Fetching contributors…
Cannot retrieve contributors at this time
47 lines (40 sloc) 2.07 KB
This file documents how to install Crowbar on a basic install of
Ubuntu or CentOS without needing to create a crowbar ISO first.
* Online install mode still has plenty of bugs to work out.
* Especially where the Crowbar web UI is concerned. Doing an online
install as described by this README will result in a broken web
UI. The Crowbar CLI still works, though.
* Online install mode is primarily intended as a development aid for
now. If you want to deploy something more production oriented,
build the ISO using the usual build process.
* Your primary NIC has an IP address of
* You have at least 20 gigs of disk space and 4 gigs of ram.
* You are running either as root or as crowbar.
1: Install a basic install of Ubuntu Server or CentOS.
2: Do whatever is needed to be able to install packages from the Internet.
3: Install git, rubygems, the ruby development packages, rpm, sudo,
debootstrap, and the json gem. If you are not running as root,
give yourself passwordless sudo rights
4: Clone the Crowbar repository from, and cd into the
newly-created crowbar directory.
5: In the crowbar repository, run ./dev switch. This will check out
the barclamps we need.
6: Run ./install, and wait.
7: Profit.
Neato things:
* You don't need to have a direct connection to the Internet at all
to deploy Crowbar with this code. The install process pulls
all the packages it needs entirely over http, so all you need
is access to an http proxy that does have access to the Internet.
If you export an appropriate http_proxy before starting the
install, the install process and any nodes you bring up will wind
up using that proxy for all package fetching.
* The current code deploys its own caching proxy before installing
anything. This allows us to minimize the amout of data we have to
pull from the Internet.
* You don't need to have ISO images of the operating systems you want
to install as long as you have an active Internet connection.