hibernate a single disk out of many?

[bwanaaa]

As I see it, windows desires to spin down ALL hard drives for hibernation. Too bad each hard drive cant be individually designated with a time out to sleep depending on last access. There must be an easy way to write such a utility that could run as a service?

 

[MichaelD]

Interesting. I say post this over in Operating Systems. It would be a .vbs script, I'd imagine.

My install of WinXP pro hangs about 20% of time when the HDs spin down...they just don't wake up. The fact that my HDs are on a SCSI HBA might have something to do w/it...

 

[EeyoreX]

This has been asked before. Several times. In several different guises. The answer is still the same. Windows uses an all-or-nothing approach. That (AFAIK) is because the hardware uses the same all-or-nothing ideas. The computer hardware/OS can not predict when one disk will be needed rather than another, so it either hibernates all of them or none of them. Assume you have 4 IDE drives on IDE channels 1 - 4. You always use your system disk (IDE 1) when running Windows. Windows and the hardware will not know that you want to save your current data file to the drive on IDE channel 3 and thus keep the drives "sleeping" on IDE channels 2 and 4. It just can not be done. Current hardware does not work this way.

\Dan

 

[bwanaaa]

from what i've read so far, linux can do this with HDPARM

 

[VirtualLarry]

Originally posted by: bwanaaa
As I see it, windows desires to spin down ALL hard drives for hibernation. Too bad each hard drive cant be individually designated with a time out to sleep depending on last access. There must be an easy way to write such a utility that could run as a service?

You could try installing Adaptec EZ-SCSI 5.01, it includes a power-management app that allows you to set spindown times for SCSI drives individually. After installing the proper ASPI layer, your IDE drives will show up as SCSI IDs, so this could theoretically work for them. I personally haven't tested it, however. I believe that spinning down HDs increases wear and tear on them, and is totally unnecessary for a desktop system.
(Also, Hibernate is largely broken, don't use it if you value the integrity of your filesystem data.)

 

[Mark R]

On my Win2k machine, the drives spin down individually. A single timeout for all drives is specified (e.g. 1 hour), and each drive will be spun down 1 hour from when it was last accessed.

I frequently find that if I open 'my computer' after working on a document for a while that my PC will pause for a few seconds and one or more HDDs will spin up.

 

[beatle]

Originally posted by: Mark R
On my Win2k machine, the drives spin down individually. A single timeout for all drives is specified (e.g. 1 hour), and each drive will be spun down 1 hour from when it was last accessed.

I frequently find that if I open 'my computer' after working on a document for a while that my PC will pause for a few seconds and one or more HDDs will spin up.

Same here. In my backup server, I have no need for all of the drives to be spinning, so I let Windows power management spin them down after 20 minutes. The only drive that keeps spinning is the system drive which isn't very loud anyway.

 

[corkyg]

You could also use a hardware approach. You can have power switches on all drives by using mobile racks - or - make them all external except for your boot/system drive. Then, with either Firewire or USB 2, you can simply pull the plugs on the ones you don't need.

I deliberately avoid hibernation - it is a buggy program that sometimes forgets to wake up when you want it to.

 

[VirtualLarry]

Originally posted by: corky-g
You could also use a hardware approach. You can have power switches on all drives by using mobile racks - or - make them all external except for your boot/system drive. Then, with either Firewire or USB 2, you can simply pull the plugs on the ones you don't need.

If you do that, make sure that each drive is the only drive on each IDE channel. In fact, with certain IDE controllers (ones that electrically share some of the same signal lines for both channels, this may still not be enough) - those I would not recommend going this route. But just make sure that you don't have two (one Master, one Slave) HD connected to a single IDE channel when you power one of them off. The onther may get confused, and the IDE channel may hang, unless the OS can detect that and issue an IDE controller hard-reset to that IDE channel. I've personally found that Win98se seems to deal with hung IDE devices much better than W2K SP2 does, and recovers much faster (on the order of 2-3 minutes, not 10-20).