Making HTTP calls to local InferenceGraph

Hi folks,

I’m following this guide to setup the iris sequence inference graph. My graph is running and I can make requests to its inference services succesfully, but I can’t figure out how to make an http request to the graph itself. I am developing locally and would like to just use port forwarding.

The provided guide uses curl http://model-chainer.default.10.166.15.29.sslip.io -d @./iris-input.json, but I don’t have DNS setup. Could someone point me to how to achieve the same using the Ingress gateway with host header?

I tried the below without success. curl -v -H "Host: ${SERVICE_HOSTNAME}" http://${INGRESS_HOST}:${INGRESS_PORT} -d @./iris-input.json

Many thanks! Appreciate it.

[rachitchauhan43]

First do port-forwarding like this:

kubectl port-forward --namespace istio-system svc/${INGRESS_GATEWAY_SERVICE} 8080:80
# start another terminal
export INGRESS_HOST=localhost
export INGRESS_PORT=8080

Then, make a call: curl -v -H "Host: ${IG_NAME}.${NAMESPACE}.svc.cluster.local" http://localhost:8080/v1/inferencegraphs/${IG_NAME} -d @./input.json -v

[MichalPitr]

@rachitchauhan43 Yes, closing!