Skip to content

Latest commit

 

History

History
54 lines (47 loc) · 3.7 KB

MANUAL-METHOD.md

File metadata and controls

54 lines (47 loc) · 3.7 KB
Raw

The manual method

This way requires you to do more manual work on your end.

  • You will create deployments in each individual cloud provider; you can use my provided scripts or do it manually
  • Gather the IP addresses; via scripts or manually
  • Run setup.py to manually start the install process

Create your VMs

Choose your own adventure: do it yourself or use some scripts that are in the repo.

Go to a few of your favorite public cloud providers and create some VMs!

  • Make note of the public and private IP addresses of all of the VMs
  • Use the same public key on all of the servers and put the private key on your laptop;
    • Either use an existing one or create a new one: ssh-keygen -C "ubuntu"
  • Use the same username on all of the servers (e.g. ubuntu)
  • Make sure all of the DSE appropriate ports are open. For a list of all of the ports go to: https://docs.datastax.com/en/dse/6.0/dse-admin/datastax_enterprise/security/secFirewallPorts.html
  • If you don't want to open up the specific ports for DSE, then you can do the nuclear option and open all all of the following ports: 7000-65535; fine for demos, but not fit for production by any means

Use the scripts in the repo

  • Check out the iaas directory in this repo for a full README on how this works
  • This is complete automation for provisioning the IaaS for Azure, GCP, and AWS

Create a text file that has your different VMs that you want in the cluster:

Again, choose your own adventure...

  • If you're leveraging the scripts in the iaas folder to create your infrastructure, then you probably already put your list together via the gather_ips.sh script. If you already have your list, then skip the rest of this section. If you do not have your list yet, then go back to the iaas/README.md and use revisit the gather_ips.sh section.
  • Otherwise, if you need to put your list together manually:
    • Create a generic text file and format your VMs like this: public-ip:private-ip:dc-name:node-number for example, I call it as server-list below:
18.236.78.240:172.31.16.53:ops:0
18.236.78.240:172.31.16.53:aws:0
34.217.211.58:172.31.21.235:aws:1
34.208.176.38:172.31.17.235:aws:2
35.224.38.177:10.128.0.2:gcp:0
35.193.235.66:10.128.0.3:gcp:1
35.192.167.240:10.128.0.4:gcp:2
104.42.173.94:172.16.0.4:azure:0
104.42.168.14:172.16.0.4:azure:1
104.42.173.219:172.16.0.4:azure:2
  • Very important: decide which VM you want to be acting as your OpsCenter node:
    • Make a note of the public IP address
    • Delete that entry from your server-list file - We don't want to make your OpsCenter VM a DSE node as well

From your laptop, here is an example command to get everything setup command:

It is time to setup your cluster using the setup.py script. Here is an example: python setup.py -u ubuntu -k keys/ubuntu -n dse-cluster -s server-list

Let's break down the switches:

  • -u --> Argument needed: username that you'll use to log into all of the servers (required) (e.g. ubuntu)
  • -k --> Argument needed: file of the private key on your laptop (required) (e.g. ../keys/ubuntu)
  • -n --> Argument needed: name of the DSE cluster and cloud deployments that you'd like to create (required) (e.g. jbreese-test)
  • -s --> Argument needed: file of the list of servers that you created in the previous step (required) (e.g. server-list)
  • -p --> Flag only: phased deployment will configure all DCs but will only install the first DC. You'll still be able to see the other DCs within LCM, though. You just need -p for this flag and nothing else. To install the subsequent DCs: go into LCM, click on the cluster -> DC -> click on the ... and then install. This will kick off an install job for that DC (optional) (e.g. -p)

The script could take a few minutes to deploy so be patient.