Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
AskSQLTeam
Ask SQLTeam Question
0 Posts |
 Posted - 2000-12-13 : 11:24:15
|
Peter writes "I have SQL Server 7 SP2 running transactional replication and stumbled across a serious bug.
If the publisher contains a cascade update trigger which is fired because a subscriber has changed a foreign key in a parent table the cascade updates of the child tables are sent back to all other participating subscribers except the one that initiated the parent table update.
The 2 phase commit process does not need to send back the parent table update since the subscriber has already done so but should send any cascading transactions the publisher performed on its own since the cascading triggers don't and shouldn't exist on the subscriber side.
The results are catastrophic since it creates orphan records for that subscriber and is out of sync with the publisher.
Regards
Peter" |
|
|
|
|
|
Serious Bug in Replication