I'm looking for recommendations on how to deal with the possibility of
filesystem corruption.
As in the subject, I'm on 2.4.22, with SATA drives.
From what I've read, the safest thing to do is to turn off the write cache and
use a journalled filesystem. However the hardware guy says that turning off the
write cache also turns off the automatic error correction on writes, so the
write may return an error rather than being remapped silently.
What's the best way for me to deal with this?
Chris
On Maw, 2004-09-21 at 16:27, Chris Friesen wrote:
> use a journalled filesystem. However the hardware guy says that turning off the
> write cache also turns off the automatic error correction on writes, so the
> write may return an error rather than being remapped silently.
I don't know how vendors handle this however he may be right, or it
may be vendor dependant. Have you asked which vendors do this and which
don't. Maybe you just need the right drive vendor.
The Linux IDE layer will retry a failed write not sure about the 2.4.22
SATA code although the early versions probably were not business ready
anyway.
> What's the best way for me to deal with this?
For IDE you either
1. turn off write caching
2. buy a hardware raid card (which will turn off write caching or
have battery backup or both 8))
3. accept that its unlikely to get burned in normal usage especially
with a UPS.
Alan