5 Mar
2009
5 Mar
'09
3:42 a.m.
Den Wednesday 04 March 2009 22.07.57 skrev alex stone:
I have a question about running linux in ram.
The Scenario.
I'm researching building a dedicated Lsampler box, running headless, with
netjack doing the ins and outs.
Given the usually large templates i run (full orchestra), including a lot
of articulations, i'm wondering how feasible it would be to build a
barebones Linux OS, with linuxsampler as the sole app, more or less, (and
required libs, etc...), and run the entire thing in ram, that is, the OS,
Lsampler, and samples.
Is this possible, and if someone's already tried this in some form or other
successfully or otherwise, could you share some experiences, pros and cons,
etc... ?
Hi Alex,
We have a home brew at work that is a Liveubuntu with some extra mayonnaise.
PXE-boot (net boot) of the kernel through a boot server (Ubuntu 8.04). All
runs in memory. Everything you install (from the boot server) on desktop OS
(Ubuntu) can be saved as a file with apt-get install set and reinstalls
automatically when you boot next time. This makes us a hardware independent
up to the point whats in our kenrel. I was not the designer of the system so
I don't have the intricate details of it but if you're interested I could ask
(not that hard. the designer is sitting 3 doors down) about them. I do
remember a caveat about booting. The initrd can't hold more than xx amount of
data and that made the designer do some trick to get around that. What I can
about the system is that you'll need at least 2G of memory to run smoothly.
/bengan