Subject: Re: raidframe parity always dirty after reboot
To: Matthias Buelow <mkb@mukappabeta.de>
From: Greg A. Woods <woods@weird.com>
List: netbsd-users
Date: 04/27/2003 22:54:21
[ On Monday, April 28, 2003 at 02:37:52 (+0200), Matthias Buelow wrote: ]
> Subject: Re: raidframe parity always dirty after reboot
>
> well.. I'm not that convinced that swap "needs" that; I've always
> thought of shutdown rather being a wrapper to inform users about
> a timed shutdown than being necessary to keep things clean.

No, "shutdown" really is the proper way to shut down any system.

The "halt" and "reboot" commands are just very low-level tools used by
"shutdown" to do the very last step of its overall job.  A systems
administrator should never unthinkingly use "halt" or "reboot".  Even if
the system is already in single user mode great care must be taken to
ensure the system really is 100% ready to actually halt or reboot.

In my mind the job of informing users is just a very minor side effect
of what "shutdown" really does.

> IMHO it's a workaround; the kernel should properly unconfigure
> swap partitions/-files by itself before unconfiguring RAIDs.

I most certainly do not think so.  There are many many other tasks that
need far more than just the "kill -15; sleep 10; kill -9" sequence that
you may be lucky to get "halt" or "reboot" via their interaction with
"init" -- the proper use of "shutdown" is criticlal to the care and
feeding of systems, especially production systems providing services,
even if there normally are no direct user logins on them.

Indeed I would not be bothered at all if "halt" became "halt -q" by
default and all the cleanup functionality moved out of "init" (thus
making this most critical program even smaller and simpler and more
elegant).

-- 
								Greg A. Woods

+1 416 218-0098;            <g.a.woods@ieee.org>;           <woods@robohack.ca>
Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>