Hoping someone can give me some options on this one as it is driving me crazy
Transactional Publication publishing a single table (56m rows)
Table is two column (date and id) primary key (4 column table (value and archive flag)
No Filter columns, no filter rows
@ins_cmd = N'SQL', @del_cmd = N'CALL usp_NoAction3', @upd_cmd = N'SQL'
usp_NoAction3 is an empty SP with a single print command in (no Action taken)
Subscriper is a pull subscription
Publisher ServerA is where all data is entered
Subscriber ServerB is a reporting layer, data is not meant to be manipulated and I can't find any evidence that anyone is manipulating it (although i can't rule it out)
Problem we are having is when certain historic records are updated on ServerA we get a key violation on the subscription where it is attempting to insert a new record (rather than performing an update)
insert into "tbl_data" values ({ts '2015-05-08 01:52:24.000'}, '2720YC0100G6', 3.6000000e+002, NULL)
Cannot insert duplicate key row in object 'tbl_Analyt_data' with unique index 'PK___1__13'.
Checking the table on both servers the record exists on both tables with the same ID and the same timestamp.
The only difference is the value field has been updated from a value of 400 to 360
The datestamp on the field is 08-05-2015 01:52:25.000 on both tables so it was originally inserted fine.
Anyone care to offer a hypothesis as to this behaviour?
Cheers
Smitty
MCDBA, MCAD, MCITP, MCTS