The "Provision a Nomad cluster in the cloud" works in AWS with these updates:
- use an available ubuntu version
- uses hashicorp packages where possible
- updates Nvidia installation
- installs CNI plugins
Also format terraform scripts with hclfmt, equivalent to terraform fmt.
I opted not to use terraform fmt, because I didn't want to introduce dev dependency on the terraform CLI.
Also, I've optimized the find command to ignore spurious directories (e.g. .git, node_modules) that seem to be populated with too many files! make hclfmt takes 0.3s on my mac down from 7 seconds!
The systemd configs spread across our repo were fairly out of sync. This
should get them on our best practices.
The deployment guide also had some strange things like running Nomad as
a non-root user. It would be fine for servers but completely breaks
clients. For simplicity I simply removed the non-root user references.
* update versions of all products and add consul-template
* update client and server user_data scripts
* modify README.md and terraform.tfvars
* fix typo in consul-template system unit file
* update AMI id
