[Date Prev][Date Next]
Re: segfaults with syncrepl
Ok, I solved my problem with glueing databases.I added ldif file
with ldapadd command then, removed database files from my /var/lib/ldap/sub-rid-00X directories and restart ldap. It worked. After strugling with this problem for a day it appeared that this is kind of very basics of openldap quick startup.
Ok, as I understand I have nothing to change at my master, right?
Do I miss something but I expect my replica server to display all my glued databases as one. Slapd.conf man page says, that slapcat should dump all glued databases, but onnly slapcat -b gives me results.
Why do you wrote syncrepl directive for suffix dc=example,dc=com? As I understand glueing works for databases under the same suffixes, so the last suffix dc=example,dc=com is needed only to create a global naming context and glue all replicated databases under the same tree.
I'm very new to subordinate and glueing databases so I probably miss something, but I can't get search to be performed under both databases.
I didn;t change config for master, and for slave I have (omited some unimportant things)
This solution gives me possible answer to my question above " why do I need add syncrepl directive to dc=example,dc=com database?". The replication of this database could let me get this database replicated, and won't need it to add manualy.
Am I right?
What is important - I have alias objects on server pointing from leaf in BranchX to leafs in BranchMain. Then having separate backends in replicas these alias won't work anymore. Will glueing databases solve this problem? Or should I think about other options? What could be the options in this case? I do auhtorization on BranchX uids, and aliases lets me have users to be able to authenticate in all the servers.
Alias problem is not solved. Although I can use my luma browser and see replicas as if they were one database, aliases are not working. Is this by design? Is there any way to make them work in glued databases? I just need a confirmation if somebody thinks I'm not reading manuals :)
And if i should forget aliases, what could be a work arround or alternative solution to get uids stored in o=BranchMain,dc=example.dc=com apear in o=BranchX,dc=example,dc=com on each replica?
Thanks in advance twice