Container doesn't start for a second time, because default database changed

[Kritzefitz]

If prepopulation data is provided, the container only starts once and then never again:

$ mkdir prepopulate
$ cat > prepopulate/test.ldif
dn: ou=example,dc=example,dc=org
objectClass: organizationalUnit
ou: example
$ podman run --name test-ldap -e 'SLAPD_PASSWORD=12345' -e 'SLAPD_DOMAIN=example.org' -v "$PWD/prepopulate:/etc/ldap.dist/prepopulate" docker.io/dinkel/openldap:latest
657312c4.38b3a632 0x7f474441e200 @(#) $OpenLDAP: slapd 2.5.13+dfsg-5 (Feb  8 2023 01:56:12) $
	Debian OpenLDAP Maintainers <[email protected]>
657312c4.38f56d00 0x7f474441e200 slapd starting
^C657312c7.37ac3eae 0x7f4702fff6c0 daemon: shutdown requested and initiated.
657312c7.37b0a93b 0x7f4702fff6c0 slapd shutdown: waiting for 0 operations/tasks to finish
$ podman start --attach test-ldap
Info: Container already configured, therefore ignoring SLAPD_xxx environment variables and preseed files
mdb_id2entry_put: mdb_put failed: MDB_KEYEXIST: Key/data pair already exists(-30799) "ou=example,dc=example,dc=org"
=> mdb_tool_entry_put: id2entry_add failed: err=-30799
=> mdb_tool_entry_put: txn_aborted! MDB_KEYEXIST: Key/data pair already exists (-30799)
slapadd: could not add entry dn="ou=example,dc=example,dc=org" (line=1): txn_aborted! MDB_KEYEXIST: Key/data pair already exists (-30799)

This is because the check that guards loading the propupulation data checks for /var/lib/ldap/DB_CONFIG, but that file is only created by the berkeley db backend of openldap, but the new default backend in openldap after version 2.5 has been changed to mdb. This means a freshly populated /var/lib/ldap looks like this:

$ ls -l /var/lib/ldap
total 64
-rw------- 1 openldap openldap 61440 Dec  8 13:04 data.mdb
-rw------- 1 openldap openldap  8192 Dec  8 13:04 lock.mdb

and the prepopulation code path will be run every time, failing on every start but the first.