FW: Back up Berkeley dbb files may cause file corruption

["Howard Chu" <hyc@highlandsun.com>]

Why don't we add a signal handler that will temporarily turn off write
access
to all backends, to allow online backup? Add a backend_sync handler entry
point,
etc. that will be called as well, and another signal to re-enable writes.
Should
be simple enough. There are still synchronization issues to think through,
re:
in-progress operations, but that may not be too bad.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc

-----Original Message-----
From: owner-openldap-software@OpenLDAP.org
[mailto:owner-openldap-software@OpenLDAP.org] On Behalf Of John Madden
Sent: Tuesday, September 25, 2001 6:58 AM
To: mrwm@tid.es
Cc: tnguyen55@socal.rr.com; openldap-software@OpenLDAP.org
Subject: Re: Back up Berkeley dbb files may cause file corruption


>    I had the same problem (data files get corrupted) but I don't know
> why. What version of Berkeley are you using? I'm using 3.1.17 and I
> want to know if it's the same with 3.3.11.
>
>> Hi, Just want to make sure that backing up dbb files (BTree structure
>> is used) while slapd are still running will not cause the data files
>> to get corrupted. Does anyone have similar experienced with OpenLDAP ?
>> If so, what are the most effective ways to back up your dbb files ?

Sorry, but I thought that this was a given -- it states in the documentation
that this won't work.  The state of the db files on disk isn't necessarily
what's in memory as long as slapd is running.  If you want to backup your
directory, kill -INT slapd and back them up.  If you have to do live backups
(no downtime), either export to LDIF or use replication...

John



--
John Madden
UNIX Systems Engineer
Ivy Tech State College
jmadden@ivy.tec.in.us