Skip to content

Making inferenceModel optional #1024

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jun 19, 2025
Merged

Making inferenceModel optional #1024

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 8 additions & 4 deletions pkg/epp/requestcontrol/director.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,12 +91,16 @@ func (d *Director) HandleRequest(ctx context.Context, reqCtx *handlers.RequestCo
return reqCtx, err
}

// NOTE: The nil checking for the modelObject means that we DO allow passthrough currently.
// This might be a security risk in the future where adapters not registered in the InferenceModel
// are able to be requested by using their distinct name.
modelObj := d.datastore.ModelGet(reqCtx.Model)
if modelObj == nil {
return reqCtx, errutil.Error{Code: errutil.BadConfiguration, Msg: fmt.Sprintf("error finding a model object in InferenceModel for input %v", reqCtx.Model)}
logger.Info("No associated inferenceModel found, using default", "model", reqCtx.Model)
sheddable := v1alpha2.Sheddable
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default should probably be standard, not sheddable.

modelObj = &v1alpha2.InferenceModel{
Spec: v1alpha2.InferenceModelSpec{
ModelName: reqCtx.Model,
Criticality: &sheddable,
},
}
}

reqCtx.ResolvedTargetModel = reqCtx.Model
Expand Down
39 changes: 21 additions & 18 deletions pkg/epp/requestcontrol/director_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -268,6 +268,27 @@ func TestDirector_HandleRequest(t *testing.T) {
},
wantMutatedBodyModel: "resolved-target-model-A",
},
{
name: "nonexistent target defined, use default inference model",
schedulerMockSetup: func(m *mockScheduler) {
m.scheduleResults = defaultSuccessfulScheduleResults
},
wantReqCtx: &handlers.RequestContext{
Model: "food-review-1",
ResolvedTargetModel: "food-review-1",
TargetPod: &backend.Pod{
NamespacedName: types.NamespacedName{Namespace: "default", Name: "pod1"},
Address: "192.168.1.100",
},
TargetEndpoint: "192.168.1.100:8000",
},
wantMutatedBodyModel: "food-review-1",
reqBodyMap: map[string]interface{}{
"model": "food-review-1",
"prompt": "test prompt",
},
mockSaturationDetector: &mockSaturationDetector{isSaturated: false},
},
{

name: "request dropped (sheddable, saturated)",
Expand Down Expand Up @@ -298,24 +319,6 @@ func TestDirector_HandleRequest(t *testing.T) {
},
wantErrCode: errutil.BadRequest,
},
{
name: "invalid model defined, expect err",
reqBodyMap: map[string]interface{}{
"model": "non-existent-model",
"prompt": "test prompt",
},
mockSaturationDetector: &mockSaturationDetector{isSaturated: false},
wantErrCode: errutil.BadConfiguration,
},
{
name: "invalid target defined, expect err",
reqBodyMap: map[string]interface{}{
"model": "food-review-1",
"prompt": "test prompt",
},
mockSaturationDetector: &mockSaturationDetector{isSaturated: false},
wantErrCode: errutil.BadConfiguration,
},
{
name: "scheduler returns error",
reqBodyMap: map[string]interface{}{
Expand Down