Re: [rdiff-backup-users] Weird behavior of rdiff-backup : founding diffs while there isn't

[Matthieu . Rioteau]

Hi Bob,

> From: address@hidden
> To: address@hidden
> Date: 20/11/2012 15:39
> Subject: Re: [rdiff-backup-users] Weird behavior of rdiff-backup : 
> founding diffs while there isn't
> Sent by: address@hidden
> 
> I would say there is no "probably" about it.  Another option might 
> be to put the
> database in read-only mode during the backup and then restart the server 
in
> read/write mode afterward, but that would still leave a momentary 
> gap in service
> while the server is being restarted.  You might need to delay the 
> restart until
> _both_ backups were complete.  I'm no LDAP expert (never used it, 
infact), so
> take that with a grain (or two) of salt.

Actually, LDAP server is down during backup of the machine that runs it 
because the server itself is backed up (settings, database, ...).
But second backup (faulty one) won't never ever touch it, except for Uname 
resolution (read only access). Thus I don't think it is a problem if it is 
running during this particular one.

 
> As I said, I have doubts about whether that will help.  The names are 
still
> recorded in the mirror_metadata file even when 
"--preserve-numerical-ids" is
> used.
 
On my side I think is that it could work. Even it should work if option 
description in the manual is correct. ;-)
That's why I want to try it. Actually I will then time shift one of the 
backup, but keep this option as a "safety option" if it works.

> No problem.  Those of us who lurk on this list do it because we enjoy
> tracking down issues like this.

Fact that you enjoy helping people like me is not a sufficient thing so I 
don't have to thank you. :-)

I mistook my test on past night. I will take some days to perform some 
different tests.
Then I will come back to you with an updated clear and complete status 
about what helps and what don't in situations like mine (expect update end 
of this week/beginning of next).

See you,
Matthieu