Page Comparison

Introduction

This guide has as a purpose to guide the user through the steps necessary to deploy ModelOp Center from end to end. It will involve creating resources in Kubernetes, ensuring that specific utilities and docker images are available to support the installation, as well as walking through and understanding the configuration items in the Helm Charts that are provided by ModelOp. Note that access to any ModelOp resources, e.g. Docker images, is dependent on having been provided access after having been onboarded as a ModelOp customer.

As part of this walkthrough, it is important to note that the items chosen for deployment via Helm, can be added or removed as seen fit, given the chosen configuration of the environment. For instance, Spring Cloud Configuration Server, which is the backing configuration mechanism for ModelOp Center. If You can find additional information regarding advanced configurations in the "Additional Configuration" section of this documentation.

Pre-Requisites

Before beginning, determining which environment ModelOp Center will be deployed into will be important. If this installation is occurring locally, or in a development environment, ModelOp suggests making use of Docker Desktop with Kubernetes enabled if possible to test deploying with Helm prior to targeting a full-fledged Kubernetes cluster, please see the Docker Desktop installation and documentation here: https://www.docker.com/products/docker-desktop/

There are a few key items that will need to be available prior to deploying ModelOp Center. Firstly, a Kubernetes cluster that has access to either DockerHub via the web, or an image repository that has the ModelOp Center images stored. Second, a workstation or jumpbox/server that has a properly credentialled and configured kubectl cli utility which allows for interaction with the Kubernetes cluster. Thirdly, the Helm utility, version 3+, will be necessary for deployment. Kubectl will require a properly configured context will be necessary to connect to the Kubernetes cluster, which can be done using the kubectl utility itself. Lastly, the ModelOp Center images will be necessary for the deployment. The images can either be accessed directly from DockerHub, as well as from a repository of your choosing, which may be inside your network as well.

To validate that the above is ready, please complete the following:

1. Ensure that a Kubernetes cluster has been allocated and is running, available, and usable with credentials that have adequate access. This will be likely managed by enterprise IT, whereby either an existing cluster or a new one is made available to use. Please ensure that this step is completed, though the following steps can be accomplished in parallel to having the Kubernetes cluster created/allocated. The link to installing Kubernetes has been provided below, though as mentioned, a specific enterprise team will likely create the cluster/infrastructure.

2. Ensure that the utility kubectl is installed and functional on the workstation or chosen jumpbox/server. Installation of the utility can be accomplished by following the guide here: https://kubernetes.io/docs/tasks/tools/

3. Ensure that a context is properly set so that connectivity can be established with your cluster following this guide: https://kubernetes.io/docs/tasks/access-application-cluster/configure-access-multiple-clusters/

4. To ensure that kubectl is working correctly, the output of the following command should result in the displaying of the namespaces for the context in which you are operating: kubectl get namespaces Note that if this command is not generating output correctly, please invoke the assistance of your enterprise support teams to assist in the installation.

5. Ensure that Helm is installed and operational by following the guide found here: https://helm.sh/docs/intro/install/

6. To test that Helm is ready and available, please run: helm version which should generate output similar to: version.BuildInfo{Version:"v3.9.0", GitCommit:"7ceeda6c585217a19a1131663d8cd1f7d641b2a7", GitTreeState:"clean", GoVersion:"go1.18.2"}

7. Ensure that the version of Helm that is currently being used is 3.0+. e.g. Version:"v3.9.0"

Useful Links

Kubernetes Installation Documentation:
https://kubernetes.io/docs/setup/
Kubectl Installation Documentation:
https://kubernetes.io/docs/tasks/tools/
Creating a kube-config Context using kubectl:
https://kubernetes.io/docs/tasks/access-application-cluster/configure-access-multiple-clusters/
Helm Installation Documentation:
https://helm.sh/docs/intro/install/
ModelOp DockerHub Repository:
https://hub.docker.com/u/modelop

Getting the Helm Charts

There are two methods for fetching the Helm Charts from ModelOp, either by requesting a tar.gz compressed file of the content of the Git repository where the Helm Charts reside from your ModelOp Account Executive or Sales Engineer, which will be delivered via bucket link.

Bring the ModelOp Center Helm Charts onto the workstation or the operating system with both the kubectl and helm utilities installed, as per the previous section.

From the provided tar.gz file: tar -xvf helm-installer.tar.gz

Despite which method of retrieving the Helm Charts is chosen, the directory helm-installer will be created with a number of files and folders inside in the directory where the tar command is executed.

Configure Helm Charts

Before deploying an instance of ModelOp Center, choices need to be made in specific configuration files related to which services are deployed, which affects how ModelOp utilizes enterprise resources in a variety of fashions. Prior to understanding which files to modify, it is important to understand the directory structure of the Helm Charts, please read through the following items.

The "installer" directory contains all of the necessary files for installing using Helm.

Within the "installer" directory, the installer/generatePullSecrets.sh script is used to generate secrets within Kubernetes in order to allow the cluster to import images.

The installer/3.0-sccs-compat.yaml file contains certain default settings for a ModelOp Center v.3.0 installation which include SCCS (Spring Cloud Configuration Server) which is used as the backing configuration service for ModelOp services.

The installer/moc directory is a default configuration ModelOp Center deployment, using database pods to support the core application of ModelOp Center. It is suggested that a copy of this directory be made for every release that would need to be deployed. e.g. cp -R installer/moc installer/example-release

This directory can be referred to as an "environment", and each environment directory can be used to deploy any number of Helm "releases", which is the way Helm manages deployed applications to Kubernetes.

Within the copied directory, there are a number of files that are created, but at the root of the environment directory there are two primary files:

values.yaml contains the bulk majority of the configurations that will need to be modified and made specific to the environment they are targeting. Please find information related to the values file here: https://helm.sh/docs/chart_best_practices/values/

Chart.yaml contains details about what will describe the release once installed, it is a place to track details on what is deployed.

The my-config directory contains all of the base yaml files that will be used by Spring Cloud Configuration Server to configure the ModelOp Center application.

Modifying the values.yaml file is necessary to tailor the installation to the targeted environment. The main section of values.yaml that should be modified is moc, which contains a number of key/values that need to be set to adjust the deployment to specific needs. Please refer both to the following documentation, with general guidelines on the sections, as well as the in-line comments that are part of the file itself.

Please refer to the document Configuring values.yaml before proceeding to the section "Complete the Installation"

Pre-Flight Check with MOC Watcher

**Please refer to this documentation (to be scraped and published): https://modelop.atlassian.net/wiki/spaces/~906983076/pages/2326790145/MOC+Watcher+-+Help**

Complete the Installation

Please only begin this section after having tailored the values.yaml file to the specific deployment needs.

In order to successfully pull the images intended for the ModelOp Center deployment the generatePullSecrets.sh script performs the addition of credentials into a Kubernetes secret which is leveraged to pull the images down from either DockerHub or an internal Docker image registry. Please refer to the documentation concerning this script in the "Additional Configuration" section. Note that credentials to access DockerHub to pull ModelOp images would be provided by ModelOp, whereas credentials for in-house registries would likely be provided by enterprise IT. To execute, perform the following:

From a shell terminal (bash/zsh): ./generatePullSecrets.sh --apply

Output will be similar to the following:

$ ./generatePullSecrets.sh --apply
Enter username:
modelopdevops
Enter password:

secret/modelop-regcreds created

Then trigger the installation by running the following command from the installer directory inside the helm-installer directory that was provided:

helm install RELEASENAME DIRECTORY --set moc.version=DESIREDVERSION --namespace TARGETNAMESPACE -f 3.0-sccs-compat.yaml

Whereby:

RELEASENAME = The name given to the release, tracked by Helm, which can be anything relevant to the deployment itself. e.g. "moc-dev-helm1" Please see the following documentation that describes the concept of a Helm Release: https://helm.sh/docs/intro/using_helm/#three-big-concepts

DIRECTORY = The directory in which the targeted Chart.yaml resides, no slashes necessary. e.g. "moc-dev"

DESIREDVERSION = The version of ModelOp Center desired for the Helm deployment. e.g. "3.0"

TARGETNAMESPACE = The namespace into which you would like to deploy ModelOp Center within your current context in Kubernetes.

Example command structure:

helm install moc-test-rel1 moc-test --set moc.version=3.0 --namespace moc-test-1 -f 3.0-sccs-compat.yaml

This example command assumes the existence of the directory %helm-installer/installer/moc-test which is a copy of the %helm-installer/installer/moc directory with all values changed to what is desired for the environment.

Output will resemble this example:

NAME: release
LAST DEPLOYED: Tue Jun  7 08:28:01 2022
NAMESPACE: namespace
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
.__  __           _      _  _____           ____           _
|  \/  | ___   __| | ___| |/ ____\ _ __    / ___|___ _ __ | |_ ___ _ __
| |\/| |/ _ \ / _` |/ _ \ | /    \| '_ \  | |   / _ \ '_ \| __/ _ \ '__|
| |  | | (_) | (_| |  __/ ||      | |_) | | |__|  __/ | | | ||  __/ |
|_|  |_|\___/ \__,_|\___|_|\\____/| .__/   \____\___|_| |_|\__\___|_|
                                  |_| Version3.0

Install complete!

The following services have been installed:
  - sccs
  - registry
  - gateway
  - reporting-service
  - model-manager
  - mlc-manager
  - runtimes
  - document-service
  - minio

App name: modelop-center
Namespace: namespace
Profile: noauth

At this point, the instance of ModelOp Center that has been configured, has been deployed, and should be accessible directly by simply accessing the ModelOp Gateway pod's IP address, on port 8090, the default configurable installation port.

To fetch the IP of the ModelOp Gateway pod, simply ensure you are in the targeted namespace chosen for deployment, and use the following command:

`$ k get services | grep gateway`

Output should resemble the following as an example:

$ k get services | grep gateway
gateway               LoadBalancer   10.100.8.109     abb3888bb91fc49d5a101f1a1340d5b8-116610426.us-east-2.elb.amazonaws.com   8090:30285/TCP   70d

Note the IP address and the port that is allocated by default. To connect to the newly instantiated deployment of ModelOp Center, access the landing page via:

<http://<IP> Address>:<Port>

Example using the above output:

<http://10.100.8.109:8090>

The landing page of ModelOp Center, the Dashboard, should be presented after loading. Congratulations, you have just deployed an instance of ModelOp Center! To ensure that all services are visible to the application registry, please consult the "Status" link in the bottom lefthand corner of the landing page of the ModelOp Gateway.