On Fri, Jul 11, 2008 at 6:19 PM, David Mathog <mathog at caltech.edu>
wrote:
> This is on a CentOS 5 system.
You may have better luck asking in a RedHat or CentOS forum.
Startup/shutdown scripts are highly distribution-specific.
> So far I have found
> three possibilities, but I do not know which one (if any) will still be
> valid that late in the halt procedure:
>
> 1. /dev/md0 (from posts on the net)
> 2. /dev/mapper/VolGroup00-LogVol00 (from /etc/mtab)
> 3. /dev/VolGroup00/LogVol00 (from /etc/fstab)
>
> Also, this system has SWAP on an LVM as well. Will that still be
> mounted right before "halt" tries to remount?
Look for 'swapoff -a' calls in your shutdown scripts.
Also (other NUT users/devs, please correct me if you see something
that looks wrong), upon re-reading the shutdown.txt instructions, I
think that may be overkill if your BIOS is configured to power on the
computer on every time power is applied (not just staying in the last
commanded state), *and* your UPS will cycle the outlets off and on if
power returns after shutdown has been commanded (after step 4 below).
Consider the following scenario:
1) Power fails
2) UPS driver sets LB flag
3) upsmon tells UPS to shutdown after delay
4) upsmon shuts computer down with 'poweroff'
5) UPS turns off outlets
6) Power returns
7) UPS turns on outlets, possibly after charging delay
8) Computer sees power, boots.
Note that many desktops will not turn on automatically in step 8,
since the 'poweroff' command tries to use ACPI to cut power, and most
BIOSes will stay in the last commanded state when power is reapplied
(on if the power failed during use, or off if the poweroff command was
used).
The advantage to this method is that you are relying on the kernel to
shut down the system as usual. The drawbacks, of course, are the
possible race conditions if power comes back early, and your UPS is
not smart enough to detect that power has returned just after a
shutdown has been requested.
--
- Charles Lepple