<br><br><div class="gmail_quote">On Fri, Sep 24, 2010 at 10:39 PM, Nick Rout <span dir="ltr"><<a href="mailto:nick.rout@gmail.com">nick.rout@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br><div class="gmail_quote"><div><br>surely in a netboot+nfs scenario you don't need a full filesystem for each client? I've not set one up in a while, but that certainly wasn't my recollection.<br></div></div>
</blockquote></div><br><br>It depends on how you do it. I'm using the LTSP based stuff from Mythbuntu. I gather that the GUI in control-centre doesn't work anymore, but google found me some docs on it. It creates a compressed filesystem image that it shares out over nbd, then uses an NFS mount and AUFS I believe to handle the custom stuff for each machine. The main root is mounted RO, so you can only change it on the boot host. Just update the files using chroot, then run ltsp-update-image and it compresses the new filesystem. Then just reboot the FEs. So I run an "aptitude update; aptitude upgrade" on the backend, then on the chroot FS for the FEs, then do the ltsp-update-client. Pretty simple, keeps me from having to update all the boxes individually. <br>
<br>The downside is that the compressed image takes a little while to build. So if you're testing changes it can be tedious. It is multi-threaded though, so if you have a big CPU in the server and a decent I/O rate for the data, it's not too bad. <br>