-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
The Monday 2007-05-07 at 13:55 +0100, David Bolt wrote:
> I'd use something like this:
>
> par2 c -s1024000 -c235 -l <basename.for.par2.archives> *
>
> c is to create the recovery files
> -s1024000 gives a recovery block size of a little under 1MB
> -c235 says to create 235 recovery blocks
> -l limits the size of the par2 recovery files to just a bit bigger than
> the largest file.
>
> That should create a few recovery files which, with the par2 overheads,
> occupy about 235MB and leave around 15MB free.
I had already done one run using simply the default options (ie, none,
meaning 5% redundancy, except for a -500). Even though I compiled with
"-O3 -march=pentium4" it is terribly slow, almost one hour.
Right now I'm running it like (275M free):
time nice par2 c -m500 -s1024000 -c250 -l recovery *avi
....
Block size: 1024000
Source file count: 11
Source block count: 4305
Recovery block count: 250
Recovery file count: 8
...
Memory usage is currently:
NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
10 250m 247m 664 R 53.3 24.4 2:26.62 par2
> Once it's finished, and
> if you're the sort of person that just has to be sure, you can verify
> the freshly created files using:
>
> par2 v <basename.for.par2.archives>
Yep, I did that on the dvd after burning it. It took 15 minutes.
>
> And, if there's a failure after the contents has been burnt, copy the
> contents off the DVD using either dd or ddrescue, and then use:
>
> par2 r <basename.for.par2.archives>
>
>
> Now, the bad news is that for a dozen files, totalling a bit over 4GB,
> you may not be able to rebuild a broken file with only 235 blocks
> without rescuing as much data as possible from the DVD. My guess is that
> the files are around 350MB[0], which means you'd need at least 350-360
> recovery blocks to rebuild a completely missing file. As long as only
> one file is broken, and you manage to recover more than a third of the
> data, there _should_ be enough to rebuild it.
Well, I assume I would be able to recover a damage of less than 5%, ie,
about 210MiB. If the damage affects only some sectors and I can read the
damaged file with errors ignored (supposedly, dd_rescue does that), it
should be repairable. And, in any case, it's better than nothing ;-)
Also, my usual practice is to burn two copies on DVD (ie, 2 DVDs), so if a
file is damaged I can probably recover it from the other copy.
> There are ways to reduce this problem, and the one I chose was to limit
> the size of files to 100MB[1]. That, combined with my using 535 blocks
> means I can have 5 completely unreadable files before I am unable to
> recover. And when I want to recombine the split files, I just use cat
> :-)
A possibility, yes...
>
>
> [0] After rounding to the nearest MB:
> 4.35GB - 250MB = 4.1GB
> 4.1GB / 12 = 350MB
Yes, your assumption is correct.
>
> [1] split -b 100M -a 3 -d <filename> <filename>.
> ^
> That is to allow creation of names in the format filename.000,
> filename.001, etc.
Yep, I use the basename "recovery", making it obvious what they are.
- --
Cheers,
Carlos E. R.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)
Comment: Made with pgp4pine 1.76
iD8DBQFGQh7vtTMYHG2NR9URAhefAJoCnXL+ud3nJpp8/baZhEZ0flQ7rQCfTXP0
R5/UuwMgmdjxAan8JNzf4XA=
=dMSb
-----END PGP SIGNATURE-----
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
