Re: ..weeding out bad debs off mirrors, was: Can I get some help, please?

[Arnt Karlsen <arnt@c2i.net>]

On Fri, 25 May 2007 14:13:45 -0300, Carlos wrote in message 
<[🔎] 18007.6473.431879.8283@fisica.ufpr.br>:

> Arnt Karlsen (arnt@c2i.net) wrote on 22 May 2007 14:54:
>  >On Tue, 22 May 2007 07:01:39 +0200, Klaus wrote in message 
>  ><46527933.6020207@gmx.de>:
>  >
>  >> Arnt Karlsen wrote:
>  >> > ..one thing I haven't found, is an easy way to weed out bad files off
>  >> > my (big bad disk, replacement ordered) mirror after fsck's, my plan is
>  >> > use md5sum down the debian/tree and weed out all the bad files, instead
>  >> > of wasting another week re-downloading all the good stuff (right now
>  >> > I have 71GB of i386 and source) to build a full mirror.  Pointers?
>  >> 
>  >> Isn't that what debmirror does anyhow to verify the integrity of the mirror?
>  >> 
>  >> At least I seem to remember that bad or missing files were fixed with the next remirror of my mirror.
>  >> 
>  >> Cheers,
>  >> Klaus
>  >> 
>  >On Tue, 22 May 2007 08:25:38 +0100, Jonathan wrote in message 
>  ><cf3b996d0705220025lda0a341xa273204c1e16519c@mail.gmail.com>:
>  >
>  >> doesn't rsyncing take care fo this by itself updating the mirror?
>  >> 
>  >> mirror.ac3bf1.org
>  >> John
>  >
>  >..IME no (but I use the anonftpsync script), as long as the file names 
>  >matches, all files are accepted as good, I see no md5sum kinda check
>  >done. I appreciate the "do not dump that load upstream", so I'm looking
>  >for some kinda find *.md5sum* ->xarg md5sum-c -> rm -vf baaad.debs
>  >oneliner or somesuch to verify my mirror update, pointers?
> 
> rsync will check the integrity of existing files if used with option -c.
> However this puts a high load on the server so many of them disable

..aye.  ;o)

> its use. A debian mirror does not need to allow it.
> 
> The default is to not use -c; in this case only the date and length
> are used to decide if a file should be updated.
> 
> One can use the checksums in indices/md5sums.gz to verify the files
> locally. Very useful indeed... I do it periodically, because a mirror
> is a significant responsibility.

..aye, but _precisely_how_ do you do it?  Playing with Rafal's
'zcat indices/md5sums.gz |md5sum -c |grep -v ": OK" |grep -v ' etc
from the mirror root, I keep finding new innovative variations of
": Failure".

-- 
..med vennlig hilsen = with Kind Regards from Arnt... ;o)
...with a number of polar bear hunters in his ancestry...
  Scenarios always come in sets of three: 
  best case, worst case, and just in case.