[Date Prev][Date Next]
Re: Multi-LDBM Replication
Garry Thomas wrote:
> I have setup a multi-ldbm directory service with replication,
> and have found that the version (Redhat's 2.0.11-8 src rpm) of SLURPD I
> have been using needs to be 'contained' to a single LDBM. The problem is
> when you have multiple replica instances in SLAPD.CONF (one for each
> LDBM) SLURPD will try and perform replications to all LDBM's ignorant of
> the fact that each replica statement is associated with different
> To fix this, and 'contain' it, I created individual config files
> for a separate SLURPD instance per LDBM, and make SLAPD create an
> LDAPREPLOG file for each LDBM. So if I have 3 LDBM's, I create 3 config
> files with replica config statements in each, and force SLURPD to use
> this config file instead of SLAPD.CONF with the -f switch. Everything
> works fine.
> My question is two-fold:
> 1)Is this the 'proper' way of doing this?
> 2)I notice on the CVS that SLURPD has added the
> suffix=<dn> parameter to the replica config directive. Should I use this
> now, and spawn a single SLURPD, and if so, what version of OpenLDAP
> should I use?
Point 1) is fine, although not straightforward. You may implement
it in a somewhat "smart" way by heavily using "includes", which
are supported also by slurpd since 2.0.12.
Point 2): it works correctly since 2.0.13; in this case you need
to put everything under a single database (i.e. you need a
least common naming context for all your databases, which could be "").
In this case, you can instruct each replica to follow only some subtrees
by using multiple occurrences of the "suffix" parameter. Note that
in this case the slaves can have the "suffix" they handle as naming
Dr. Pierangelo Masarati | voice: +39 02 2399 8309
Dip. Ing. Aerospaziale | fax: +39 02 2399 8334
Politecnico di Milano | mailto:email@example.com
via La Masa 34, 20156 Milano, Italy |