I have about a dozen rather beefy machines on a separate network. This network is not (will not be, cannot be) connected to the internet. I’d like to deploy Openstack on them so that they can be used as a sort of internal cloud hosting solution. This will be a sort of “prototyping environment”, so while I can tolerate things being a bit rough around the edges, the main aim is still getting work done with the machines, not merely messing around with Openstack.
Here’s the options I’ve come up so far:
- Run a local mirror of the relevant repositories and install a script-based installation solution such as Alamo 3+. I’m not entirely sure how to find out what all the relevant repositories are though, and I’m not looking forward to finding out using trial-and-error.
- Use crowbar to build a stand-alone ISO, or use one of the provided ones. I’m not completely clear on whether these are supposed to work without Internet access though.
- Use the Alamo 2.0 ISO, which uses the older Exeter version of Openstack.
I’m sure I’ve missed about a dozen other options. I’d like to stay away from commercial vendors if possible.
What is the best way to achieve this, in terms of effort, functionality and future-proofness?
Consider creating a local private mirror of your favorite Openstack-containing distribution, such as CentOS, Fedora, Ubuntu, etc. Once you have set up the local mirror (on a networked machine) using the distribution’s instructions, walk it over the air gap on an external hard drive to the disconnected network for use there.
To obtain updates, you can simply update the mirror on the network, and then walk it over the air gap again.
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.