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

Re: RFC2589 implementation

On Mon, 2006-01-02 at 10:49 +0100, Pierangelo Masarati wrote:
> I refreshed the overlay a bit and now also replication seems to work
> fine.  Submitted as ITS#4293.

Open issues:

- what happens if a non-leaf object expires?  in the proposed
implementation, it survives until all subordinates (which have to be
dynamic objects) expired.

- what happens if a dynamic object expires on the master and the client
re-adds it, but it did not expire yet on the replica?  in the proposed
implementation, when the slave detects an add of a dynamic object that
is already present, it first tries to delete it, under the assumption
that since it's a slave this can only occur for replication delay timing

- what happens if a dynamic object is deleted on the master, but it
already expired on the replica?  In the proposed implementation this
case is not handled; a .rej is generated, but the replica should remain
consistent with the master.

I see many replication related issues; in general, however, the
assumption that dynamic objects will expire some time should lead to
eventual convergence.



Ing. Pierangelo Masarati
Responsabile Open Solution
OpenLDAP Core Team

SysNet s.n.c.
Via Dossi, 8 - 27100 Pavia - ITALIA
Office:   +39.02.23998309          
Mobile:   +39.333.4963172
Email:    pierangelo.masarati@sys-net.it