martin f krafft wrote at 2009-03-09_15:32 -0600: > also sprach green <greenfreedom10@gmail.com> [2009.03.09.1716 +0100]: > > So you are saying that it is a BAD THING for RAID1 array devices to spin down? > > You are saying that I SHOULD NOT expect to be able to use power management with > > a RAID1 array? I understand that it would be unreasonable to spin down drives > > in a server but this is more of a desktop system with RAID1 providing for the > > case of a hard drive failure. > > I understand your need and agree that it would make sense in this > situation. I also use RAID1 at home for the same reason. I just > don't think mdadm is the right tool. Oh, well I can easily agree with that (mdadm is not the right tool). > Have you tried hdparm? Spinning down the drives is not my problem. Or rather I don't think it is. I have set the spindown timeout in hdparm.conf and just need to find out what processes are accessing the filesystems such that the drives don't spin down after that timeout (I use noatime, except for /home which is relatime). Surely "hdparm -C" wouldn't spin up a drive? I am using that to check periodically and the logging that to /tmp (tmpfs). > It would be nice to have harddrives spin down and up as needed, even > with RAID1, and theoretically, there is no reason that couldn't > happen. I just don't want to start thinking about the problems that > come from powering down and RAID1 interacting. Well, I just assume that Linux can handle it. The mdadm 'layer' should be operating above that of the block device anyway, right? And I haven't had any trouble during those times when hdparm said an array device was in standby mode. > I retract my previous statement. :_) mdadm is not the right tool, so don't bother retracting anything! ;) (Unless I missed something else.)
Attachment:
signature.asc
Description: Digital signature