chore: use separate docker image for mlflow server (#101)

* Use mlflow-skinny instead of mlflow
* Update example dockerfile to have multi-stage; one for udf and one for mlflow
* Update docs to use quay.io images, instead of requiring build

Signed-off-by: Avik Basu <avikbasu93@gmail.com>

.flake8

@@ -1,6 +1,5 @@
 [flake8]
-ignore = |
-    E203, F821
+ignore = E203, F821
 exclude = .git,__pycache__,docs/source/conf.py,old,build,dist
 max-complexity = 10
 max-line-length = 100

docs/quick-start.md

@@ -72,10 +72,8 @@ Once Numaflow is installed, create a simple Numalogic pipeline, which takes in t
 
 For building this pipeline, navigate to [numalogic-simple-pipeline](https://github.com/numaproj/numalogic/tree/main/examples/numalogic-simple-pipeline) under the examples folder and execute the following commands.
 
-1. Build the docker image, import it to k3d, and apply the pipeline. *Note Make sure the pipeline and, numaflow controllers and isbsvc pods are running in the same namespace (`default` in this case).*
+1. Apply the pipeline. *Note Make sure the pipeline and, numaflow controllers and isbsvc pods are running in the same namespace (`default` in this case).*
 ```shell
-docker build -t numalogic-simple-pipeline:v1 . && k3d image import docker.io/library/numalogic-simple-pipeline:v1
-
 kubectl apply -f numa-pl.yaml
 ```
 2. To verify if the pipeline has been deployed successfully, check the status of each pod.
@@ -101,9 +99,9 @@ numalogic-simple-pipeline-in-0-tmd0v               1/1     Running   0
 
 Once the pipeline has been created, the data can be sent to the pipeline by port-forwarding the input vertex.
 
-1. Port-forward to the http-source vertex
+1. Port-forward to the http-source vertex. From the above pod output, this would be:
    ```shell
-   kubectl port-forward numalogic-simple-pipeline-in-0-xxxxx 8443
+   kubectl port-forward numalogic-simple-pipeline-in-0-tmd0v 8443
    ```
 
 2. Send the data to the pod via curl

examples/numalogic-simple-pipeline/.dockerignore

@@ -0,0 +1,6 @@
+.git/
+__pycache__/
+**/__pycache__/
+*.py[cod]
+*$py.class
+.idea/

examples/numalogic-simple-pipeline/Dockerfile

@@ -1,8 +1,13 @@
-FROM python:3.9.12-slim
+####################################################################################################
+# builder: install needed dependencies
+####################################################################################################
+
+FROM python:3.10-slim-bullseye AS builder
+
 ENV PYTHONFAULTHANDLER=1 \
   PYTHONUNBUFFERED=1 \
   PYTHONHASHSEED=random \
-  PIP_NO_CACHE_DIR=off \
+  PIP_NO_CACHE_DIR=on \
   PIP_DISABLE_PIP_VERSION_CHECK=on \
   PIP_DEFAULT_TIMEOUT=100 \
   POETRY_VERSION=1.2.2 \
@@ -13,6 +18,7 @@ ENV PYTHONFAULTHANDLER=1 \
   VENV_PATH="/opt/pysetup/.venv"
 
 ENV PATH="$POETRY_HOME/bin:$VENV_PATH/bin:$PATH"
+
 RUN apt-get update \
     && apt-get install --no-install-recommends -y \
         curl \
@@ -25,14 +31,36 @@ RUN apt-get update \
     # install dumb-init
     && wget -O /dumb-init https://github.com/Yelp/dumb-init/releases/download/v1.2.5/dumb-init_1.2.5_x86_64 \
     && chmod +x /dumb-init \
-    \
-    # install poetry - respects $POETRY_VERSION & $POETRY_HOME
     && curl -sSL https://install.python-poetry.org | python3 -
 
+####################################################################################################
+# mlflow: used for running the mlflow server
+####################################################################################################
+FROM builder AS mlflow
+
+WORKDIR $PYSETUP_PATH
+COPY ./pyproject.toml ./poetry.lock ./
+RUN poetry install --only mlflowserver --no-cache --no-root && \
+    rm -rf ~/.cache/pypoetry/
+
+ADD . /app
+WORKDIR /app
+
+RUN chmod +x entry.sh
+
+ENTRYPOINT ["/dumb-init", "--"]
+CMD ["/app/entry.sh"]
+
+EXPOSE 5000
+
+####################################################################################################
+# udf: used for running the udf vertices
+####################################################################################################
+FROM builder AS udf
 
 WORKDIR $PYSETUP_PATH
-COPY ./pyproject.toml ./poetry.lock $PYSETUP_PATH
-RUN poetry install --without dev --no-cache --no-root && \
+COPY ./pyproject.toml ./poetry.lock ./
+RUN poetry install --without mlflowserver --no-cache --no-root && \
     rm -rf ~/.cache/pypoetry/
 
 ADD . /app

examples/numalogic-simple-pipeline/numa-pl.yaml

@@ -18,7 +18,7 @@ spec:
       min: 1
     udf:
       container:
-        image: docker.io/library/numalogic-simple-pipeline:v1
+        image: quay.io/numaio/numalogic/example-udf
         env:
           - name: WIN_SIZE
             value: "12"
@@ -31,7 +31,7 @@ spec:
       min: 1
     udf:
       container:
-        image: docker.io/library/numalogic-simple-pipeline:v1
+        image: quay.io/numaio/numalogic/example-udf
         env:
           - name: WIN_SIZE
             value: "12"
@@ -44,7 +44,7 @@ spec:
       min: 1
     udf:
       container:
-        image: docker.io/library/numalogic-simple-pipeline:v1
+        image: quay.io/numaio/numalogic/example-udf
         env:
           - name: WIN_SIZE
             value: "12"
@@ -65,7 +65,7 @@ spec:
         env:
           - name: WIN_SIZE
             value: "12"
-        image: docker.io/library/numalogic-simple-pipeline:v1
+        image: quay.io/numaio/numalogic/example-udf
         args:
           - python
           - starter.py
@@ -106,7 +106,7 @@ spec:
         app: mlflow
     spec:
       containers:
-        - image: docker.io/library/numalogic-simple-pipeline:v1
+        - image: quay.io/numaio/numalogic/example-mlflow
           name: mlflow
           args:
           - server