>> >> +Entry cn=changelog5,cn=config is added >> +Processing 03replica.ldif ... >> +++check_and_add_entry: Entry not found cn=replica,cn="o=NetscapeRoot",cn=mapping tree,cn=config error No such object >> +ERROR: adding an entry cn=replica,cn="o=NetscapeRoot",cn=mapping tree,cn=config failed, error: No such object >> dn: cn=replica,cn="o=NetscapeRoot",cn=mapping tree,cn=config >> objectclass: top >> objectclass: nsDS5Replica >> objectclass: extensibleObject >> cn: replica >> nsds5replicaroot: o=NetscapeRoot >> nsds5replicaid: 1 >> nsds5replicatype: 3 >> nsds5flags: 1 >> nsds5replicapurgedelay: 604800 >> nsds5replicabinddn: cn=Replication Manager,cn=config >> >> +ERROR: There was an error processing entry cn=replica,cn="o=NetscapeRoot",cn=mapping tree,cn=config >> +Cannot continue processing entries. >> Error adding entry 'cn=replica,cn="o=NetscapeRoot",cn=mapping tree,cn=config'. Error: No such object >> Error: Could not create directory server instance '389-master01'. >> Exiting . . . >> >This means the parent entry cn="o=NetscapeRoot",cn=mapping >tree,cn=config does not exist - yes, this is a problem because, when >setting up the server to be a configuration directory server, the >o=NetscapeRoot suffix and database are added later, after the initial >instance creation, when the ConfigFile directives are processed. I >suppose you could add those as well - use the file >/usr/share/dirsrv/data/template-suffix-db.ldif - make a copy - replace >%ds_bename% with NetscapeRoot and %ds_suffix% with o=NetscapeRoot - use >that file as the first ConfigFile directive. Then, when >setup-ds-admin.pl tries to add those, it should be ok if they already exist. Thanks I will try the re-ordering and see if I can get it working that way. I discovered a previous thread with the same file which also mentions a bug in the documentation. The bug still exists I will raise an bugzilla to get the documentation clarified on the requirements of creating a root entry. The thread I am referring to is: http://www.mail-archive.com/fedora-directory-users at redhat.com/msg08032.html Something else occurred to me. If you have a shared/replicated NetscapeRoot database and lets say 12 servers over 3 datacentres, 6 providers and 6 consumers. You will end up with 12 servers in a multimaster group for the netscaperoot database but only 6 servers in a multi master setup for the userdata database. That seems quite a lot of masters for NetscapeRoot. Is that considered to many and/or is there any recommendations when having that many servers? Regards ________________________________________________________________________ In order to protect our email recipients, Betfair Group use SkyScan from MessageLabs to scan all Incoming and Outgoing mail for viruses. ________________________________________________________________________
