Hosting other Distros on my Linux

Hey Linuxmanagers,

I've been mulling this one over, googling a lot and reading whatever I can.

I want to run a few linux distros on my host box.  My host box is one of 
the recently-without-official-support RedHat 9 systems, but I do a lot 
of packaging for RH and OL, and CL, MDK, and also UL at gunpoint.

The thing is, I don't want to have to boot up some multiboot machine if 
at all possible.  I have an idle system here that I use as a 
file-server, and it's grossly overpowered: it does absolutely nothing 
that raises its usage out of the noise.  It's also been treated with 
nexus and zalman products to make it really, really quiet.  I'd really 
like to work on a lot of automated packaging under the various distros, 
but I need these machine to have cron and sshd running, and I need the 
virtual machines to respond and build code as if they were installed by 
themselves on the metal.  I'd also rather spend time on labour than on 
the electric bill or a rack full of shiny, quiet space-heaters in 3U format.

I looked at VMWare, and even demoed the product.  If I paid a whack of 
money, it looks like I can, after a reboot, find a monitor and plug it 
into my fileserver, log in by hand, start vmware, load up 2-8 hosted 
machines and maybe, just maybe, I could then log into them or their 
cronds would run.  My demo ended some time after I lost interest in 
going that route, as I don't usually even use X when I'm doing any real 
work.

I looked at Bochs and plex86, but only from afar.  Unfortunately, I got 
the feeling that I'm the only person to have even looked at them 
recently, let alone developed them.  More importantly, I couldn't make 
head nor tails of most of the documentation.  Oh, too, one of them may 
not be open-source, or some other similarly political row went on a year 
or two ago.  Beyond that and a plea for donations, both projects look 
stagnant.

Tonight I discovered UML, or I should say re-discovered it.  It looks, 
though, like it'd be running its own special kernel inside a process, 
and not be running, say, Redhat's 200-patches kernel.  This is something 
that's important:  the machine's gotta be running the distro's kernel 
and actually think it is, be it a 2.2.19 kernel, 2.4.20, 2.6.3, 
2.5.38-mm3-ac2-aa12, whatever.  The code I build can't be missing some 
element of a structure member introduced on line 1752 of the AC 
super-patch tossed into the 2.4.20-31 kernel RPM, or it won't run on RH9 
on release; by contrast, a kernel without that patch should not assume 
that structure will be there, with the improper member and bad size.

I'd be willing to start up a machine, temporarily for the install, and 
make the DD image afterwards.  I have the spare machine, constantly 
being reinstalled with Linux, Unix, whatever for the 
problem-of-the-week, the extra hardware that's temporarily idle is not a 
problem.  I just want to move forward, here, and stop reinstalling.

Anyone else found a need for a similar set-up in their own work/spare 
time, and, if so, what solutions have been tried for a virtual operating 
environment on a free-software budget?

Thanks for your time.  I hope to summarise some really inventive stuff 
within a week, and I hope I didn't leave anything out.

  - bish
_______________________________________________
LinuxManagers mailing list - http://www.linuxmanagers.org
submissions: LinuxManagers-35TzE1X9F6582KRnZfj+bdi2O/[email protected]
subscribe/unsubscribe: http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers