#125 Re-synchronization is not implemented
Closed: Fixed None Opened 4 years ago by pspacek.

In-memory database should be periodically re-synchronized with LDAP database so inconsistencies caused by bugs will be resolved eventually.

The resynchronization mechanism itself is required for covering corner cases during ordinary re-connection to LDAP, see [wiki:BIND9/Design/RBTDB#Re-synchronization Resynchronization section in RBTDB design page].

Results of planning meeting held on 2014-11-04.

Implementation requires additional information from meta-database - every object will need some kind of generation id or something like that (for detection of deleted records).

For metadata storage see [[Design/MetaDB#Re-synchronization]]. We need to know also object class of the deleted object otherwise we would not be able to call correct deletion function.

I think that that very similar functionality might be required in order to support run-time changes to default TTL (#70).

Moving tickets as triaged on 2016-02-11 with mkosek.

This ticket was not fully closed, the commits above are just first attempts to fix the huge mess inside plugin.

Metadata Update from @pspacek:
- Issue assigned to pspacek
- Issue set to the milestone: Fedora 24

2 years ago

Login to comment on this ticket.