models-web-app
models-web-app copied to clipboard
kserve
KServe models web UI
Currently there is no activity in this repo. I have a good understanding of the code and am able to make significant updates to it (see my currently open PRs)....
Right now the manifests are using the hardcoded variables for the Istio gateways https://github.com/kserve/models-web-app/blob/a1cef56b037d6b308937854870ebeac25001f0e6/config/base/istio.yaml#L8 We should update the manifests to directly utilize the required Gateways
PR https://github.com/kserve/models-web-app/pull/27 updated the app's prefix in the VirtualService to `kserve-endpoints`, but it didn't configure the backend to use that value as well. The manifests should also set the `APP_PREFIX`...
Going forward we will need to have a versioning policy for the web app. This will also help integrating the app with the KServe project and making it more clear...
Since we now have a distinct repo for the app we will need to re-evaluate the way we will hanvle CI/CD for: * Conintuously running unit tests on PRs *...
Lets use this issue to discuss the roadmap of the newly added [web app](https://github.com/kubeflow/kfserving/tree/master/web-app) and document the next steps. Also really looking forward to hear users' feedback and what features...
The kubeflow models web app (v0.6.0) does not display models although it gets a 200 response for `/models/api/namespaces/namespace/inferenceservices` ``` { "inferenceServices":[ { "apiVersion":"serving.kubeflow.org/v1beta1", "kind":"InferenceService", "metadata":{ "annotations":{ "autoscaling.knative.dev/maxScale":"5", "autoscaling.knative.dev/minScale":"1", "kubectl.kubernetes.io/last-applied-configuration":"{""apiVersion"":"serving.kubeflow.org/v1beta1",""kind"":""InferenceService"",""metadata"":{""annotations"":{""autoscaling.knative.dev/maxScale"":"5",""autoscaling.knative.dev/minScale"":"1"},""labels"":{""app"":""iris-amy"",""app.kubernetes.io/instance"":""iris-amy"","controller-tools.k8s.io":"1.0",""env"":"mlk8s-dev",""sidecar.istio.io/inject"":"false",""version"":"0.0.20"},""name"":""iris-amy"",""namespace"":""abachir""},""spec"":{""predictor"":{""containers"":[{""env"":[{""name"":""PATH_TO_MODEL"",""value"":""/mnt/models/iris/models/2021-05-05T18":"47":18.359991/"},{""name"":""STORAGE_URI"",""value"":""pvc":"},{""name"":""MODEL_VERSION"",""value"":"0.0.20"}],""image"":""iris-kfserving":0.0.20",""name"":""kfserving-container"",""resources"":{""limits"":{""cpu"":"1",""memory"":"6Gi"},""requests"":{""cpu"":"250m",""memory"":"4Gi"}}}]}}} "...
/kind bug **What steps did you take and what happened:** 1. create an inference service ```yaml apiVersion: "serving.kubeflow.org/v1beta1" kind: "InferenceService" metadata: name: "flower-sample" annotations: serving.kubeflow.org/raw: "true" spec: predictor: tensorflow: storageUri:...
/kind feature **Describe the solution you'd like** We should add e2e tests to ensure the web app is compatible with the changes in the `InferenceService` CRD. I believe these could...
/kind feature The web app currently seems to assume that grafana will be accessible via the same ingress as kubeflow under the path `/grafana`. We have a separated prometheus+grafana that...