> Yes that's a really good idea - using a bind mount from another > filesystem in the LVM2 ? > something like: > mount --bind /etc /origetc > mount --bind /newetc /etc > Where /origetc is simply a mountpoint in the root filesystem (to be able > to maintain /etc after the next mount) > and /newetc is an LV within LVM2.
Pretty much. The base /etc gets one /etc/init.d/boot.d/ script entry that remounts the "other" /etc over the top of the one in the root, and startup proceeds from there using the "other" /etc. If that fails, then you go directly to single-user mode with no network enabled to let you go fix whatever it's whinging about. Our setup also runs a small script to use CP MSG to scream to a PROP machine to let some human know about the problem. Bill Scully at CA did a nice paper/presentation on the basic idea, which I've tinkered with a bit more to make it use more VM facilities. I actually don't mount the base /etc; I want that centrally maintained and as R/O as I can make it. That shouldn't change often, if ever -- probably once a major release when I build a new template machine. Much easier to deal with overall, and scales nicely without having all that / in LVM mess involved. I suspect one could get really clever and put the initial / in a DCSS, which would promote even more sharing -- it's basically configuring a primarily diskless workstation with some local disks at that point. Hmm. I feel a white paper or a presentation coming on. -- db ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390