Once, long ago--actually, on Mon, Oct 29, 2012 at 12:59:44PM CDT--Robert
Nichols (rnicholsnos...@comcast.net) said:
> A sector that is unreadable even after retries CANNOT be remapped until it
> is written, and any attempts to read it MUST return an I/O error until
> that remapping has occurred.
On a full failure, yes. The firmware should, however, detect sectors that
are failing and remap them on an ongoing basis. You wouldn't get
notifications about those, and they should be far more frequent than
undetected full read failures.
> If the drive were to go ahead and immediately remap that unreadable
> sector, what data would you suggest that it return when the sector
> is read? All-zeros with no indication of error is NOT acceptable.
Of course not, and just what the firmware should do with a full read
failure of a previously unsuspected bad sector, I'm sure, has been the
subject of design meetings at the various disk manufacturers. My suspicion
is that such a full failure would have to be exempt from automatic
remapping, resulting in reported failures before all of the available spare
sectors are allocated.
I would, however, expect such a condition to be either a rare
occurance--due to physical damage/shock, unexpected power failure,
etc.--making it a class of general "one-off" events, or part of an
increasing cascade of detected predictive failures resulting in automatic
remapping in the case of a failing disk.
And I suspect we've gotten much more deeply into the topic than I expect
most of the list cares about.
Cheers,
--
Dave Ihnat
dih...@dminet.com
--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org
