total-cluster

A tool for running enterprise-grade application stacks powered by MongoDB running in any Kubernetes cluster.

total-cluster is designed to improve the quality of life for technology professionals, giving them a simple out-of-box cloud-native Kubernetes environment running securely with MongoDB Enterprise Data Services. Once installed, total cluster is ready to provide data services for your apps.

total-cluster should only be used for development and testing; it's ideally suited for demonstration and proof-of-concept tasks.

NOTE: total-cluster is still a work in progress. If you'd like to get involved making total-cluster better, then heck out the addons page!

Get started

To install all base components and start a MongoDB database:

helm install mongodb .
kubectl port-forward mongodb-ops-manager-0 8080:8080

TODO add info to create apikey cloud.mongodb.com

You can connect to the database with the uri found in the binding secret.

What is total-cluster?

The total-cluster project is basically 1 main Helm chart along with sets of addons, quickups, and tools.

• addons - Additional Helm charts to add more functionality to the cluster. For example, minio or the YCSB load test.
• quickups - Helper scripts to spin up clusters and similar utilities.
• tools - Various other experimental items.

The main Helm chart builds the following:

templates
├── binding.yaml                               # Secret holding user database connection string
├── cloud-manager-config.yaml                  # Cloud Mgr credentials (optional)
├── cluster-backup.yaml                        # db cluster for Ops Mgr backups
├── cluster-cluster1.yaml                      # the db for user to use
├── deployment-mail.yaml                       # simple local smtp for Ops Mgr
├── deployment-ui.yaml                         # Experimental simple Operator ui
├── operator-roles.yaml                        # Standard MDB K8S operator roles
├── operator.yaml                              # MDB operator
├── ops-manager-admin.secret.yaml              # Creds for ops mgr
├── ops-manager.yaml                           # Ops mgr crd instance
└── service-ui.yaml                            # service for experimental ui

The end result of running this chart will be:

1. A 3-node replica set for use by users and apps
2. The MongoDB k8s operator
3. MongoDB Ops Mgr or connection to Cloud Manager
4. A Secret you can use to connect your apps

This means you can starting running your apps with MongoDB in Kubernetes with the push of a button.

Setups

There are various levels of 'sophisication' you can choose for your total cluster, each level adds more locally deployed data platform components.

The minimalist setup is extra-lite.

This uses MongoDB Cloud Manager an enterprise db devops tool, and the MongoDB Kubernetes operator.

• extra-lite.values.yaml
  • MongoDB Kubernetes Operator
  • MongoDB Cloud Manager connection
  • Local 3-node MongoDB replica set

The lite setup switches to a minimalist local (MongoDB Ops Manager) deployment. This setup does not support Ops Manager backups out-of-the-box. (But you can always add it yourself, now or later.)

• lite.values.yaml
  • MongoDB Kubernetes Operator
  • MongoDB Ops Manager
    • 3-node app db
  • Local 3-node MongoDB replica set

The first almost prod-ready option is the standard package. This adds local Ops Manager backups, HA Ops Manager backing datastores. When properly configured this option can support air-gapped environments.

Note: The values.yaml file is a copy of standard.values.yaml and therefore is the default option for total-cluster.

• standard.values.yaml
  • MongoDB Kubernetes Operator
  • MongoDB Ops Manager
    • 3-node app db
  • Local 3-node MongoDB replica set
  • Node affinity for Ops Manager and dbs
  • DB pod spec overrides
  • Local Mail server

Needed next levels:

Add Atlas Service Broker & Service Catalog. Add support for locally defined "plans" to support Enterprise database-as-a-service requirements.

Setup a quick & easy local filesystem snapshot store for Ops Manager.

helm install --set accessKey=myaccesskey,secretKey=mysecretkey \
    stable/minio

Then add like http://minio:9000 and myaccesskey and mysecretkey into the Ops Mgr S3 backup config.

GKE

Very simple way to get started. Install the gcloud cli and create a Cloud Manager apikey at http://cloud.mongodb.com. Then,

git clone https://github.com/jasonmimick/total-cluster
cd total-cluster
gcloud container clusters create --zone us-central1-b total-cluster
gcloud container clusters get-credentials total-cluster
# installs to 'default' namespace; change that if you want to
helm install -f extra-lite.values.yaml \
             --set cloudManager.publicApiKey=<PUBLIC_KEY> \
             --set cloudManager.privateApiKey=<PRIVATE_KEY> \
             --set cloudManager.orgId=<CLOUD_MANAGER_ORG_ID> \
             total-cluster .

GCE

Installs an n-node k3s cluster into GCE vms (assumes you have the gcloud cli all setup.

./clusters/mongodb-k3sup-gcp.sh up

Currently, setup to demonstrate a minimal production-readly HA/DR platform. So, there a multiple dedicated Kubernetes worker nodes deployed across mutiple GCP zones.

Replicated KOTS

Replicated KOTS enables you to manage the app lifecycle of MongoDB Enterprise such as installs, upgrades, rollbacks, troubleshooting, snapshots, etc.

• Install kots as a kubectl plugin.

curl https://kots.io/install | bash

You can rerun the above curl command to upgrade kots.

• Install MongoDB Total Cluster

kubectl kots install mongodb-enterprise

• TODO: Download the license from the repo.