On 10/30/2011 05:16 PM, Charles Polisher wrote: > On Wed, Oct 26, 2011 at 06:11:02PM -0700, Eric Shubert wrote: >> On 10/26/2011 04:56 PM, Bob Hoffman wrote: >> Wrong. You're making this way more difficult than it is. >> >> Just set up your host as in the directions, and when you get to the >> point of creating a VM guest, jump to the part about setting up >> virt-manager, but set that up on a workstation/laptop. On ubuntu for >> instance, you simply install the virt-manager package. Then create your >> VM guest using virt-manager. It runs on the workstation, but the VM is >> created on the server, via a network connection. It's really pretty slick. >> >> You're right to not want to put X on your server. And you don't need to. > > It strikes me that OP is trying to do something worth doing -- > install from the (apparently broken) command line on the local > host from local media. > > When some critical infrastructure has broken, I often depend on > the ability to work with minimal dependencies. Requiring an > operational network plus a 2nd, remote host that has a running X > environment seems like too many dependencies for my taste. That > it /can/ be accomplished that way does not mean that it /should/ > be. I don't necessarily disagree, in theory. All it takes to connect remotely though is a laptop with a crossover cable. I wouldn't consider that to be too many dependencies. Pretty much just commodity (think crash cart) stuff. My servers are typically all headless, with this sort of access. Of course, if the CLI command has a bug, we should do what we can to help to get it fixed. In this case though, creating a VM using virt-manager is a good deal easier than using the CLI, so it's just easier to do it that way. > As the OP has so much time invested, the next step could be to > get strace, gdb, and the source code and start bug hunting. But > one has to pick one's battles. Agreed. -- -Eric 'shubes'