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

List:       openldap-software
Subject:    Re: Importing LDIF's exported by ldbmcat
From:       "Kurt D. Zeilenga" <Kurt () OpenLDAP ! org>
Date:       1999-05-24 23:35:19
[Download RAW message or body]

At 08:58 PM 5/22/99 +0200, Jochen Wiedmann wrote:
>
>Hi,
>
>we have recently exported an OpenLDAP servers contents with ldbmcat
>for manual modification of some attributes. Reimporting with
>ldif2ldbm seemed to work fine at first glance. However, it turned
>out that a lot of objects could not be found with ldbmsearch although
>they have been part of the database contents, as proofed by another
>ldbmcat run.
>
>When looking at the LDIF file we observed that the object numbers
>had the numbers 1, 257, 2, 258, 3, 259, ..., perhaps due to some
>hash sorting. We found a workaround by changing these numbers to
>1, 2, 3, ... and reimporting again.

That's odd.  The output of ldbmcat should be in index order.

I suggest submitting an ITS with appropriate details (including
version numbers of the underlying database system, operating
system, c translator, c library, etc.).

>Now my questions:
>  - Is our simple workaround ok? Or is some kind of object sorting
>    required?

Object need to be index order to ensure children entries always
have their parents.

>  - Would patches be accepted that change either of ldbmcat or
>    ldif2ldbm to deal with the situation? (An ldbmcat patch
>    seems quite straighforward, as we only need to change the
>    numbers printed.)

The problem is likely not in either of these two codes but
in the underlying database handling of the ldbmcat queries.

Kurt

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