Replication Distribution Subsystem agent failed in pull replication

Question

Replication Distribution Subsystem agent failed in pull replication

gautam.expounder

SSC Enthusiast

Points: 102
More actions
March 25, 2010 at 4:57 am

#219811

We have two nodes N1 and N2 with N1 being the distributor and N2 being the Subscriber.
We have Transactional replication running.
Error we got is:
Replication-Replication Distribution Subsystem: <server details> : The row was not found at the Subscriber when applying the replicated command.

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply

mike mcneer Ten Centuries Points: 1189 More actions · Answer 1

So what is the question? The row was not found at the subscriber like the error says. In another post I detail how to find the offending record then you will need to figure out why it is not in sync with your publisher.

Thanks,

Mike McNeer

josebrwn Old Hand Points: 393 More actions · Answer 2

we have the same problem - did you figure it out?

Log Name: Application

Source: MSSQLSERVER

Description:

Replication-Replication Distribution Subsystem: agent .... failed. The row was not found at the Subscriber when applying the replicated command.

noeld SSC Guru Points: 96590 More actions · Answer 3

Your primary probably issued a *delete* and your row "was already not present" at the subscriber!

* Noel

josebrwn Old Hand Points: 393 More actions · Answer 4

I think you are right!

The puzzling bit is that replication agents continue happily along, as though nothing were the matter. You have to look in the application event log to see the problem.

So, we're in a push model, as opposed to a pull ... is there no other way to restart things other than to reinitialize from a new snapshot?

noeld SSC Guru Points: 96590 More actions · Answer 5

josebrwn (4/29/2010)
I think you are right!
The puzzling bit is that replication agents continue happily along, as though nothing were the matter. You have to look in the application event log to see the problem.
So, we're in a push model, as opposed to a pull ... is there no other way to restart things other than to reinitialize from a new snapshot?

If your agents continue happily along is because they were set to use -SkipErrors !!! (Dangerous option)

The approach to not reinitialize is to use a time in which you can run a diff between primary and replica and apply the necessary commands to the replica ( some people use datacompare from RedGate some others tablediff from MS)

* Noel