-
Notifications
You must be signed in to change notification settings - Fork 835
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.
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
Internal error occurred: failed calling webhook #877
Comments
Can you try kubectl describe sdep , you can see if what is the deployment failed or succeed there. |
I think the error relates to the webhook for KFServing. It checks all Pods that get deployed, including Pods created by the Seldon operator (and any other operator too). I guess this is a kubeflow v0.6 install?
|
@ryandawsonuk I have only these pods in kubeflow namespace
and I installed seldon-core in separate workspace: |
@elouayas I do not see any failed or success status!
|
You should have something like that
I think something is wrong in your yaml definition |
Could you check |
@ryandawsonuk I get this:
Can you explain, why did you need to remove kfserving? My deployment only needs to use Seldon Operator really and not kubeflow. Is there any way to separate those two out? |
The error getting reported is "failed calling webhook* "kfservice.kfserving-webhook-server.deployment-mutator". It may be reported by seldon but it's not actually coming from Seldon. The seldon operator is just trying to create resources and they're failing because the kfserving webhook intercepts the requests to the kubernetes API and tries to modify the resources. |
It was to do with kubeflow installation - It was not done properly. Some components were not up. Uninstalling it helped getting rid of that issue - at the very least. |
Glad it fixed! |
Hi Sandhya @SandhyaaGopchandani |
My seldon deployment appears when I check:
kubectl get sdep
but not whenkubectl get deployments
Also, Nothing gets returned when I check the status of the deployment:
kubectl get seldondeployments.machinelearning.seldon.io seldon-model -n seldon-model -o jsonpath='{.status.state}'
Also, there are no pods allocated to the deployment.
Further investigating
seldon-core operator controller manager pod
, I found this error message:.....{"level":"error","ts":1568906205.4712648,"logger":"kubebuilder.controller","msg":" *Reconciler error* ","controller":"seldondeployment-controller","request":"seldon-model/seldon-model","error":" *Internal error occurred: failed calling webhook* \"kfservice.kfserving-webhook-server.deployment-mutator\": Post https://kfserving-webhook-server-service.kubeflow.svc:443/mutate-deployments?timeout=30s: service \"kfserving-webhook-server-service\" not found","stacktrace":"github.com/seldonio/seldon-operator/vendor/github.com/go-logr/zapr.(*zapLogger).Error......."}
The same deployment works okay on my local machine. So, I do not think there is issue in yaml file. But this server does have kubeflow installed. I installed seldon-core separately using Helm.
The deployment seem to be looking for kubeflow component kfserving-webhook. Is there a way to create a deployment without it interacting with kubeflow?
The text was updated successfully, but these errors were encountered: