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

Re: Sync Replication searchbase muset be a suffix of a db?



Pierangelo Masarati wrote:

Things go smooth if the searchBase in the consumer is equal to the suffix
(but subtree deletion fails because of ITS#3385, fixed in HEAD.



On a related note: if using refreshOnly, context suffix deletion at the producer side results in no deletion at the consumer's side. I guess that when refreshing after deletion at the provider side, the consumer is supposed to get noSuchObject, which results in no further action. Is this intended, or should the consumer decide to commit suicide as soon as it gets noSuchObject from the provider?

Good question. I guess the most useful behavior would be for the consumer to delete the entry and keep on with its usual refresh schedule, in anticipation of a new entry taking its place on the provider.

In HEAD, if a persist session runs into this situation, the persistent search is terminated with a SyncRefreshRequired error code. Of course, in this case, the next Refresh would probably fail.

--
 -- Howard Chu
 Chief Architect, Symas Corp.       Director, Highland Sun
 http://www.symas.com               http://highlandsun.com/hyc
 Symas: Premier OpenSource Development and Support