This repository packages the Auto Labeler Labeler stack (MongoDB, FastAPI backend, Vite frontend) for deployment on Kubernetes.
Use the helper script to build and push timestamped backend and frontend images:
./build.shYou can override defaults by exporting variables before running the script, for example:
export BRANCH_NAME=beta
export TAG=stable
./build.shThe script pushes images to:
us-docker.pkg.dev/teknoir/gcr.io/auto-labeler-backendus-docker.pkg.dev/teknoir/gcr.io/auto-labeler-frontend
Make note of the generated tags so you can reference them in your deployment values.
Edit deploy-manifest.yaml and set the backend and frontend image tags (and any MongoDB or Istio settings) under valuesContent. The default manifest ships with sensible placeholders; replace them with the tags produced by the build step. If you create the GCS credentials secret separately, point backend.gcsServiceAccountKeySecret at auto-labeler-backend-gcs (or your chosen name).
Apply the Rancher HelmChart resource:
./deploy.shBy default, the accompanying VirtualService routes traffic under /dataset-curation/auto-labeler-labeler through the teknoir/teknoir-gateway. Adjust the base path or host list in deploy-manifest.yaml if your environment differs.
You can install the chart with Helm once the repository is published:
helm repo add teknoir-auto-labeler https://teknoir.github.io/auto-labeler-labeler-helm/
helm install auto-labeler teknoir-auto-labeler/auto-labeler-labeler \
--namespace dataset-curation \
-f charts/auto-labeler-labeler/values.yamlOverride values.yaml as needed for credentials, persistence, Istio routing, and GCS signing. Refer to charts/auto-labeler-labeler/values.yaml for the full list of configurable options.
Install the secret with
kubectl create secret generic auto-labeler-backend-gcs \
--from-file=auto-label-backend.json=/path/to/auto-label-backend.json \
-n <namespace>