[prev in list] [next in list] [prev in thread] [next in thread] 

List:       amanda-users
Subject:    Re: Problem with SAMBA and NT
From:       "Lishtovny Denis" <lishtovny () rtsoft ! msk ! ru>
Date:       2002-06-27 11:35:51
[Download RAW message or body]

(I need to backup NT Server 4.0 shares with Russian characters)
I'm tryed everything.
1)
my envirment in Linux (Debian) are:
$ set | grep '^LC'
LC_COLLATE=en_US.ISO8859-1 or ru_RU.KOI8-R or ru_RU.CP1251
LC_CTYPE=en_US.ISO8859-1 or ru_RU.KOI8-R or ru_RU.CP1251
LC_MONETARY=en_US.ISO8859-1 or ru_RU.KOI8-R or ru_RU.CP1251
LC_NUMERIC=en_US.ISO8859-1 or ru_RU.KOI8-R or ru_RU.CP1251
LC_TIME=en_US.ISO8859-1 or ru_RU.KOI8-R or ru_RU.CP1251

2)
samba is installed correctly:
smbclient //NT/Shares "pass" "-U" "user_who_can_full_access"
smb: \> ls
........ I see russian chars correctly

3)
smbtar -s NT -p pass -x SHARE  -u user_who_can_full_access -t tar.out
tar -xvvf tar.out
ls
I see russian chars correctly

after that I do backup
4) amdump DailySet1
5) amrecover DailySet1
6) amrecover> setdisk //NT/SHARE
7) amrecover> ls
2002-06-27 \305\311\321\272\321\241\324\341\265\277\277/
2002-06-27 \350\341\324\341\275\253\372 \253\337\321\241\375 2001/
2002-06-27 \350\341\324\341\275\253\372 \363\321\337\241\341 2002/


:((((((

is that trouble in amanda software?

> 1) create a file with line like:
>
> ./foo
> ./foo/bar
> ./abc/def/ghi/wxyz
>
>    make sure some of the names include your culprit characters.  Also
>    make sure that when you view this file whatever viewing software you
>    use doesn't convert the character into octal for ease of viewing.
>
> 2) pass the file through sed
>
> sed -e 's/^\.//'  < OrigListFile  > ModifiedListFile
>
> 3) check if the special characters are converted or unmodified.


I test this advice and the sed work correclty !.

Soory for my bad english, but i am realy need you help.
Thanks!


> > I can not backup NT shares with amanda because of the special characters
> > (e,a,e,...).
> > The indexer seems to store them in octal ("\350" for example) in the
files.
> >
> > So the backup program can not access to directory with special
character.
> >
> > If someone have any experience with Amanda and NT share ....
> >
> > I tried to create a share with samba and the backup program can access
to
> > all
> > the files but the indexer still storing them with octal, and so the
> > amrecover
> > program can not access the files on the tape.

> $ env | grep '^LC'
> LC_MONETARY=en_US.ISO8859-1
> LC_TIME=en_US.ISO8859-1
> LC_MESSAGES=C
> LC_CTYPE=en_US.ISO8859-1
> LC_COLLATE=en_US.ISO8859-1
> LC_NUMERIC=en_US.ISO8859-1
>
> Opps, my messages are still set to my old value 'C'.
>
> My suggestions would be:
>
> 1) check (however that is done) that all your computers are using the same
>    character set.  The system can have a default (at least unix systems),
>    but each user can customize their own values.  Make sure you check the
>    NT and UNIX systems as whomever does the backups
>
> 2) if you compiled your own samba suite, make sure it gets compiled with
>    the same character set in use as will be in use during backup.  If you
>    did not compile it yourself, check with your provider to see if there
>    is a version specific for your locale
>
> 3) ask the samba crowd about your problem
>
>
> Seems like I'm picking on samba right?  Here is the reason.  Amanda
doesn't
> do much except run the backup software.  The backup and the index are
> created by samba and amanda is just collecting it.  The index is just
> a second smbtar making a table of contents of the backup file.
>
> I can only think of one piece of software on the unix end that modifies
> the index after samba supplies it to amanda.  That program is sed.
> It takes the index made by samba's smbtar, which is of the form "./<path>"
> and strips off the leading "dot".
>
> You might be able to check if sed is the problem by the following.
>
> 1) create a file with line like:
>
> ./foo
> ./foo/bar
> ./abc/def/ghi/wxyz
>
>    make sure some of the names include your culprit characters.  Also
>    make sure that when you view this file whatever viewing software you
>    use doesn't convert the character into octal for ease of viewing.
>
> 2) pass the file through sed
>
> sed -e 's/^\.//'  < OrigListFile  > ModifiedListFile
>
> 3) check if the special characters are converted or unmodified.
>
> good luck





[prev in list] [next in list] [prev in thread] [next in thread]