Jake wrote:
>>afaik, the push job pushed out deferred transactions (i.e.
>>transactions not cascaded to all snapshots), the purge job
>>purges transaction/data records from the master tables, only
>>for those transactions that have been propagated to all
>>registered snapshots.
>>If you picture refresh on demand, the mechanism, and the need
>>for it, becomes obvious.
>>
>>Grtz, Frank
>>
>
>
> Maybe I'm being thick but are you saying:
> 1) when I refresh, the changes that occurred at my local (snapshot) don't
> get pushed to the master?
> 2) the local changes (at the snapshot) are only applied at the master site,
> during the push job?
>
> I thought during a refresh, first the local changes are pushed to the
> master, then all the changes are pulled to the snapshot
>
> Jake
>
>
Correct. What if I just got back from holidays, and logged into
the network the first time since 2 weeks? All those transactions
would still be waiting for me. Nevertheless, if the replication
on demand fails, there's a job scheduled to make sure it's attempted
over and over again.
Normally, this job will have no work.
Received on Sun Jan 12 2003 - 07:45:52 CST
