Re: read and write back full disk to trigger relocation

> ----------------------------------------
> From: David Christensen <dpchrist@holgerdanske.com>
> Date: May 29, 2023, 12:04:41 AM
> To: <questions@freebsd.org>
> Subject: Re: read and write back full disk to trigger relocation
> 
> 
> >  From the dd(1) manpage:
> > 
> >       Do a refresh of a disk drive, in order to prevent presently recoverable
> >       read errors from progressing into unrecoverable read errors:
> > 
> >             dd if=/dev/ada0 of=/dev/ada0 bs=1m
> 
> 
> gmirror(8) examples indicate the device node should be:
> 
> /dev/mirror/name
> 
> 
> Where "name" is the argument provided when the mirror was created:
> 
> # gmirror label name prov ...

One of the examples is:

           gmirror label -v -b split -s 2048 data da0 da1 da2

Which tells me to look for /dev/da0, /dev/da1, /dev/da2 for the dd command.

>>
> > That won't trigger relocation, but it will refresh the decaying data.
> 
> 
> Do you have a URL for an engineering document or standard that describes 
> drive sector reallocation?  STFW "drive sector reallocation" I do not 
> see anything that looks authoritative (e.g. disk manufacturer or 
> standards body).
> 

None that I'm aware of, but perhaps the ZFS source code might be illuminating.
Whatever standard they used might lead to documentation for UFS.


-- 
Sent with https://mailfence.com  
Secure and private email