Jay the renaming of the servers is a tough one for replicated scenarios.

If you can take the old db offline while the new come online you "could" potentially getaway with...

there is a view that allows you to see the count of (non)distributed commands in distribution database

View name: MSdistribution_status

Column of interest: UndelivCmdsInDistDB

I would think the last bullet would be to run a full dbcc checkdb on it but it does not seems logical ...

Because some seem "bad" and some others don't I would then check the clients maybe that garbage is actually being sent into the db for a mis-behaving client!

Ali (3/31/2009)

---

Correct,

In all the cases I've seen, they're SP calls that simply use the UPDATE statement to update the NTEXT values from the value passed in the parameter. I see...

Replication has as much impact on the subscriber as the activity on the primary... You can use READ_COMMITTED_SNAPSHOT to minimize locking issues with possible queries coming from the linked server...

Oh Brother... good luck 🙁

The updates are being done with standard "UPDATE" commands, right? not UPDATETEXT, correct ?

I have seen occasions in which the rows do NOT reach the subscribers when Reindexing is run on the publisher but never seen that a row is sent twice.

By the...

Good Luck!

🙂

Maybe you have a "corrupted" Source too then.

You should check the source table and see if there is any Error regarding the LogReader.

Well that is correct if you want the index generated. So I can on take another "guess":

There is also another place to check. If the table articles are not marked...

This is a shot in the dark: Do you have "max text repl size (B)" setup to 2GB?

If someone ( or some program) inserted the row on the subscriber this error can happen.

If you manually delete it replication picks up where it left off.

There is no...

Believe me, I am not :w00t: