[Nottingham] Big biiig problem regarding IDE-RAID

[Robert Davies]

On Wednesday 16 Jul 2003 11:00, Lee wrote:

> What I can't seem to find out is how IDE RAID handles bad sectors, okay,
> I can keep going in the event of a total drive failure, but what happens
> if one drive develops a clump of bad sectors, I know SCSI raid does hot
> sector sparing, and relocates bad blocks with an allocation map of
> somekind (yeah my knowlage is sketchy here) , but IDE??? most IDE drives
> I've seen with bad sectors just wig out, and fail causing the hosts o/s
> to wig out too....

For info on IDE drives see http://www.storagereview.com/, it goes into things 
like sector remapping and LBA.  Basically as you state with modern drives you 
don't see bad sectors, until the drive is bad and there's no sectors left to 
remap.  Then it's time to run Disk manufacturers fault diagnosis and repair 
utilities, which amazingly often fix the apparently borked drive. 

One of the aims of the Block layer redesign in 2.5 was improved error 
reporting and handling, http://lwn.net/2001/features/KernelSummit/ has a 
summary, OSDN still has mp3's of some of the talks.  It's clear that the 
Linux error handling needed clearing up, and upper driver layers tend to only 
know an error occured, not it's type and if some action could be taken for 
recovery.

You can also choose drive enclosure with mind to maintenance, hot plug and 
enough capacity to allow hot spares to standby.  These may internally use 
multiple IDE controllers, but provide one or more SCSI interfaces.

Rob