Re: Pls Help: Recover snapshot after the refresh group is broken

Hi Jian,

If you force the jobs to broken by taking the master site down and waiting for the 16th refresh to break the jobs then snapshots will remain broken after the master is started up.

To resolve you call dbms_job.broken(<job>, false) on each replication job that is marked as broken.
The broken jobs run immediately and catch up as long as the snapshot logs are still present at the master.

Calling dbms_refresh.refresh(name=>'group') will only force a refresh of the snapshot group by whatever method is default for the snapshots; - the jobs still remain broken
dbms_job.run(job) might be a better way to do it.

incidentally , 10seconds is a pretty tight refresh rate for a snapshot group - you are probably better going up to a interval of 5mins (minimum) if you can spare the time lag in the replica. This is more relevant if you have replication running across small bandwith network links.

another thing to bear in mind with fast refresh snapshots is the following scenario..

master table has 1000 rows; master and snap are in sync with fast refresh snapshot
link between master and snapshot site goes down master undergoes updates to every row 3 or 4 times - snapshot log grows to 4000 rows

link comes back between master and snapshot site.

now whats quicker - a full refresh of the snapshot or a fast refresh processing 4000 row changes?

hope this helps,

Dave. Received on Fri Jul 02 1999 - 17:34:09 CDT