Hi Andrew, > After spending days(!) fscking and trying to decode all kinds of stuff > I stepped back and saw the light - I MUST have a duff disk!!!!.
`smartctl -a /dev/sda' can be useful to get the drive's own stats on how things are going. (Does fsck(8) still only check a filesystem's metadata?) smartctl(8) can, I think, be used to get the drive to do some non-destructive tests on all sectors. May be something to try now you've nothing to risk losing from them, although I don't know how well the SMART commands work through non-[PS]ATA interfaces. > I therefore went out and replaced BOTH disks (which were Hitachi > Desktar 200GB IDE units which on extracting them I find are dated 2005 > (so fair enough!). Post IBM's Deathstar then. http://en.wikipedia.org/wiki/IBM_Deathstar ;-) > "The "green" credentials of the new disks will mean they will power > down after (20? secs) of inactivity and take too long to power up > causing linux Raid to fail a disk and detach it." !!! hdparm(8) has options, e.g -S (capital), to control aspects like idle spin-down time; perhaps that can help. I didn't realise any modern drives took too long to spin up though. Cheers, Ralph. -- Next meeting: Blandford Forum, Wednesday 2011-03-02 20:00 Meets, Mailing list, IRC, LinkedIn, ... http://dorset.lug.org.uk/ How to Report Bugs Effectively: http://goo.gl/4Xue