Terry, in order for me to help you, please check the items that I have
outlined in my reponse to your previous post.
By not using drop as the pre-creation command, you are making the snapshot
process as difficult for us as possible since we have to work through the
constraints you have placed on the subscriber schema (the presence of
indexes can also make the bulk-load process [very] sub-optimal). As such,
can you please explain why you have to use delete as the pre-creation
command? Perhaps I can suggest another way of achieving the same result?
As I mentioned in my previous response, one of the possibilities of snapshot
process having trouble with foreign keys is that you have a table outside of
the publication referencing a table inside the publication (or you have
circular fk references?) Again, please look through the possibilities that I
have outlined in my previous post and see if any of them applies to your
scenario.
You may also want to pursue no-sync subscriptions if you don't need to move
the initial data-set between the two servers. This is relative trivial to do
in the SQL2005->SQL2000 direction but requires a moderate amount of work in
the SQL2000->SQL2005 direction.
-Raymond
[quoted text, click to view] "Terry Mulvany" <terry.mulvany@rouseservices.com> wrote in message
news:OR2YqFaQHHA.1860@TK2MSFTNGP06.phx.gbl...
> when you implement transactional replication how many times does the
> inital snapshot have to run? i would assume from the word 'initial' it
> would be just once, but just asking to clear this up.
>
> i am also wondering because it seems like running this initial snapshot is
> causing the most problems. ordering seems to be a problem. i have my table
> articles using the delete data option rather than dropping and for
> instance, i have a table that when data deletion an error is reported
> saying it can't because it contains a foreign key to another table. well
> of course it does, so why isn't it smart enough to delete records from
> that table FIRST so there are no errors.
>
> can someone explain the best and smoothest way to accomplish the initial
> snapshot given that i have two identitcal databases on each side (schema)
> and i need to maintain all keys, constratins, indexes, etc. for all table
> articles in the publishing.
>
> also, if i have identified say 10 tables to be replicated and they each
> have dependencies (other tables have a foreign key into this table) or are
> depended on by other tables outside of the tables needing to be
> replicated, do all tables need to be included?
>
> thanks all.
>
