|
Home > Archive > Data Storage > October 2004 > "Time Limited Error Recovery"?
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
"Time Limited Error Recovery"?
|
|
| Thor Lancelot Simon 2004-10-16, 2:45 am |
| Can someone tell me what this buzzword^H^H^H^H^H^H^H^Hfeature of the
Western Digital "RAID Edition" disk drives actually is?
--
Thor Lancelot Simon tls@rek.tjls.com
But as he knew no bad language, he had called him all the names of common
objects that he could think of, and had screamed: "You lamp! You towel! You
plate!" and so on. --Sigmund Freud
| |
| Tim Shoppa 2004-10-18, 5:45 pm |
| tls@panix.com (Thor Lancelot Simon) wrote in message news:<cknl4c$gdm$1@panix5.panix.com>...
> Can someone tell me what this buzzword^H^H^H^H^H^H^H^Hfeature of the
> Western Digital "RAID Edition" disk drives actually is?
"Traditional" drive firmware can take very long when reading a bad
block before they report that the block was unreadable (time spent
doing retries, recalibrations, etc.). When you
don't have RAID, this is probably the right thing. But when you have
RAID, it's almost always better to quickly mark the block as failed
and use the RAID redundancy to get to the real data.
There's some kind of out-of-band handshake between the RAID controller
and the drive that enables this. With SCSI drives it's mode page settings,
with IDE drives I don't know the details.
SCSI drives intended for RAID applications had spindle sync signals, maybe
"RAID Edition" IDE drives can do this through some channel?
Tim.
| |
| Malcolm Weir 2004-10-19, 2:45 am |
| On 18 Oct 2004 06:47:54 -0700, shoppa@trailing-edge.com (Tim Shoppa)
wrote:
>"Traditional" drive firmware can take very long when reading a bad
>block before they report that the block was unreadable (time spent
>doing retries, recalibrations, etc.). When you
>don't have RAID, this is probably the right thing. But when you have
>RAID, it's almost always better to quickly mark the block as failed
>and use the RAID redundancy to get to the real data.
.... and then write that back to the problem drive, triggering
remapping. If the write succeeds, no need to degrade the array.
(Not that that's directly relevant to the question).
Some drives I worked with would do 15 "declared" retries, and one
hidden one before admitting to any kind of issue (so 16 in total, but
if the first one succeeded, the drive wouldn't admit that it had
needed a retry at all.
[ Snip ]
>Tim.
Malc.
|
|
|
|
|