"Fotis Mylonas" <fmylon@hotmail.com> wrote in message
news:%238PpOPdBIHA.4956@TK2MSFTNGP06.phx.gbl...
> Hallo dear all,
> We have a merge replication topology with a central server at the role of
> publisher and distributor and 8 subscribers around him. All machines have
> windows server 2003, and sql 2005 entrprise edition.
> the problem is this:
> the replication was not initalized by the publisher, instead we backed up
> the db, copied the zip backup file and restored it at every subscriber.
> after this i made subscriptions stating that i dont want the data to be
> initialized again.
> so far so good.
> the problem was that there were certain tables with data in the backup
> file, that were missing the rowguid column.
> because of this, the replication setup couldnt complete, so this column
> was added with a script that we run on every server.
> this allowed the replication setup to be completed, but it created a
> hidden synchronization problem for these tables, since due to the default
> value at the rowguid column (newid()) each row, now had a diferent value
> for this column on every server, although it was the same row actually.
>
> Newly inserted rows, dont have a problem at all, since the same rowguid
> travels to all the sql servers
> the problem is with the existing data on some of those tables.
>
> so i need to somehow manage to synchronize all the guid columns on all the
> servers and put the same value into them for every row.
>
> any idea how this can be done, since i only get an exception when i try
> it?
> i tried to delete the replication triggers and also drop the rowguid col
> property for the specific column, but im afraid this will mess my
> replication topology when im finished.
> my other option is to backup the central db, delete everything from all
> the tables and insert again the data row by row with a utility or
> something, so that each row will have a newly created guid, same on all
> servers.
> the only problem with this is that my tables are really big, so this will
> bomb the db andf the network with transactions.
>
> if i could only update the roguid column, things would be a lot easier.
> i have my own primary key i can base on.
>
> any suggestions ideas?
> any hidden magical stored procedures?
> any options like for example identity on off for the row guid col?
>
> Thank you all indeed,
>
> Love,
>
> Fotis
>

"Hilary Cotter" <hilary.cotter@gmail.com> wrote in message
news:%23twessdBIHA.4568@TK2MSFTNGP02.phx.gbl...
> sp_addtabletocontents
>
> --
> RelevantNoise.com - dedicated to mining blogs for business intelligence.
>
> Looking for a SQL Server replication book?
> http://www.nwsu.com/0974973602.html
>
> Looking for a FAQ on Indexing Services/SQL FTS
> http://www.indexserverfaq.com
> "Fotis Mylonas" <fmylon@hotmail.com> wrote in message
> news:%238PpOPdBIHA.4956@TK2MSFTNGP06.phx.gbl...
>> Hallo dear all,
>> We have a merge replication topology with a central server at the role of
>> publisher and distributor and 8 subscribers around him. All machines have
>> windows server 2003, and sql 2005 entrprise edition.
>> the problem is this:
>> the replication was not initalized by the publisher, instead we backed up
>> the db, copied the zip backup file and restored it at every subscriber.
>> after this i made subscriptions stating that i dont want the data to be
>> initialized again.
>> so far so good.
>> the problem was that there were certain tables with data in the backup
>> file, that were missing the rowguid column.
>> because of this, the replication setup couldnt complete, so this column
>> was added with a script that we run on every server.
>> this allowed the replication setup to be completed, but it created a
>> hidden synchronization problem for these tables, since due to the default
>> value at the rowguid column (newid()) each row, now had a diferent value
>> for this column on every server, although it was the same row actually.
>>
>> Newly inserted rows, dont have a problem at all, since the same rowguid
>> travels to all the sql servers
>> the problem is with the existing data on some of those tables.
>>
>> so i need to somehow manage to synchronize all the guid columns on all
>> the servers and put the same value into them for every row.
>>
>> any idea how this can be done, since i only get an exception when i try
>> it?
>> i tried to delete the replication triggers and also drop the rowguid col
>> property for the specific column, but im afraid this will mess my
>> replication topology when im finished.
>> my other option is to backup the central db, delete everything from all
>> the tables and insert again the data row by row with a utility or
>> something, so that each row will have a newly created guid, same on all
>> servers.
>> the only problem with this is that my tables are really big, so this will
>> bomb the db andf the network with transactions.
>>
>> if i could only update the roguid column, things would be a lot easier.
>> i have my own primary key i can base on.
>>
>> any suggestions ideas?
>> any hidden magical stored procedures?
>> any options like for example identity on off for the row guid col?
>>
>> Thank you all indeed,
>>
>> Love,
>>
>> Fotis
>>
>
>