Mlflow and k8s from bitnami (2.18.0)

[DaniilAnd]

Hi! I use mlflow k8s from bitnami (2.18.0-debian-12-r0),
image:
registry: docker.io
repository: bitnami/mlflow
tag: 2.18.0-debian-12-r0

and i rebuild Dockerfile (screenshot 1).
After i add extraenvs in helm chart (screenshot 2) and install new version with plugin OIDC, but mlflow running with error from sqlalchemy (screenshot 3), I also added enable auth (screenshot 4).

Could there be a conflict between the first and fourth screenshot ?

)

[kharkevich]

Hello.
You need to configure databases for both the store and the users.
Regarding the Bitnami image, I have no information about its compatibility or configuration. It would be better to contact them directly.

[DaniilAnd]

@kharkevich Hi! I use for users external database. Can I take out the store and add external storе ?

[ion-elgreco]

The bitnami chart is compatible, I have it working

[DaniilAnd]

@ion-elgreco Hi! Are you using the Entry ID cloud or a local ADFS ?

[DaniilAnd]

This answer from ADFS

[rojo1997]

@ion-elgreco can you share an example of a values.yaml? Thanks in advance

[ion-elgreco]

@rojo1997 Here is the bitnami values.yaml

tracking:
  enabled: true
  auth:
    enabled: false
  command: []
  args: []
  extraArgs:
    - --app-name=oidc-auth
  extraEnvVars:
    - name: OIDC_CLIENT_ID
      value: -
    - name: OIDC_DISCOVERY_URL
      value: "-/.well-known/openid-configuration"
    - name: OIDC_REDIRECT_URI
      value: "-/callback"
    - name: OIDC_PROVIDER_DISPLAY_NAME
      value: "Login"
    - name: OIDC_SCOPE
      value: "openid email profile"
    - name: OIDC_ADMIN_GROUP_NAME
      value: "-"
    - name: OIDC_GROUP_NAME
      value: "-"
    - name: REQUESTS_CA_BUNDLE
      value: "/usr/local/share/ca-certificates/bundle.crt"

extraDeploy:
  # Create secret for mlflow deployment
  - apiVersion: v1
    kind: Secret
    metadata:
      name: mlflow-secrets
    stringData:
      AZURE_STORAGE_ACCESS_KEY:
      OIDC_CLIENT_SECRET:
      OIDC_USERS_DB_URI: 

And then custom image:

ARG VERSION
FROM docker.io/bitnami/mlflow:$VERSION
ARG VERSION

USER root
RUN update-ca-certificates
RUN pip install mlflow-oidc-auth==3.2.0 mlflow-skinny==$(echo $VERSION | awk -F'-' '{print $1}')

# Create a writable directory for flask_session
RUN mkdir -p /tmp/flask_session
RUN chmod -R 777 /tmp/flask_session

# Set the environment variable for the writable directory
ENV SESSION_FILE_DIR=/tmp/flask_session

# Switch back to the original user
USER 1001

# Set the working directory to the writable directory
WORKDIR /opt/bitnami/mlflow