2024-04-30T08:46:46Z DEBUG Start of dirsrv@IPA-TEST.service complete
2024-04-30T08:46:47Z DEBUG retrieving schema for SchemaCache url=ldapi://%2Frun%2Fslapd-IPA-TEST.socket conn=<ldap.ldapobject.SimpleLDAPObject object at 0x7fa2b3ef4b60>
2024-04-30T08:46:47Z INFO Local roles match globally used roles, proceeding.
2024-04-30T08:46:47Z INFO Backing up ipaca in IPA-TEST to LDIF
2024-04-30T08:46:47Z INFO Waiting for LDIF to finish
2024-04-30T08:46:47Z INFO Backing up userRoot in IPA-TEST to LDIF
2024-04-30T08:46:47Z DEBUG   File "/usr/lib/python3.12/site-packages/ipapython/admintool.py", line 180, in execute
    return_value = self.run()
                   ^^^^^^^^^^
  File "/usr/lib/python3.12/site-packages/ipaserver/install/ipa_backup.py", line 343, in run
    self.db2ldif(instance, 'userRoot', online=options.online)
  File "/usr/lib/python3.12/site-packages/ipaserver/install/ipa_backup.py", line 525, in db2ldif
    raise admintool.ScriptError(

2024-04-30T08:46:47Z DEBUG The ipa-backup command failed, exception: ScriptError: Unable to add LDIF task: This entry already exists
2024-04-30T08:46:47Z ERROR Unable to add LDIF task: This entry already exists

systemctl start ipa.service
ipa-backup --data --online

I think this is because there are two tasks being added within the same second under each other and they don't have name difference. One for CA backend and one for the main tree.

the fix would be to add a backend name into the 'cn' -- e.g. change line 504 to cn = time.strftime(f'{backend}_export_%Y_%m_%d_%H_%M_%S')

es, I think it is a bug:

            if os.path.exists(paths.VAR_LIB_SLAPD_INSTANCE_DIR_TEMPLATE %
                              instance):
                # Check existence of ipaca backend
                dbpath = (paths.SLAPD_INSTANCE_DB_DIR_TEMPLATE %
                          (instance, ""))
                output = run_dbscan(['-L', dbpath])
                if 'ipaca/' in output:
                    self.db2ldif(instance, 'ipaca', online=options.online)
                self.db2ldif(instance, 'userRoot', online=options.online)
                self.db2bak(instance, online=options.online)

see there are two self.db2ldif() calls each after other

and since strftime() use includes only second level resolution, that's not enough