Mark Winter

There is a PR for it here https://github.com/kserve/models-web-app/pull/7 Still waiting for a review @kimwnasptd

AFAIK, offline messaging will work by nodes closest to the receiver storing the messages. When a node comes online, it will then receive these messages and check if any are...

Disabling UDP will help, and there was an update yesterday that may help things further with regards to battery and data usage, but otherwise you will need to open an...

@MimeZoe0628 What uri were you using? http?

I noticed env vars were also being applied twice ``` spec: predictor: model: env: - name: MLSERVER_MODEL_NAME value: core - name: MLSERVER_MODEL_URI value: /mnt/models - name: MLSERVER_MODEL_NAME value: core -...

When applying the annotation on `InferenceService` it gets propagated to `spec.template.metadata.annotations` in the ksvc. But `serving.knative.dev/rollout-duration` is only supported in the top level `metadata.annotations` of a ksvc unlike other settings...

@pvaneck I also had to add `/v1/models/*` path to allow Transformer->Predictor requests. Without it the Transformer gets a 403 forbidden ```BASH $ curl -v -H "Cookie: authservice_session=$SESSION" -H "Host: ${SERVICE_HOSTNAME}"...

To add to my comment above I separated the policy into two and limited them to the Predictor and Transformer with a selector. ``` selector: matchLabels: component: predictor ``` ```...

@jiaozhentian It should look like this (and one more with `component: transformer`) ```YAML apiVersion: security.istio.io/v1beta1 kind: AuthorizationPolicy metadata: name: allowlist-by-paths namespace: istio-system spec: selector: matchLabels: component: predictor action: ALLOW rules:...

@goswamig The `AuthorizationPolicy` has a selector for pods with the label `component: predictor` to limit its affect to kserve pods only.