Improving halt package interaction...

[Jesse Keating]

On Wed, 14 Nov 2007 09:31:29 -0300
Thomas M Steenholdt <tmus at tmus.dk> wrote:

> Other than what the other guys have mentioned, like perhaps making it 
> possible to run scripts after ro remounting "/", a delayed UPS
> poweroff from a normal initscript is less precise, since you have no
> reliable measure of how long the system will take to power down
> (perhaps you're running large databases?). You could end up powering
> off the UPS outlets before the system is ready for it. Or specifying
> a long delay, the system might need to wait for far too long for the
> system to restart, even though power returned during the shutdown. It
> would be much better to call the UPS poweroff command as the very
> last thing with a 5-10 second delay, just before the system halts,
> rather than having to specify a 5 minute delay for an initscript
> based solution. It's more of a "hook" based approach, which IMO is
> better for this.

Wouldn't you just make sure that the UPS shut down script is
scheduled /after/ the database shut down, and all other critical
services?  IE make it close to last in the shutdown schedule.

-- 
Jesse Keating
Fedora -- All my bits are free, are yours?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-devel-list/attachments/20071114/47ec2694/attachment.sig>