Nebula on Packet global deployment example
The docker-compose tutorial is great for getting a feeling and doing local development on the Nebula API but as a repeated request has been to have a simple way to confirm Nebula global scalability a terrform module has been created which will do the following:
- Install a MongoDB & Nebula manager on a single packet devices
- Create multiple devices (by default 5) over multiple packet facilities (a list of 5 different facilities by default)
- Create and configure an
exampleapp & device_group on Nebula
Running instructions
- clone the repo
- cd into the repo folder
- Change the run.tf (in repo root folder) to your variables
- run
terraform apply.
You now have a nebula cluster that includes a single manager and as many workers you gave it (default 9) spread out across packet regions (default 9 regions).
Requires:
- terraform 0.12.0 or higher
- PACKET_AUTH_TOKEN envvar set to your packet token
- a packet organization
- curl installed on the machine running terraform
Configuration variables :
- server_size - the size of the Nebula manager server - defaults to
c1.small.x86 - server_region_device - the region where the nebula manager server will reside - defaults to
ams1 - worker_count - the number of Nebula workers to create - defaults to
5 - worker_region_device - the regions where the workers will be deployed on - defaults to
["ams1","dfw2","ewr1","nrt1","sjc1"] - worker_size - the size of the servers used in Nebula - note that at least 1GB is needed as it uses a RancherOS for simplicity sake - defaults to
c1.small.x86 - PACKET_AUTH_TOKEN - the token used to connect to your packet account - needs to be declared as an envvar - required
Should you want to ssh to the devices for any reason (not needed for the system to work) you can do so with the private key & the root user for the manager and rancher user for the workers.