Re: Replication sequence of events

Jake,
The sequence of events which you described does not sound to me like it should cause the quiescing problem you are seeing. I have some example scripts on my website which you might want to look at. You can compare your scripts with mine and see if you notice any differences in your setup. (I also have some info on conflict resolution and replication monitoring which could be helpful too.)

http://www.dotcomsolutionsinc.net/products/repgen/repgen_files.html

http://www.dotcomsolutionsinc.net/support/repgen/repgen_tn_replication_desig n.html

David Simpson
www.dotcomsolutionsinc.net
(Oracle Replication Consulting & Tools)

"Jake" <me_at_heyjay.com> wrote in message
news:aol48m$a6l$1_at_bob.news.rcn.net...
> Hi,
>
> I'm working on 8.1.7.2/win2k, I've got some replication problems
> (I'm stuck in quiescing, but that's not my question). I've looked through
> metalink, and I'm re-reading the Oracle8i Replication manual, but I'm
> not sure if I've got this straight.
>
> I'm doing n-way master/master, Is this sequence of events correct?
>
> --- No conflict resolution has been implemented ---
>
> 1) User U1 on db A updates row 1. The row is updated locally
> (e.g. on db A), the information necessary to make the RPC to db B is
> put into the deferred transaction queue, and everything is committed.
>
> 2) User U2 on db B updates row 1. The row is updated locally
> (e.g. on db B), the information necessary to make the RPC to db A is
> put into the deferred transaction queue, and everything is committed.
>
> 3) db A pushes its queue. The transaction fails at db B due to U2's
> actions. The transaction is rolled back on db B. The error is logged
> in deferror on db B. The transaction is put into the deferred queue on
> db B with a destination of db B. So that every time db B pushes it's
> queue it attempts this failed transaction on itself.
>
> And, does this situation/error prevent me from suspending my master
> group activity, e.g. causing my quiescing for infinity problem.
>
> Thanks
> Jake
>
>
Received on Tue Oct 22 2002 - 14:57:43 CDT