Clarification Required
The patch for 1.0.1 Kubeflow manifests now makes the ingress gateway used for prediction for Kubeflow KFServing to be kfserving-gateway instead of istio-ingressgateway.
kubeflow/manifests#1017
Refer Issues:
Whereas when we install kfserving independently without the help of kubeflow/manifests we still route the requests through istio-ingressgateway.
The current documentation refers to the istio-ingressgateway as the gateway for prediction requests.
An example is shown here:
https://github.com/kubeflow/kfserving/blob/master/docs/samples/tensorflow/README.md#run-a-prediction
CLUSTER_IP=$(kubectl -n istio-system get service istio-ingressgateway -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
Whether should we refer to kfserving-gateway or istio-ingressgateway in the docs?
/cc @yuzisun @animeshsingh @ellis-bigelow @cliveseldon
Issue-Label Bot is automatically applying the labels:
| Label | Probability |
| ------------- | ------------- |
| question | 0.76 |
Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback!
Links: app homepage, dashboard and code for this bot.
+1 to this question.
Related to this, the basic tensorflow flower example has been broken for a while. (https://github.com/kubeflow/kfserving/issues/734)
I think we shall provide both, and users choose based on the situation.
/assign @animeshsingh
/close
@yuzisun: Closing this issue.
In response to this:
/close
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
It sounds that there is a typo :
the name of the service is kfserving-ingressgateway instead of kfserving-ingressgatway (the letter 'e' is missing in gateway)
Use kfserving-ingressgatway as your INGRESS_GATEWAY if ...
should be
Use kfserving-ingressgateway as your INGRESS_GATEWAY if ...
@jal06 fixed here
https://github.com/kubeflow/kfserving/pull/752
@animeshsingh Thanks for fixing the typo.
Getting the CLUSTER_IP works now .
So I agree that this issue must be closed.
However, if I follow the documentation of the flowers-sample example, I still have a problem when getting CLUSTER_IP variable.
The problem I have now is described in 734
I am still not able to run the flowers sample (or any other example).
Please let me know if you need more infos.
Most helpful comment
+1 to this question.
Related to this, the basic tensorflow flower example has been broken for a while. (https://github.com/kubeflow/kfserving/issues/734)