Skip to content

tluck/OpsMgr-in-Kubernetes-and-Docker

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OpsMgr K8s Demo

In Kubernetes Summary:

TL;DR: To run the whole demo (assumes you have the GCP gcloud cli/api and other tools installed)

git clone https://github.com/tluck/OpsMgr-in-Kubernetes-and-Docker.git
cd OpsMgr-in-Kubernetes-and-Docker/k8s
cp sample_init.conf init.conf
./0_make_k8s.bash 
./_launch.bash
./_mc_launch.bash

This demo will install OM and a few MDB Clusters into a Kubernetes cluster.

  • Ops Manager v7 is the current version.
    • Application DB - aka App DB - a Cluster for OM data
    • For Backup, 2 Clusters are built: a blockstore Cluster for data and an oplog Cluster for continous backups
  • 2 Production Clusters
    • an example replica set cluster.
    • an example sharded cluster.
    • TLS Certs are created using a self-signed CA.
  • An openLDAP server is included and configured to support OM users and/or DB users.
  • There is the option to deploy a BiConnector for each cluster.
  • And Queriable backup is available too!

Step 1. Configure a K8s Cluster

  • Setup the K8s cluster
  • Install the needed command tools: kubectl, cfssl (to generate a CA).

Note: see the multiCluster section below to create 4 seperate K8s clusters which can be used for this part.

* Kubernetes - the demo is compatible with RH Openshift, Docker-Desktop, Minikube, AWS EKS, GCP K8S.
* For a small Production/Full Deployment:
	* 48-64 Cores
	* 192-256 GB Memory
	* 2000-5000 GB Disk 

* Minimum resources required for a demo: 
	* 8 Cores
	* 11GB Memory (2GB of swap)
	* 50GB Disk 

Step 2. Launch the Services

The _launch.bash script runs several "deploy" scripts for each of the following steps.

Important Before you run the _launch.bash script or any of the "deploy" scripts, copy the sample_init.conf file to init.conf and customize the parameters, such as adding your username (your email) and/or a new password.

Note: For a simple demo, you may need to change serviceType to NodePort vs LoadBalancer.

  • Script 1: deploy_Operator.bash

    • Setup the environment in the namespace defined in init.conf
    • Deploys the MongoDB Enterprise K8s Operator
    • Deploys the Cert-Manager if TLS is on (default)
  • Script 2: deploy_OM.bash

    • Setup the Ops Manager enviroment
    • Deploy the OM Resources
      • OpsManager
      • AppDB
    • Monitors the progress of OM for Readiness
  • Script 3: deploy_Cluster.bash

    • The script creates various TLS certificates, mdb users, secrets, and related configmaps. In other words, it deploys a cluster.
    • the _launch script deploys several MDB clusters.
    • The Oplog and Blockstore ReplicaSet Clusters complete the Backup setup for OM
    • The cluster "myreplicaset" is a "Production" ReplicaSet Cluster and has a splitHorizon configuration for external cluster access
      • connect via bin/connect_external.bash script
    • The cluster "mysharded" is a "Production" Sharded Cluster using either NodePort or LoadBalancer for external cluster access
    • The script monitors the progress of the "mongodb" resource to attain "ready" before it exits.
    • Note: for convenience, the k8s cluster node names are used for the split horizon external access.

Step 3. Login to Ops Manager

  • To login to OM, connect with a browser using the user/password defined in init.conf.

  • The admin user credentials and various other settings are held in init.conf

    • The scripts also create a hostname entry such as:

      127.0.0.1 opsmanager-svc.<namespace>.svc.cluster.local

      or

      34.168.131.193 opsmanager-svc.mongodb.svc.mdb.com opsmanager-svc om.mongodb.mdb.com

      into the /etc/hosts file for your convenience.

    • Note: Add the custom TLS certificate authority (certs/ca.crt) to your keystore. This will allow seamless unchallenged secure https access.

Step 4: LDAP Server (Optional)

  • Run bin/deploy_ldap.bash to create the server
    • This creates an openLDAP server and pre-configures OM for users and groups.
    • There are several DBusers: dbAdmin, User01, User02 (password is Mongodb1)
    • There are several OpsMgr Users: e.q Thomas.Luckenbach
    • There are 4 groups:
      • dn: cn=dbadmins,ou=users,dc=example,dc=org
      • dn: cn=dbusers,ou=users,dc=example,dc=org
      • dn: cn=readers,ou=users,dc=example,dc=org
      • dn: cn=managers,ou=users,dc=example,dc=org
    • The "manager" group is intended for configuring LDAP for OM users
    • The "dbusers" group is intended for configuring LDAP for DBusers
    • The "dbadmim" group is intended for configuring LDAP for DBadmins for clusterAdmin etc.

Step 5: BiConnector Server (Optional)

  • Run bin/deploy_BIC.bash -n <cluster> -p <NodePort> to create the connector server
    • This creates and configures a BiConnector server for a Cluster on NodePort N (e.g. 30307).

Step 6: MultiCluster K8s resources (Optional)

  • If you didn't already make multiple k8s clusters, run the 0_make_k8s.bash to create 4 K8s in GCP.

    Note: this script builds a central cluster and 4 member clusters and then installs Istio mesh. The central cluster also has a VPC-wide DNS server.

  • run _mc_launch.bash

    • Creates two 5-node replicaSets across 3 the K8s "member" clusters - one is externally accessible using an external domain. Note: this should be run afer you build OM resources as outlined above.

Ops Manager Demo Environment (in Docker)

Docker Image Repo: https://hub.docker.com/repository/docker/tjluckenbach/mongodb

In Docker Summary:

These scripts run/deploy several containers in Docker. They create an OpsManager instance with severval agent-ready "empty nodes" to provision a cluster - leverage the Automation, Backup, etc.

Step 1. StartUp OpsMgr (step1.bash)

  • Build or pull the image (add -b option to build a new image with step1.bash script)
  • Will create a docker network called mongonet and deploy OpsMgr and its AppDB

Step 2. Configure OpsMgr and get API key for the agent (step2.bash)

  • Add host alias to 127.0.0.1 localhost as OpsMgr in /etc/hosts
  • Login and configure OpsMgr
    • Set the server to http://opsmgr:8080
    • Use a fake mail server, smtp, port 25, and use defaults for other values
    • Click on Admin to et up Backup DB blockstore (localhost:27017)
  • In OpsMgr, to the project and create an API key
  • Add this Group key and API key to the file 6.x_node/data/automation-agent.config

Step 3. Create the containers for a replicaSet - 3 nodes (step3.bash)

  • Create some nodes - their agent will "register" with OM as available servers.
  • For 3 nodes:
    • run these commands: run 1; run 2; run 3

Step 4.