<quote who="Scott Classen">

...

...

I thought by installing openldap 2.4.6 on a second machine and using a short 10-15 line seed.ldif file it would talk to the master LDAP server, get the cn=config from the master and see that in addition to the cn=config base there is also another BDB base (dc=example,dc=als,dc=lbl,dc=gov), and then it would sync up that guy too.

But if you replicate the config, you are creating another master. You have to create a seperate config for the slave. There is no harm from then on to create a seed.ldif and replicate your cn=config from another slave.

Gavin.

Yes, I see your point. This was a concern of mine. I was suspicious that simply replicating the cn=config from the master would not work as a seed for a slave I was envisioning.

So I need a slave-specific cn=config?

Or just one master to replicate from.

Is it possible to store a second cn=config on the master server that is specific for slave machines? How would you do this? What would it look like?

Hmmm..., you could do (not tried this):

slapcat -n 0

Do this on a configured slave to get the right config.ldif etc.

change all DNs to whatever you plan to store the data under, and then use the rwm overlay to deliver it all out of the directory in the original cn=config format.

You could then replicate from that.

Just a thought.

Could I make a cn=slave,cn=config and use that as the cn=config for all slave machines.

Thanks, Scott