samba - Mar 2015 - Users and groups named in Russian disappeared after upgrade samba4 PDC

Thanks for the answer.
> Did you got errors during the classicupgrade?
I didn't do classicupgrade. My Samba AD DC was installed from scratch some 
months ago using the instructions from this page 
https://wiki.samba.org/index.php/Samba_AD_DC_HOWTO and everything had been 
working well. I just upgraded some deb-packages with samba to the same version 
but different buid-release. Regarding to the changelog there were some bug- and 
security-fixes.
> @all: Didn't we had someone here on the list with a similar problem?
> Accounts with chinese characters didn't worked or something like that?
I supposed that it could be a characters encoding issue because only users and 
groups named in Russian disappeared, default administrative accounts and user 
"dhcpd" that I created manually were available.
> About how many accounts do we talk about? Is it worth investigating or
> easier recreating?
You right. It was 36 accounts and 2 groups, not too many. After all my 
attempts to fix it I have recreated whole my domain and rejoined all clients. 
It was very tiresomely but I had no more time to try to repair it.
 
I am still conserned if some day it may happen again. It was strange that 
restoring backup didn't help. I thought that my operating system can be 
damaged, so I tried to restore the backup inside an lxc-container. I tried to 
restore some different backups that have been made in different days and months.
No one of them worked. Have all my backups been making broken? To backing up I 
use the instruction from this page 
https://wiki.samba.org/index.php/Backup_and_Recovery with little modifications, 
but I didn't try to restore it before the fault.

Am 11.03.2015 um 16:16 schrieb Vitaly Pyslar:
>> Did you got errors during the classicupgrade?
> I didn't do classicupgrade. My Samba AD DC was installed from scratch
some
> months ago using the instructions from this page 
> https://wiki.samba.org/index.php/Samba_AD_DC_HOWTO and everything had been 
> working well. I just upgraded some deb-packages with samba to the same
version
> but different buid-release. Regarding to the changelog there were some bug-
and
> security-fixes.
Hm. if you could create this accounts and they worked, then I don't see
a reason, why it should stop working.




> I am still conserned if some day it may happen again. It was strange that 
> restoring backup didn't help. I thought that my operating system can be
> damaged, so I tried to restore the backup inside an lxc-container. I tried
to
> restore some different backups that have been made in different days and
months.
> No one of them worked. Have all my backups been making broken? To backing
up I
> use the instruction from this page 
> https://wiki.samba.org/index.php/Backup_and_Recovery with little
modifications,
> but I didn't try to restore it before the fault.
What modifications?

And what was wrong with the restored backup? Did it has the same
problems? Or why it wasn't usable?

Do you just have one DC?


Regards,
Marc

> > I tried to restore some different backups that have been made in
different
> > days and months. No one of them worked. Have all my backups been
making
> > broken? To backing up I use the instruction from this page 
> > https://wiki.samba.org/index.php/Backup_and_Recovery with little
> > modifications, but I didn't try to restore it before the fault.
> 
> What modifications?
There is written in instruction:

How LDB and TDB files can be backed-up/restored.
you can backup *.ldb & *.tdb files online with 'tdbbackup'

But the provided script doesn't make baskups of *.tdb files, only *.ldb so I
decided to backup both of them. This is my backup script:

=============================================================================
#!/bin/sh

BACKUPDIR=/var/backup/samba
DAYS=90

DATE=$(date +%y%m%d)

if [ ! -d $BACKUPDIR ]
then
        echo "Missing backup directory $BACKUPDIR"
        exit 1
fi

if [ -f $BACKUPDIR/$DATE.tar.bz2 ]
then
        echo "Backup $DATE.tar.bz2 already exists"
        exit 1
fi

if [ -d $BACKUPDIR/$DATE ]
then
        rm -rf $BACKUPDIR/$DATE
fi

find /var/lib/samba -type f \( -name "*.ldb.bak" -o -name
"*.tdb.bak" \) -exec
rm -f {} \;

for db in $(find /var/lib/samba -type f \( -name "*.ldb" -o -name
"*.tdb" \))
do
        tdbbackup $db

        if [ $? -ne 0 ]; then
                echo "Error while backuping $db"
                exit 1
        fi
done

mkdir $BACKUPDIR/$DATE
rsync -aX --exclude='usershares' --exclude='*.ldb'
--exclude='*.tdb'
/var/lib/samba/* $BACKUPDIR/$DATE/

mkdir $BACKUPDIR/$DATE/etc
cp -dpR /etc/samba/* $BACKUPDIR/$DATE/etc/

cd $BACKUPDIR/

for db in $(find . -type f \( -name "*.ldb.bak" -o -name
"*.tdb.bak" \))
do
        mv $db ${db%.bak}
done

tar -cjf $DATE.tar.bz2 $DATE >/dev/null 2>&1

rm -rf $DATE

find /var/lib/samba -type f \( -name "*.ldb.bak" -o -name
"*.tdb.bak" \) -exec
rm -f {} \;

find . -name "*.tar.bz2" -mtime +$DAYS -exec rm {} \;

exit 0

=============================================================================
Is it correct?
It is launched by cron every night.

To restore a backup I stopped the samba service, removed contents of the 
/var/lib/samba directory, untared the archive there and started the service 
again. I tried to restore the backup of the previous day, the prevous week and 
the previous month.
> And what was wrong with the restored backup? Did it has the same
> problems? Or why it wasn't usable?
There was no success, the problem didn't disappear. I know it's a some
kind of
mistery. Only one logical explanation that I see is that in some point of time 
my database has become corrupted (don't know why and how) and all the
backups
have been made with broken data. Long running DC has been working somehow 
until upgrade restarted it and after that the problem appeared. Hense the 
reason of the fault might be not the upgrade, it could have happened before 
somewhere else. I don't know what could it be, all that I've done since
then I
have installed the DC added/removed users via ADUC.
> Do you just have one DC?
Yes. I know I had to have a secondary domain member server for redundancy but 
unfortunately I don't have it right now. The company I talk about is very 
small, the hardware of the current server is quite powerful and reliable so 
they won't to buy one more server only for another samba instance. I decided
to run both DCs - primary and secondary in separate lxc-containers or in  
virtual maschines and snapshot them periodically. I hope it will help me to 
restore quickly the whole environment in case of a fault.

Thanks a lot for your help.