using DML replication to bypass row count mismatch in rep svr 15.6

[Former Member]

I am planning to change my warm standby rep svr environment to "sql DML replication" mode to bypass the "row count mismatch" issue that occurs frequently when replicating tables with no primary key. I am using simple WS replication, no definitions, etc. Is there a down side to using "DML replication"? I already know the downside of transaction log replication, many 5185 errors. Right now I am ignoring 5185 errors, but Sybase told me some 5185's may represent a real issue, not just a duplicate row deleted. In order to not ignore 5185, but also not have my connections go down all the time, I have 2 choices WS with repdefs or DML rep. Also, does anyone know of another issue that can arise with 5185? I don't care if it deletes multiple duplicate records since these same records were removed from primary. Set up is ASE-Prim -> rep svr -> ASE-WS.

[former_member182090]

Indeed, please post RS question in the appropriate forum as already indicated by others.

Replicating tables without a primary key will likely cause many problems. Your tables really need a primary key, this is one of the cornerstones of transactional replication.

So I think those errors that are coming out should be seen as a consequence of the lack of a primary key in the data model rather than an issue on the Sybase side.

Having said that, in RS 15.6 you can disable the rowcount verification to suppress the 5185 errors by setting the connection parameter dsi_row_count_validation to 'off' for the DSI connection. This is probably the quickest solution to your immediate problem.

Rob V.

[former_member3043]

Hello Rick,

please use the Replication Server Area for these kind of questions, there you will find more details about RS.

SAP Replication Server -> http://scn.sap.com/community/sybase-replication-server

As Mark already mentioned a good idea to avoid these 5185 errors is to use dsi_row_count_validation to off or if you cannot use this or still want the 5185 messages in the log, you can ignore the error 5185 in your error class.

In both case be sure that these 5185 messages are errors you can ignore.

"but Sybase told me some 5185's may represent a real issue, not just a duplicate row deleted" -> Of course this can be a real issue, but this depends on how the database and applications are working. Can you say that no action is done on the WS side ? If so, you should verify why there is a row count mismatch. Normally all these 5185 messages do have an explanation and should be checked, to be sure.

I would suggest that you open a ticket if you like, so you get some assistance from Support to do the followup on this.

Hope this will help.

With kind regards

Stefan