This Helm chart contains the Signalen classification service. It automatically classifies nuisance reports using machine learning.
The classification service expects a trained machine learning model in the folder /models
on the persistent volume. For more information how to train a model see the documentation.
Parameter | Description | Default |
---|---|---|
image.repository |
The repository of the Docker image | signalen/frontend |
image.tag |
The tag of the Docker image | latest |
signalsCategoryUrl |
This prefix that will be added to every category slug | http://api.signals.example.com/signals/v1/public/terms |
replicaCount |
The number of API replicas | 1 |
persistence.enabled |
Enable persistence | true |
persistence.size |
Specify the size of the PVC | 1Gi |
persistence.existingClaim |
Name of an existing PVC to use | null |
persistence.accessModes |
The accessModes of the PVC | { ReadWriteOnce } |
persistence.storageClassName |
The storageClassName of the PVC | "" |
ingress.enabled |
Expose the API through an ingress | true |
ingress.annotations |
Additional annotations on the API ingress | {} |
ingress.hosts |
The hosts of the ingress | {} |
ingress.tls |
The TLS settings of the ingress | {} |
Check values.yaml for all the possible configuration options.