Volunteer to be the official image maintainer - I had emacs-nox and (screen) installed fleet wide for my own utility :)
I had a friend that even had his public keys added to the /root/.ssh/ but I didin't go that far -I didn't even put my own .emacs out - but I at least could use good tools to look at the tcpdump output or giant log files if needed. "Eight Megs and Constantly Swapping" is not that big of a deal anymore.
And if people want to just use some default open source image, just point out that in modern cloud environments, you don't want each node to customize itself, you want to pre-run that process one time per node type in your "directed graph of image delta pipelines" which takes the input image and publishes the cloud ready app-specific images (with your DNS configs, LDAP integration, whatever, plus emacs/neovim and screen/tmux :)
This will definitely work, but it's not really even necessary. Just have some pre-connect script, that checks if the host is already "configured", and if not, then one-shot some Ansible playbook (or bash even) that installs what's needed. Use /tmp if root is not available. Also works for Kubernetes, though there we have better options.
Volunteer to be the official image maintainer - I had emacs-nox and (screen) installed fleet wide for my own utility :)
I had a friend that even had his public keys added to the /root/.ssh/ but I didin't go that far -I didn't even put my own .emacs out - but I at least could use good tools to look at the tcpdump output or giant log files if needed. "Eight Megs and Constantly Swapping" is not that big of a deal anymore.
And if people want to just use some default open source image, just point out that in modern cloud environments, you don't want each node to customize itself, you want to pre-run that process one time per node type in your "directed graph of image delta pipelines" which takes the input image and publishes the cloud ready app-specific images (with your DNS configs, LDAP integration, whatever, plus emacs/neovim and screen/tmux :)
This will definitely work, but it's not really even necessary. Just have some pre-connect script, that checks if the host is already "configured", and if not, then one-shot some Ansible playbook (or bash even) that installs what's needed. Use /tmp if root is not available. Also works for Kubernetes, though there we have better options.
One of my favs: https://github.com/nicolaka/netshoot