Subject: Re: scsi hd sleep
To: Adam Fritzler <>
From: Simon Burge <>
List: port-pmax
Date: 01/30/1998 10:48:57
On Thu, 29 Jan 1998 14:57:05 -0700  Adam Fritzler wrote:

> At 03:47 PM 1/29/98 -0200, you wrote:
> >
> >           What kind of drive are you using ? Isn't it whit spin up
> >jumper setted to start at power up ? 
> >   Check It.
> I believe he meant power it down after it had powered up.  Once it gets a
> START UNIT, it stays STARTed.  He (and I) need a way to give a STOP UNIT
> after a certain length of idle time and automatically make sure the device
> is ready when trying to read (ie, send a START UNIT before reading).  This
> would delay the initial access, but reduce drive wear from idle use.  I'm
> guessing this would take a kernel modification, much like the APM BIOS stuff
> for the x86s (except without the BIOS).

Only guessing here, but wouldn't the prime motivation for this type of
thing be saving power (not getting sleep :)?  I would have thought that
constant spinning up and down would cause a degradation of drive life...

Some things that come to mind (but haven't checked too closely) would
be jobs that cron starts (atrun, etc) and update (does the ffs/ufs
superblock get written out with _every_ sync?).