +optional
+	SeedReplicas *intstr.IntOrString `json:"seedReplicas,omitempty"`
+
+	// SourceFanOut is the number of role replicas per role that may be starting
+	// (created but not yet Running) for each ready replica of the same role.
+	// +kubebuilder:validation:Minimum=1
+	// +kubebuilder:default=1
+	// +optional
+	SourceFanOut *int32 `json:"sourceFanOut,omitempty"`
+
+	// ModelExpress configures the ModelExpress provider. If unspecified, no MX
+	// environment variables are injected and users configure the engine themselves.
+	// +optional
+	ModelExpress *ModelExpressConfig `json:"modelExpress,omitempty"`
+}
+
+// ModelExpressConfig holds values injected into the Pods of in-scope roles.
+type ModelExpressConfig struct {
+	// EngineContainers names the regular containers running the inference engine
+	// in entry and worker Pods of in-scope roles. Only these containers receive MX envs.
+	// +kubebuilder:validation:MinItems=1
+	// +listType=set
+	EngineContainers []string `json:"engineContainers"`
+
+	// ServerAddress is the gRPC address (host:port) of the ModelExpress server. It is
+	// injected as MX_SERVER_ADDRESS only if that variable is not already
+	// set in the container.
+	// +optional
+	ServerAddress string `json:"serverAddress,omitempty"`
+
+	// ReadyURL is the engine endpoint that MX polls before a source publishes its
+	// metadata; it may differ from the Pod readiness probe. It is injected as
+	// MX_ARTIFACT_READY_URL only if that variable is not already set in the container.
+	// If empty, MX uses the engine's default (http://127.0.0.1:8000/health for vLLM).
+	// +optional
+	ReadyURL string `json:"readyURL,omitempty"`
+}
+```
+
+```yaml
+spec:
+  replicas: 16
+  bootstrapAccelerateStrategy:
+    roles: ["prefill"]
+    seedReplicas: 1
+    sourceFanOut: 2
+    modelExpress:
+      engineContainers: ["engine"]
+      serverAddress: modelexpress-server.modelexpress:8001
+      readyURL: http://127.0.0.1:8000/health
+  template:
+    roles:
+    - name: prefill
+      replicas: 1
+      entryTemplate:
+        spec:
+          containers:
+          - name: engine
+            image: <vllm-0.23+-with-modelexpress>
+            args: ["serve", "/models/deepseek-v4", "--load-format", "modelexpress", "--tensor-parallel-size", "8"]
+            readinessProbe:
+              httpGet: {path: /health, port: 8000}
+```


Heres my example.Plugin way i think is more convient to introduce more P2P backends in future

// ModelServingSpec type ModelServingSpec struct { // ...existing fields (replicas, schedulerName, plugins, template, ...) // BootstrapStrategy controls how role replicas are created during startup, // scale-up, recovery, and rolling-update re-creation. // If unset, all missing replicas are created as today. // This field is revision-neutral: edits do not trigger a rollout. // +optional BootstrapStrategy *BootstrapStrategy `json:"bootstrapStrategy,omitempty"` } // BootstrapStrategy configures source-aware creation pacing of role replicas. // It does not select a weight-transfer implementation; engines/plugins do that. type BootstrapStrategy struct { // Roles limits pacing to the listed roles. Empty means all roles. // +listType=set // +optional Roles []string `json:"roles,omitempty"` // SeedReplicas is the max role replicas per role that may be created while // no ready source exists. Int or percent of (spec.replicas * role.replicas), // rounded up. Effective budget is at least one gang core when gang scheduling // is enabled. // +kubebuilder:validation:XIntOrString // +kubebuilder:validation:XValidation:rule="type(self) == int ? self >= 1 : self.matches('^([1-9]|[1-9][0-9]|100)%$')" // +kubebuilder:default=1 // +optional SeedReplicas *intstr.IntOrString `json:"seedReplicas,omitempty"` // SourceFanOut is the max starting (created, not yet Running) role replicas // allowed per ready replica of the same role. Effective budget is at least // one gang core when gang scheduling is enabled. // +kubebuilder:validation:Minimum=1 // +kubebuilder:default=1 // +optional SourceFanOut *int32 `json:"sourceFanOut,omitempty"` } // ModelExpress stays out of Spec strategy. It is a BuiltIn plugin config: // // plugins: // - name: modelexpress // type: BuiltIn // scope: { roles: ["prefill"], target: All } // config: { engineContainers, serverAddress, readyURL } // // Suggested JSON shape for plugin Config (not a top-level CRD struct field): type ModelExpressPluginConfig struct { EngineContainers []string `json:"engineContainers"` ServerAddress string `json:"serverAddress,omitempty"` ReadyURL string `json:"readyURL,omitempty"` }

spec: replicas: 16 schedulerName: volcano # Controller: When to Create bootstrapStrategy: roles: ["prefill"] seedReplicas: 1 sourceFanOut: 2 # Plugins: How to modify the Pod (to match existing plugins) plugins: - name: modelexpress type: BuiltIn scope: roles: ["prefill"] # It's clearer when aligned with bootstrapStrategy.roles target: All config: engineContainers: ["engine"] serverAddress: modelexpress-server.modelexpress:8001 readyURL: http://127.0.0.1:8000/health template: roles: - name: prefill replicas: 1 entryTemplate: spec: containers: - name: engine image: <vllm-0.23+-with-modelexpress> args: - serve - /models/deepseek-v4 - --load-format - modelexpress - --tensor-parallel-size - "8" readinessProbe: httpGet: {path: /health, port: 8000}

Signed-off-by: Zhonghu Xu <xuzhonghu.xzh@alibaba-inc.com>

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

Copilot

🟡 Changes recommended

Bootstrap admission budgets legacy gang recovery using a reduced current replica count, causing perpetual denial when no Ready source exists and preventing partition-protected replicas from recovering.

7 open issues

6 files not reviewed

client-go/applyconfiguration/utils.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstraprole.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go: Generated file
pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go: Generated file

_{Review effort: Balanced}

+			seed = value
+		}
+	}
+	return min(desired, max(a.GangMinimum(spec), max(1, seed)))


+					TemplateHash: role.RoleTemplateHash,
+					ConfigHash:   hash,
+					HashKnown:    known,
+					Running:      role.Status == datastore.RoleRunning,


 		}
 		klog.V(4).Infof("scaleUpServingGroups: creating new ServingGroup at ordinal=%d with newRevision=%s for modelServing=%s", ordinal, newRevision, utils.GetNamespaceName(ms))
-		if err := createServingGroup(ordinal, newRevision, ms.Spec.Template.Roles); err != nil {
+		if _, err := createServingGroup(ordinal, newRevision, ms.Spec.Template.Roles); err != nil {


Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

Copilot

🟡 Changes recommended

Missing ModelExpress plugin symbols prevent compilation, while revision-unaware bootstrap admission can exceed seed budgets or indefinitely block partition-protected replica recovery.

8 open issues

6 files not reviewed

client-go/applyconfiguration/utils.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstraprole.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go: Generated file
pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go: Generated file

_{Review effort: Balanced}

+	if mxPlugin := plugins.NewModelExpressPlugin(ms, c.modelExpressServerAddress); mxPlugin != nil {
+		chain = chain.WithManagedPlugin(mxPlugin, plugins.ModelExpressPluginScope(ms))


Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

volcano-sh-bot · 2026-11-06T01:57:59Z

[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 hzxuzhonghu. 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:

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

Copilot

🟡 Changes recommended

Gang-recovery admission can deadlock when seed budgets are exhausted by incomplete gangs or conflict with historical replica layouts, leaving workloads unable to recover.

9 open issues

1 resolved since last review

Historical replicas are charged against the wrong template pool

6 files not reviewed

client-go/applyconfiguration/utils.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstrapacceleratestrategy.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/bootstraprole.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelexpressconfig.go: Generated file
client-go/applyconfiguration/workload/v1alpha1/modelservingspec.go: Generated file
pkg/apis/workload/v1alpha1/zz_generated.deepcopy.go: Generated file

_{Review effort: Balanced}

+	bootstrapAdmits := func(ordinal int, roleToApply workloadv1alpha1.Role) bool {
+		if bootstrap.AdmitterFrom(ctx).AdmitRole(roleToApply) {


+## Prerequisites
+
+- An MX-enabled inference engine (for example vLLM with `--load-format modelexpress`) that can reach a ModelExpress server.
+- The controller manager is started with `--modelexpress-server-address=<host>:<port>` (Helm: `controllerManager.modelExpressServerAddress`). The address is injected into engine containers as `MX_SERVER_ADDRESS`.


Copilot AI balanced review requested due to automatic review settings October 9, 2026 06:55

github-project-automation Bot added this to Kthena Release Tracking Oct 9, 2026

github-project-automation Bot moved this to Accepted in Kthena Release Tracking Oct 9, 2026

volcano-sh-bot added the kind/enhancement New feature or request label Oct 9, 2026

volcano-sh-bot requested review from FAUST-BENCHOU, LiZhenCheng9527, YaoZengzeng, git-malu and kube-gopher October 9, 2026 06:55

volcano-sh-bot added the size/XXL label Oct 9, 2026

Copilot started reviewing on behalf of hzxuzhonghu October 9, 2026 06:56

View session

Copilot AI reviewed Oct 9, 2026

View reviewed changes

Copilot AI balanced review requested due to automatic review settings October 9, 2026 11:14

Copilot started reviewing on behalf of hzxuzhonghu October 9, 2026 11:15

View session

Copilot AI reviewed Oct 9, 2026

View reviewed changes

Comment thread pkg/model-serving-controller/controller/model_serving_controller.go Outdated

Comment on lines +634 to +639

replicas = append(replicas, bootstrap.Replica{

Pool: bootstrap.PoolKey{

Role: roleName,

RoleTemplateHash: role.RoleTemplateHash,

ConfigHash: role.BootstrapConfigHash,

},

FAUST-BENCHOU reviewed Oct 14, 2026

View reviewed changes

hzxuzhonghu changed the title ~~Support model express bootstrap acceleration~~ 【WIP】Support model express bootstrap acceleration Oct 20, 2026

volcano-sh-bot added the do-not-merge/work-in-progress label Oct 20, 2026

hzxuzhonghu and others added 4 commits November 5, 2026 16:40

Added proposal to support model express startup accelerate

7dd1bb0

Signed-off-by: Zhonghu Xu <xuzhonghu.xzh@alibaba-inc.com>

Added api

67d717c

Signed-off-by: Zhonghu Xu <xuzhonghu.xzh@alibaba-inc.com>

remove fanout field

33a9dba

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

Implement bootstrap accelerate with mx

5646929

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

hzxuzhonghu force-pushed the mx-implement branch from c883303 to 5646929 Compare November 5, 2026 11:23

Copilot AI balanced review requested due to automatic review settings November 5, 2026 11:23

Copilot started reviewing on behalf of hzxuzhonghu November 5, 2026 11:24

View session

Copilot AI reviewed Nov 5, 2026

View reviewed changes

hzxuzhonghu added 2 commits November 5, 2026 19:44

moveplugin

0007464

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

update

1f388a7

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

Copilot AI balanced review requested due to automatic review settings November 5, 2026 12:00

Copilot started reviewing on behalf of hzxuzhonghu November 5, 2026 12:01

View session

Copilot AI reviewed Nov 5, 2026

View reviewed changes

Comment thread pkg/model-serving-controller/controller/model_serving_controller.go

Comment on lines +3029 to +3030

if mxPlugin := plugins.NewModelExpressPlugin(ms, c.modelExpressServerAddress); mxPlugin != nil {

chain = chain.WithManagedPlugin(mxPlugin, plugins.ModelExpressPluginScope(ms))

hzxuzhonghu added 3 commits November 5, 2026 22:40

xx

b7c1970

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

update

e356413

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

update

ed5cdfc

Signed-off-by: Zhonghu Xu <zhhxu2011@163.com>

Copilot AI balanced review requested due to automatic review settings November 6, 2026 01:57

Copilot started reviewing on behalf of hzxuzhonghu November 6, 2026 01:58

View session

Copilot AI reviewed Nov 6, 2026

View reviewed changes

Navigation Menu

Repository navigation

【WIP】Support model express bootstrap acceleration - #1862

【WIP】Support model express bootstrap acceleration#1862
hzxuzhonghu wants to merge 9 commits into
volcano-sh:mainfrom
hzxuzhonghu:mx-implement

hzxuzhonghu commented Oct 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

volcano-sh-bot commented Nov 6, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		if mxPlugin := plugins.NewModelExpressPlugin(ms, c.modelExpressServerAddress); mxPlugin != nil {
		chain = chain.WithManagedPlugin(mxPlugin, plugins.ModelExpressPluginScope(ms))

		bootstrapAdmits := func(ordinal int, roleToApply workloadv1alpha1.Role) bool {
		if bootstrap.AdmitterFrom(ctx).AdmitRole(roleToApply) {

Conversation

hzxuzhonghu commented Oct 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Choose a reason for hiding this comment

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Choose a reason for hiding this comment

Uh oh!

FAUST-BENCHOU Oct 14, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

volcano-sh-bot commented Nov 6, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants