last write to drive issued with write cache off?

last write to drive issued with write cache off?

[Chris Friesen]

I was looking through the official spec on the IBM 120gxp drives (looking for
more info on the 333hr/mo thing) when the following sentence jumped out at me.

"To prevent loss of customer data, it is recommended that the last write access
before power off be issued after setting the write cache off."

I was curious if we're actually doing this when someone issues a shutdown
command. Anyone know?

Just wondering,

Chris


-- 
Chris Friesen                    | MailStop: 043/33/F10  
Nortel Networks                  | work: (613) 765-0557
3500 Carling Avenue              | fax:  (613) 765-2986
Nepean, ON K2H 8E9 Canada        | email: cfriesen@nortelnetworks.com

Re: last write to drive issued with write cache off?

[Alan Cox]

> "To prevent loss of customer data, it is recommended that the last write access
> before power off be issued after setting the write cache off."

Standard 2.4 doesn't do this. 2.4.19-ac issues cache flushes.

Re: last write to drive issued with write cache off?

[Bernd Schmidt]

On Thu, 21 Mar 2002, Alan Cox wrote:

> > "To prevent loss of customer data, it is recommended that the last write access
> > before power off be issued after setting the write cache off."
> 
> Standard 2.4 doesn't do this. 2.4.19-ac issues cache flushes.

Is this what's been causing my harddrives to make funny noises on shutdown
recently?


Bernd

Re: last write to drive issued with write cache off?

[Andre Hedrick]

On Thu, 21 Mar 2002, Bernd Schmidt wrote:

> On Thu, 21 Mar 2002, Alan Cox wrote:
> 
> > > "To prevent loss of customer data, it is recommended that the last write access
> > > before power off be issued after setting the write cache off."
> > 
> > Standard 2.4 doesn't do this. 2.4.19-ac issues cache flushes.
> 
> Is this what's been causing my harddrives to make funny noises on shutdown
> recently?

Your so called funny noise is the process of flushing cache and putting
your drive to sleep in a spin down.  It only happens if you are going to
shutdown thw system.  If you are warm booting or rebooting you only get a
flush cache.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

Re: last write to drive issued with write cache off?

[Bernd Schmidt]

On Fri, 22 Mar 2002, Denis Vlasenko wrote:

> Re spindown: I don't see a reason why spindown via IDE 'sleep' command
> should sound differently from spindown due to poweroff.
> 
> So it is interesting to understand why Bernd's system started to emit
> 'funny noises'.
> 
> Maybe shutdown script does not realize that drives are spun down and cause 
> disk accesses which lead to spinup? That will be
> 
> normal -> spin down -> spin up -> poweroff and spindown
> 
> and indeed may sound funny :-)

I think the explanation may be that the drives are not spun down at the
same time, but with a one or two second delay, making it sound different
than usual.


Bernd