Learning OpenLDAP, following an "older" tutorial-- hdb vs. mdb update needed, adjusting a step to move forward
My interpretation is that it is like classes in HTML-- you combine searching for classes until you distill down to the distinct subset of elements you wanted. Which may be an incorrect view.
This would be right for 'ldapsearch', where you give it a filter that matches against certain attributes (which often includes objectClass as the 1st condition) and it spits out the found entries.
But it's not correct for 'ldapmodify'. When modifying or deleting an entry, you always specify the exact entry DN that you want to update; in LDIF that's the dn:
field. This is not an attribute-based filter at all – an entry's DN is its exact unique path, very much like a filesystem path.
(Often LDAP modifications are indeed made by first performing an attribute-based search to determine the entry DNs, then modifying each entry by its DN one by one.)
So when looking at a 'ldapsearch' or 'slapcat' output, you just take the value of the 'dn:' shown as the 1st line of each result.
In your case, the entry describing the main database is olcDatabase={1}mdb,cn=config
.