Charts: [stable/traefik] service type ClusterIP does not work with externalTrafficPolicy
I have an internal instance of the traefik chart and i have the service type set to ClusterIP. Because of this, i get an error when attempting to deploy traefik.
"Invalid value: "Cluster": ExternalTrafficPolicy can only be set on NodePort and LoadBalancer service"
Should the condition for externalTrafficPolicy be changed to only be added if the service type is not ClusterIP?
gmorse81
All 6 comments
Sound like it, yes. Although I can't quite understand the use case that would lead one to select ClusterIP as the service type. The point of an ingress controller (as the "ingress" in its name would imply) is to route traffic coming _into_ the cluster. If you're using Traefik with service type ClusterIP, it sounds like you're using it to route traffic _within_ the cluster instead. That sounds odd and inefficient.
But again, yes. If there's a legitimate use case for this, then it sounds like setting or not setting ExternalTrafficPolicy on the basis of the service type is necessary.
krancour
on 5 Jun 2018
An Ingress (object) is really just a way to define configuration for a reverse proxy (ingress controller). in my use case I have an external ingress controller which handles traffic coming into the cluster. The external ingress controller routes most of the traffic (but not all) to a couple middleware services. Those middleware services are not smart enough to route traffic based on path, so they blindly forward the traffic to the internal ingress (running as ClusterIP). The internal ingress routes to the correct service based on path.
gmorse81
on 6 Jun 2018
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Any further update will cause the issue/pull request to no longer be considered stale. Thank you for your contributions.
![stale[bot] picture](https://avatars.githubusercontent.com/in/1724?v=4&s=40)
stale[bot]
on 19 Aug 2018
This issue is being automatically closed due to inactivity.
stale[bot]
on 2 Sep 2018
Just a reminder that this is still an issue
DanielHabenicht
on 12 Sep 2019
Yup - still an issue. It seems like setting --set externalTrafficPolicy= will fix it.
I'm using a ClusterIP because I'm running keepalived VIPs in my cluster.
Tawmu
on 22 Sep 2019
Related issues
tklovett
路
3Comments
michas2
路
3Comments
musabmasood
路
3Comments
vsliouniaev
路
3Comments
kminehart
路
3Comments
Most helpful comment
An Ingress (object) is really just a way to define configuration for a reverse proxy (ingress controller). in my use case I have an external ingress controller which handles traffic coming into the cluster. The external ingress controller routes most of the traffic (but not all) to a couple middleware services. Those middleware services are not smart enough to route traffic based on path, so they blindly forward the traffic to the internal ingress (running as ClusterIP). The internal ingress routes to the correct service based on path.