I'm reviewing our backup/restore scripts and processes. The directory in question has a single suffix containing about 41,000 entries. It has a single R/W instance, with several R/O replicas, running 389-Directory/1.4.4.

Working with on-line data, if needed, we could turn one of our R/O instances into a R/W, and create some new replication agreements. Or, we could build a new R/W instance, and populate it by exporting the required information from a R/O instance. We'd have additional work to do with certs, DNS, and hostnames to put the finishing touches on.

To make an off-line backup of our R/W instance, we have a scripted process which does:

* dsconf userdata backup create * dsconf userdata backend export

makes a tar of those and the contents of:

* /etc/dirsrv/slapd-userdata (to pick up certs and scheme extensions)

and parks the resulting .tar in a safe place for a couple of months.

My questions are:

1. After looking this over, I don't see that we are capturing cn=config Is that implicit in the dsconf 'backup'? Are we just missing it? If so, how best to capture the objects we want from it? 2. Does the answer change as we shift to 389-Directory/2.1.8 ?