Merge pull request #106 from NealAnalyticsLLC/mlflow-on-azure-stack

Mlflow on azure stack

Author: initmahesh

Research/mlflow-on-azure-stack/Readme.md

@@ -1,9 +1,10 @@
 # MLflow and Kubeflow Installation On Kubernetes
 
-This guide demonstrates how to install a remote MLflow Tracking Server on Kubernetes. The instructions below demonstrate how to install using a Cloud Native Application Bundle (CNAB). Please see the document referenced below for manual installation instructions.
+This guide demonstrates how to install a remote MLflow Tracking Server & Kubeflow on Kubernetes. The instructions below demonstrate how to install using a Cloud Native Application Bundle (CNAB). Please see the document referenced below for manual installation instructions.
 
 **Reference Material:**
-- [Manual Installation Instructions](./docs/manual_installation.md)
+- [Manual Installation Instructions for Mlflow](./docs/manual_installation.md)
+- [Manual Installation Instructions for Kubeflow](/Research/kubeflow-on-azure-stack/Readme.md)
 
 **Prerequisite:**
 - Will need the k8 cluster ".kubeconfig" file on your local machine to execute commands on the k8 cluster 
@@ -17,95 +18,8 @@ Make sure you have Porter installed. You can find the installation instructions
 
 **NOTE:** be sure to add porter to your PATH so it can find the binaries
 
+## Step 2 : Install CNAB Packages
+We have CNAB packages for Kubeflow & MLFlow, use any of the below link to install 
 
-## Step 2: Build Porter CNAB
-First you will need to navigate to porter directory in the repository. For example 
-
-```sh
-cd ./research/mlflow-on-azure-stack/porter
-```
-Change the file permissions
-
-```sh
-chmod 777 mlflow.sh
-chmod 777 kubeflow.sh
-```
-Next, you will build the porter CNAB
-
-```sh
-porter build
-```
-
-## Step 3: Generate Credentials 
-This step is needed to connect to your Kubernetes cluster
-
-```sh
-porter credentials generate 
-```
-Enter path to your kubeconfig file when prompted
-
-Validate that your credential is present by running the below command. You should see something like the below output.
-```sh
-porter credentials list
-```
-
-![List Porter Credentials](./docs/img/porter-credentials-validate.png)
-
-
-## Step 4: Use Porter CNAB
-Run one of the below commands to interact with the CNAB
-
-```sh
-porter install --cred MLServicesInstaller
-porter upgrade --cred MLServicesInstaller
-porter uninstall --cred MLServicesInstaller
-```
-### Step 5: Check for pods and services
-After the installation each of the services gets installed into its own namespace, try below commands to look for pods and services:
-
-```sh
-kubectl get pods -n mlflow
-kubectl get pods -n kubeflow
-```
-### Step 6: Opening Kubeflow dashboard
-To access the dashboard using external connection, replace "type: NodePort" with "type: LoadBalancer" using the patch command:
-
-```sh
-$ kubectl patch svc/istio-ingressgateway -p '{"spec":{"type": "LoadBalancer"}}' -n istio-system
-service/istio-ingressgateway patched
-```
-Then the EXTERNAL-IP will become available from:
-
-```sh
-$ kubectl get -w -n istio-system svc/istio-ingressgateway
-NAME                   TYPE           CLUSTER-IP     EXTERNAL-IP   PORT(S)                          AGE
-istio-ingressgateway   LoadBalancer   10.0.123.210   12.34.56.78   15020:30397/TCP,80:31380/TCP,..  7m27s
-```
-![Kubeflow dashboard](./docs/img/kubeflow_dashboard1.png) 
-
-Use external-ip to open it in your browser, and make sure your firewall rules allow HTTP port 80.
-
-You can monitor Kubeflow cluster by looking at the Kubernetes status, you might need to wait to let the pods create containers and start.
-
-For more information see [Installing Kubeflow on Azure](https://www.kubeflow.org/docs/azure/deploy/install-kubeflow/) 
-
-### Step 7: Opening MLflow dashboard
-To access the dashboard using external connection, first we need to get the external-ip:
-
-```sh
-$ kubectl get svc -n mlflow
-NAME             TYPE           CLUSTER-IP    EXTERNAL-IP     PORT(S)          AGE
-mlflow-service   LoadBalancer   10.0.176.78   52.250.47.209   5000:31148/TCP   19m
-```
-Use external-ip to open it in your browser, and make sure your firewall rules allow HTTP port 5000.
-
-![MLflow dashboard](./docs/img/mlflow_dashboard.png)
-
-### Step 8: Creating a Notebook Server
-
-From the Kubeflow dashboard select "Notebook Servers". Pick the namespace you want to create the server under and select "+ New Server".
-
-Enter the desired specs for your server. Make sure the "Custom Image" checkbox is select and input `naedwebs/jupyter-mlflow` in the text field for this option. Click "Launch".
-### Step 9: Upload a Notebook
-
-Once your server is running click "Connect". A Jupyter Notebook landing page should load on a new tab. On the right hand side of this page push the "Upload" button and select the MLflow_Tutorial notebook found in the notebooks folder in this repository and hit open. Click the blue "Upload" button that has just appeard. Select the notebook to run it.
+ - [Link to Kubeflow](https://github.com/NealAnalyticsLLC/azure-intelligent-edge-patterns/blob/mlflow-on-azure-stack/Research/mlflow-on-azure-stack/porter/kubeflow/Readme.md)
+ - [Link to MLFlow](https://github.com/NealAnalyticsLLC/azure-intelligent-edge-patterns/blob/mlflow-on-azure-stack/Research/mlflow-on-azure-stack/porter/mlflow/Readme.md)

Research/mlflow-on-azure-stack/notebooks/MLflow_Tutorial.ipynb

@@ -83,7 +83,8 @@
     "  - scikit-learn=0.19.1\n",
     "  - pip\n",
     "  - pip:\n",
-    "    - mlflow"
+    "    - mlflow\n",
+    "    - pysftp"
    ]
   },
   {
@@ -364,8 +365,6 @@
     "    return rmse, mae, r2\n",
     "\n",
     "if __name__ == \"__main__\":\n",
-    "    remote_server_uri = \"http://mlflow-service.mlflow:5000\" # set to your server URI\n",
-    "    mlflow.set_tracking_uri(remote_server_uri)\n",
     "    mlflow.set_experiment(\"demo-experiment\")\n",
     "    diabetes = datasets.load_diabetes()\n",
     "    X = diabetes.data\n",
@@ -393,7 +392,6 @@
     "    print(\"  MAE: %s\" % mae)\n",
     "    print(\"  R2: %s\" % r2)\n",
     "\n",
-    "    start_run()\n",
     "    set_tag(\"author\", \"name\")\n",
     "    set_tag(\"description\", \"tutorial data based on the diabetes data set using an elastic net model\")\n",
     "    log_param(\"alpha\", alpha)\n",
@@ -406,8 +404,7 @@
     "    eps = 5e-3  # the smaller it is the longer is the path\n",
     "\n",
     "    print(\"Computing regularization path using the elastic net.\")\n",
-    "    alphas_enet, coefs_enet, _ = enet_path(X, y, eps=eps, l1_ratio=l1_ratio, fit_intercept=False)\n",
-    "    mlflow.end_run()"
+    "    alphas_enet, coefs_enet, _ = enet_path(X, y, eps=eps, l1_ratio=l1_ratio, fit_intercept=False)"
    ]
   },
   {
@@ -452,19 +449,30 @@
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,
    "metadata": {},
+   "outputs": [],
    "source": [
-    "# STOP HERE - CODE BELOW IS STILL IN DEVELOPMENT"
+    "%env MLFLOW_TRACKING_URI=http://mlflow-service.mlflow:5000"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,
    "metadata": {},
+   "outputs": [],
    "source": [
     "%%bash\n",
     "\n",
-    "mlflow run . -P alpha=0.6"
+    "mlflow run . -P alpha=0.6 --experiment-name demo-experiment"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# STOP HERE - CODE BELOW IS STILL IN DEVELOPMENT"
    ]
   },
   {

Research/mlflow-on-azure-stack/porter/Dockerfile.tmpl renamed to Research/mlflow-on-azure-stack/porter/MLflow & Kubeflow/Dockerfile.tmpl

@@ -0,0 +1,25 @@
+# MLflow and Kubeflow Installation On Kubernetes
+
+This guide demonstrates how to install a remote MLflow Tracking Server & Kubeflow on Kubernetes. The instructions below demonstrate how to install using a Cloud Native Application Bundle (CNAB). Please see the document referenced below for manual installation instructions.
+
+**Reference Material:**
+- [Manual Installation Instructions for Mlflow](./docs/manual_installation.md)
+- [Manual Installation Instructions for Kubeflow](/Research/kubeflow-on-azure-stack/Readme.md)
+
+**Prerequisite:**
+- Will need the k8 cluster ".kubeconfig" file on your local machine to execute commands on the k8 cluster 
+- Below instructions are not intended to be run from the master node, but from another Linux dev environment
+- Clone the github repo at "/home/user/" path
+
+## Step 1: Install Porter
+Make sure you have Porter installed. You can find the installation instructions for your OS at the link provided below.
+
+[Porter Installation Instructions](https://porter.sh/install/)
+
+**NOTE:** be sure to add porter to your PATH so it can find the binaries
+
+## Step 2 : Install CNAB Packages
+We have CNAB packages for Kubeflow & MLFlow, use any of the below link to install 
+
+ - [Link to Kubeflow](https://github.com/NealAnalyticsLLC/azure-intelligent-edge-patterns/blob/mlflow-on-azure-stack/Research/mlflow-on-azure-stack/porter/kubeflow/Readme.md)
+ - [Link to MLFlow](https://github.com/NealAnalyticsLLC/azure-intelligent-edge-patterns/blob/mlflow-on-azure-stack/Research/mlflow-on-azure-stack/porter/mlflow/Readme.md)

Research/mlflow-on-azure-stack/porter/MLflow & Kubeflow/Readme.md

@@ -53,7 +53,7 @@ uninstall() {
   wget ${KF_CONFIG_URI}
   ls -la
   cat $KF_CONFIG_FILENAME
-  $KF_CTL_DIR/kfctl delete -f ${KF_DIR}/${KF_CONFIG_FILENAME}
+  $KF_CTL_DIR/kfctl delete -f ${KF_DIR}/${KF_CONFIG_FILENAME} --force-deletion
 }
 
 # Call the requested function and pass the arguments as-is

Research/mlflow-on-azure-stack/porter/kubeflow.sh renamed to Research/mlflow-on-azure-stack/porter/MLflow & Kubeflow/kubeflow.sh

@@ -1,6 +1,6 @@
 # This is the configuration for Porter
 
-name: MLServicesInstaller
+name: MLServiceInstaller
 version: 1.0
 description: "Installer for ML Services on Azure Stack. Installs Kubeflow and MLflow"
 # TODO: update the registry to your own, e.g. myregistry/porter-mlflow:v1.0
@@ -13,6 +13,10 @@ mixins:
   - exec
   - kubernetes
 
+credentials:
+  - name: kubeconfig
+    path: /root/.kube/config
+
 parameters:
   - name: kf_ctl_dir
     type: string
@@ -46,10 +50,6 @@ parameters:
     type: string
     default: "https://raw.githubusercontent.com/kubeflow/manifests/v1.0-branch/kfdef/kfctl_k8s_istio.v1.0.2.yaml"
 
-credentials:
-  - name: kubeconfig
-    path: /root/.kube/config
-
 install:
   - exec:
       description: "Install mlflow"
@@ -80,8 +80,9 @@ uninstall:
       command: ./mlflow.sh
       arguments:
         - "uninstall"
+           
   - exec:
       description: "Uninstall Kubeflow"
       command: ./kubeflow.sh
       arguments:
-        - "uninstall"           
+        - "uninstall" 

Research/mlflow-on-azure-stack/porter/mlflow.sh renamed to Research/mlflow-on-azure-stack/porter/MLflow & Kubeflow/mlflow.sh

@@ -0,0 +1,19 @@
+FROM debian:stretch
+ARG BUNDLE_DIR
+RUN apt-get update && apt-get install -y ca-certificates
+RUN apt-get install -y wget
+# Use the BUNDLE_DIR build argument to copy files into the bundle
+COPY . $BUNDLE_DIR
+RUN apt-get update && \
+apt-get install -y apt-transport-https curl && \
+curl -o kubectl https://storage.googleapis.com/kubernetes-release/release/v1.15.5/bin/linux/amd64/kubectl && \
+mv kubectl /usr/local/bin && \
+chmod a+x /usr/local/bin/kubectl
+
+# exec mixin has no buildtime dependencies
+
+RUN rm -fr $BUNDLE_DIR/.cnab
+COPY .cnab /cnab
+COPY porter.yaml $BUNDLE_DIR/porter.yaml
+WORKDIR $BUNDLE_DIR
+CMD ["/cnab/app/run"]

Research/mlflow-on-azure-stack/porter/porter.yaml renamed to Research/mlflow-on-azure-stack/porter/MLflow & Kubeflow/porter.yaml

@@ -0,0 +1,6 @@
+FROM debian:stretch
+ARG BUNDLE_DIR
+RUN apt-get update && apt-get install -y ca-certificates
+RUN apt-get install -y wget
+# Use the BUNDLE_DIR build argument to copy files into the bundle
+COPY . $BUNDLE_DIR