[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Partial replication, remove branch

Le 20/03/2012 12:37, anax a écrit :

On 03/20/2012 10:54 AM, jehan procaccia wrote:

I would like to replicate only some OUs under the baseDN ; ou=people and
ou=group,ou=system, but not the remaining of OUs below ou=system =>
ou=Hosts , ou=Networks, ou=Protocol.
How can I remove those branches to replicate ?
my actual syncrepl config that replicate all the subtree branches:
syncrepl rid=001
retry="60 10 300 +"
updateref ldaps://master.domain.fr:636

Define the ACL for binddn="cn=replic,ou=System,dc=int-evry,dc=fr" such that it cannot access the ou's you don't want to sync.


Thanks, I achieved a partial replication to only wanted branches, as you suggested by restricting ACL to the replica's account on the branches/attributes I want. However that's not an easy config to set up , I noticed that as soon as I forgot to mention an attribute in a subtree object, all the objects in that subtree aren't replicate, that's the same for a branches DN node , I initially forgot the attribute associatedDomain which was part of that object for example, then that object node and all subtree objects below weren't replicated .
So I ended with many more ACLs like that :

#ou=system,dc=int-evry,dc=fr BaseDN ACL to get ou=system object node
access to dn.exact="ou=system,dc=int-evry,dc=fr"
        by dn="cn=admin,dc=int-evry,dc=fr"                      write
        by dn="cn=replic,ou=System,dc=int-evry,dc=fr"             read
        by users                                                read
#Goups and associeted attributes
access to dn.subtree="ou=Group,ou=System,dc=int-evry,dc=fr" attrs=cn,sn,memberuid,member,mail,description,entry,objectclass,associatedDomain,gidNumber,ou
        by dn="cn=admin,dc=int-evry,dc=fr"                      write
        by dn="cn=replic,ou=System,dc=int-evry,dc=fr"             read
        by users                                                read

How can I check performance issue with all the ACL I added ? is there a program to test / bench the ACLs or optimise them ?

Thanks .