Warning
This Helm chart is deprecated and won't be further maintained. Please head to deepset's Helm repository for a list of maintained charts.
Here you can read the documentation to a haystack deployment using helm for Kubernetes.
Haystack has 3 main architectural components:
- haystack API
- haystack UI
- datastore (elasticsearch)
In the following directory you can find helm charts for haystack API (haystack-api
) and haystack UI (haystack-ui
) deployments.
Elasticsearch is not included since it may vary how and where users want to run their Elasticsearch cluster.
- functioning Kubernetes cluster
- network level access to the Kubernetes cluster API
- sufficient permissions to deploy helm charts
MISSING: elasticsearch deployment guide
- Go to the helm chart directory in haystack repository (
cd $HAYSTACK_REPOSITORY/helm
) - First we deploy the
haystack-api
helm chart, for that we need to runhelm install haystack-api-deployment haystack-api
. This will deploy the API to the namespace in the current Kubernetes context. - To validate the deployment you can run
helm list
orkubectl get pods
and see if the haystack API pods are in running state. - If all worked well we can now deploy the UI component, for that we need to run
helm install haystack-ui
. This will deploy the UI to the namespace in the current Kubernetes context. Note that you should not change context between deployments if you don’t know Kubernetes that well! - Voila! All components are deployed! See the section about how to access your deployments!
MISSING
MISSING
MISSING
- Run
helm package -u .
- Run
helm repo index --url https://deepset-ai.github.io/haystack-helm/charts
- Switch to
helm-repository
branch - Move .tgz files from poin 1 into the charts folder and update
index.yaml
with the generated one from point 2 - Commit & push