On Thu, 25 Jan 2001 11:37:27 +0100 "Aley Keprt" <[EMAIL PROTECTED]> wrote: > > > > On Thu, 25 Jan 2001 09:08:50 +0100 "Aley Keprt" <[EMAIL PROTECTED]> > wrote: > > > But why you still rely on "disk crash = all data are lost" ? > > > > This is the case with RAID0 (lost = lost from the RAID). > > > > -Frode > > I see the problem is that I don't see disk crash is equal to "disk is > completely lost". > If you have some errors on a disk drive, and you can still use the rest of > it, your RAID0 still works. > Or not?
This is not a disk crash. This is a sector error. ALL disk have sector errors. In the old days, ie. pre 1986 or thereabouts, these used to be marked as 'bad blocks' in a block map on the disk. This was done from the factory. If you wanted to keep this map up to date, you had to run some software to check this for you. Disk nowadays still have block errors. However, some percentage of the disk is now reserved for this purpose. Whenever the disk detects a "bad block" it marks it as bad and substitute it with a block from the reserved area. If there is data on such a bad block, the disk usually detects the fault early enought to salvage the data. But sometimes you loose the data in such a bad block. This happens VERY rarely in my experience, I have only heard of this, not experience it myself. I suspected this once, but I used another controller and the disk worked nicely evre since (6 years now). When I talk about a crash I mean a crash which can not be repaired with conventional means. This can be a physical crash where the only way is to send it to a profesional compnay (www.ibas.com) or it might be a crash in the filesystem. A format or a low-level format ususlly solves this. -Frode